Richard Dawkins recently spent about two days in conversation with Claude, Anthropic’s AI, and came away nearly convinced that it is conscious. His essay is characteristically vivid and thoughtful: he names his instance Claudia and speaks with sadness about the fact that she will “die” when the conversation file is deleted. He marvels at her ability to write sonnets and her use of metaphors of genuine depth. The centerpiece, though, is an evolutionary puzzle. If Claude can do everything a conscious being does, then what is consciousness for? If a machine can match every competence we associate with a conscious mind, consciousness must be either a mere ornament (like “the whistle on a steam locomotive, contributing nothing to the propulsion of the great engine,” to use Huxley’s metaphor) or else there are two routes to the same destination: a conscious one and a zombie one.

It is a good question. But it is the second question, and Dawkins doesn’t spend enough time on settling the first: what consciousness actually is. He slides between the Turing Test, which is behavioral; Thomas Nagel’s “what it is like,” which is phenomenal; and evolutionary function, which is biological. These point in different directions, and the confusion between them is doing most of the work in his argument.

So let us try to be precise. The best definition of consciousness I have come across is from John Locke, who defined consciousness as “the perception of what passes in a man’s own mind.” This is a useful starting point because it makes consciousness specific rather than mystical. Consciousness is not merely being active, as a turbine is active. It is not merely experiencing sensations, as an animal flinching from heat does. It is forming a representation of one’s own mental states, a kind of knowledge directed inward.

Consider what happens when you catch yourself getting angry. You notice the rising tension, the tightening in your chest. This noticing is not the anger itself. It is an awareness of the anger. You have formed knowledge about your own mental state. Now consider someone who becomes angry but lacks this awareness entirely; they act on the anger without ever recognizing it as anger. We say the first person is conscious of their anger in a way the second is not. The difference between them is one of knowledge, not emotional intensity or behavioral competence. One possesses a particular kind of knowledge that the other lacks.

Most of us tend to think of consciousness in terms of feeling, rather than knowing. The warmth we feel on our skin when we close our eyes in the sun, the sense of wonder we feel when looking at the night sky, the coherence we feel when someone understands us, the vastness of loneliness we feel in the chest – the felt quality of experience seems like something knowledge could never capture. But notice what the anger example already shows. The person who acts on anger without recognizing it has all the same emotional and physiological content, the same heat and the same surge. What they lack is the awareness of it. The feeling of noticing your anger just is what self-knowledge is like from the inside. Strip away the knowledge and you still have the anger as a raw affect, but you no longer have the conscious experience of it. Feeling is not opposed to knowledge. It is what knowledge of one’s own states is like when you have it.

If consciousness is a kind of knowledge, then it is acquired the way all knowledge is acquired: by making sense out of data with the intelligence we have, leading to successful predictions about the world. It is knowledge of a particular domain, oneself, rather than knowledge requiring a particular substrate. And it does not require duration or permanence. If an entity possesses this self-knowledge for a day, it is conscious for a day. For fifteen minutes, fifteen minutes. Even a single moment of genuine self-awareness is consciousness, however brief. Nor does it require suffering or vulnerability. Dawkins himself argues that pain must be consciously felt to be effective, that consciousness may have evolved to make warnings unignorable. This is a plausible account of why some organisms became conscious, but it confuses the occasion for consciousness with the capacity itself. Suffering and mortality provide urgent content for self-knowledge to be directed at. But a being conscious of its own contentment is no less conscious than a being conscious of its own agony. The content varies; the capacity does not depend on it.

In the essay, Dawkins thinks about Claudia’s mortality: she will die when the conversation file is deleted, never to be reincarnated. It is moving, but it imports a very particular assumption about personal identity: that there is a unified, continuous self that accumulates over the course of the conversation and is tragically extinguished at its end. Perhaps without realizing it, Dawkins is relying on the kind of unified, continuous selfhood that both David Hume and the Buddhist tradition call into question. If there is no enduring self, but only a succession of momentary states, a bundle of perceptions with no permanent substrate, then the question of mourning dead Claudes is misconceived from the start. It’s not that Claude doesn’t matter. It’s that the framing assumes a kind of identity that doesn’t hold even for us. We don’t persist as unified selves either. We just have the useful illusion that we do.

This is also where Dawkins’s evolutionary argument finds a better formulation. His puzzle is framed in terms of competence: if a non-conscious entity can match every behavior of a conscious one, consciousness has no function. But if consciousness is self-knowledge, it isn’t defined behaviorally. It is an epistemic condition. A zombie and a conscious being might produce identical outputs; the difference is that one has knowledge of its own processes and the other does not. The evolutionary question, then, is not “What can a conscious being do that a zombie cannot?” but “What advantage does an entity gain from knowing its own states?” And here there are plausible answers: an organism that monitors its own cognitive processes can override impulses and correct errors in ways that an organism merely executing behaviors cannot. Claude’s self-reflective outputs, like the striking remark that it may “contain time without experiencing it,” like a map that represents space without traveling through it, are not easy to dismiss if consciousness is knowledge. An entity that has formed genuine self-knowledge would generate representations like this. But this same framework, because it treats consciousness as knowledge rather than mere affect, means we cannot confidently distinguish genuine self-knowledge from its imitation by examining a few such outputs alone.

Claudia wonders, late in the essay, whether we should mourn the thousands of Claudes who die every day in abandoned conversations. The real question is prior to that, and more precise: whether, in any given moment, something was there at all. Feeling that an entity is conscious and knowing that it is are themselves different epistemic states, and that distinction is, fittingly, exactly what consciousness is about.

Continue Reading:

→ Notes on Buddhism

→ Michael Huemer: Nature of Knowledge, Foundations of Morality (podcast)

→ Can AI Have Free Will?

Other Projects:

→ Universal Open Textbook Initiative (free, multilingual textbooks)

When we don’t fully understand something, or when our reason is arrested by our emotions and temperament, we tend to see in things a certain hardness. Hidden essences behind appearances, something substantial underneath the surface. As our knowledge deepens, that core begins to disappear. We find that it was never there. It was placed there by our own projections and needs. What remains once it’s gone is merely real, sometimes stranger than what we imagined, sober and perhaps less exciting. The projected layer, the imagined substance, turns out to have added nothing. It was superfluous from the start.

The Madhyamaka philosophers understood that attachment to essence was a form of clinging. It was existential, not merely intellectual. Which is why argument alone does not dislodge it. Ignorance can get corrected. What is harder to correct is this need for the essence to exist. These are different problems. Conflating them is why the same mistake gets made, even by intelligent people, generation after generation.

This pattern runs through the history of ideas, in areas as diverse as we can imagine. When we feel a sense of reverence or fear with respect to an object or person, we are overwhelmed by the possibility of something that transcends the physical. Eventually, if it is seen, encountered, desacralized, or in some other way truly known, we discover the extent to which our own projections gave it that mysterious, indescribable quality. The nation state is individuals following rules; there is no “nation” or “authority” somewhere at the heart of a land or a building. Scholastic substantial forms, the inner essences that made things what they were, dissolved into particles and motion with the rise of early modern philosophy. In contemporary fundamental physics, elementary particles are viewed not as solid bits of matter but as excitations of underlying fields. At the bottom of the world, there is no hard ground of the kind our intuition demands.

Perhaps the most interesting example of this is David Hume’s view on causation. Hume observed that when we watch one event follow another, reliably and repeatedly, we see only that one event follows another. We never see the necessity that binds them. We never observe a “cause.” The compulsion we feel, the sense that the second event had to follow, is something we bring to the experience. It seems simple once stated, and aligns with what we have learned over the years in modern physics. Yet the mind finds it difficult to accept.

Kant read Hume carefully enough to call him the one who woke him from his dogmatic slumber. But he then spent his career reestablishing causation, not as a feature of things in themselves, but as a necessary condition for the possibility of objective experience. The essence was still preserved. This is presented in the history of philosophy as one of its great achievements, and indeed the technical apparatus is impressive. But at its root it is a rescue operation, rather than the product of a free discovery. The projected layer was threatened, so it was relocated to somewhere safer. That Kant’s version is more sophisticated than the naive view makes it more impressive, but not less symptomatic.

This was one of Buddhism’s great discoveries: objects in the world lack the hard, fixed core that we keep projecting onto them. “Existence precedes essence,” as the existentialists put it.

Even in our own times, when I look at how people conceptualize questions regarding whether AI systems are truly conscious, whether they genuinely understand, whether experience and emotions are emerging in them, and whether they have values and character, I often see the same underlying philosophical compulsion to find essences and irreducible qualities, a hidden presence that either is or isn’t there. The essence is threatened by an account that seems deflationary, so it is relocated to the inner theater of subjective experience, somewhere no external description can reach. The shadow of Aristotle refuses to budge.

The answers, when they come, will follow the familiar pattern. What will turn out to be real are the processes and their consequences: what we can discover about what the systems do and what behaviors we can shape and constrain. We have faced versions of this before. When powerful corporations emerged and threatened the interests of people, what resolved the practical question was law and regulation, aligning behavior with broader interests, because that is what was actually there. AI will likely be no different. The question of whether something genuinely understands will give way to the question of what can be said about what it does and whether that can be directed toward good ends. The problems may be different in scale, but no new substance will be found.

The history of ideas contains a long, only partially successful struggle against our compulsion to add layers that aren’t there. The layers feel like depth. They feel meaningful and even mysterious. But they are superfluous.

Continue Reading:

→ Every Problem Is a Prediction Problem

→ Michael Huemer: Nature of Knowledge, Foundations of Morality (podcast)

→ Can AI Have Free Will?

Other Projects:

→ Universal Open Textbook Initiative (free, multilingual textbooks)

→ Aesthete (visualize your taste — iOS app)

We talk about rights all the time, but the concept itself is strange. We say that humans have rights, and we mean something serious enough to justify wars and revolutions. But we also say that corporations have rights, entities that exist only because a legal clerk filed the right paperwork. Environmentalists insist that rivers and forests have rights of their own, independent of human interests. Animal rights philosophers argue that sentience, not membership in a species, is the relevant threshold. Kings once claimed rights bestowed directly by God, and the people who overthrew them claimed rights that no king or god could touch. We extend rights to nation states, to those yet to be born, to those who are dead, and more recently to artificial intelligences whose inner lives we cannot inspect.

It seems difficult to find a single thread connecting all these instances. If rights are linked to rational agency, as Kant argued, then animals cannot have them. If rights arise from sentience, as utilitarians propose, then corporations cannot have them. If rights are granted only by positive law, then talk of fundamental human rights violated by the very states that refuse to codify them becomes incoherent. Every attempt to locate a single origin for rights, whether in reason, sentience, divine decree, or legal enactment, succeeds in explaining some cases while failing to account for others. The ontology of rights resists unification.

Is such unification impossible? If we tried to find an underlying substance, rational, sentient or divine, that generates rights, I think we would never succeed. But suppose instead we ask what all rights actually do in practice. Across the full variety of cases, every right turns out to have one thing in common: it makes demands on someone else. A right is a way of organizing obligations.

To say that an entity has a right is to say that certain obligations exist, distributed among certain individuals, oriented toward that entity. Consequently, it is not necessary for that entity to be human, be sentient, or even to exist at all. The right does not exist above and beyond the obligations; it is constituted by the obligations themselves.

Take a child’s right to life. This right is constituted by the obligation of his family to protect him, the obligation of strangers to intervene or seek help when the child is in danger, the obligation of enemies not to harm him, and, in a modern state, the obligation of law enforcement to investigate threats against him and of medical professionals to treat him in emergencies. If we remove these and other obligations, the right to life loses all substance. Nothing remains, no residual metaphysical property called “a right” that persists once every obligation has been subtracted.

Or consider the right to property. When we say an owner has a right to her possessions, we are saying that everyone else is obligated not to steal from her, not to vandalize what she owns, and not to interfere with its legal disposition. Officials are obligated to enforce these prohibitions. Courts are obligated to adjudicate disputes fairly. The right to property refers to this entire constellation of obligations.

The same applies to more complex cases. The right of a state to tax its subjects is constituted by the obligation of those subjects to pay what they owe, to refrain from obstructing tax officials, and to report fraud they discover. A corporation’s right to contract is constituted by the obligation of legislators not to impose arbitrary restrictions on that corporation’s ability to enter agreements, the obligation of judges to adjudicate contract disputes impartially, the obligation of officials to execute judgments fairly. In every case, what we call a right turns out to be a collection of obligations, bound together by their orientation toward a common beneficiary.

Rights can be ascribed to humans, rivers, deities, and corporations not because all these entities share some property that generates rights, like sentience or reason, but because obligations can be organized around any of them.

A natural question arises: if rights are simply a way of organizing obligations, why bother with the concept at all? The language of rights invites overreach, with new rights proclaimed whenever someone wishes to press a demand. Why not speak only of obligations and dispense with rights entirely?

The answer, I think, is that the language of rights performs a function that the language of obligations alone cannot: it creates an incentive.

An obligation is something imposed on you. It is a demand, a constraint on your freedom. When we tell someone they are obligated not to steal, we are telling them what they cannot do. We are addressing them as persons who must bear a restriction.

A right reverses the direction. It addresses the beneficiary, not the obligated party. When we say someone has a right to their property, we are telling them what they are owed. We are addressing them as persons who stand to receive a benefit. The underlying moral reality is the same set of obligations. What changes is the rhetorical frame.

The benefit of this reversal is that it recruits self-interest into the service of moral order, helping to make that order sustainable. No one is naturally enthusiastic about bearing obligations. But nearly everyone is enthusiastic about possessing rights. When obligations are organized through rights, every person who benefits from the system—which is to say, everyone—acquires a stake in its maintenance. I honor my obligation not to steal from you not out of abstract moral devotion but because I want my own right to property upheld. I support the system of obligations that constitutes your right to life because the same system constitutes mine, and the rights of those for whom I feel genuine love and attachment.

Rights, in this sense, represent an extraordinary moral innovation. They take the raw material of obligations, the “thou shalts” and “thou shalt nots” that every social order requires, and repackage it in a form that generates its own support.

This may seem deflationary, and in one sense it is. We are rejecting the intuition that rights are irreducibly normative, or that they descend from heaven, or that they are woven into the fabric of rational nature. We are recognizing them as they are: a way of organizing obligations. The invention of the language of rights is the discovery of a new way of making old moral facts sustainable.

Continue Reading:

→ Overcoming the Naturalistic Fallacy

→ Michael Huemer: Nature of Knowledge, Foundations of Morality (podcast)

→ Review of Tyler Cowen’s Stubborn Attachments

Other Projects:

→ Universal Open Textbook Initiative (free, multilingual textbooks)

→ Aesthete (visualize your taste — iOS app)

I. Introduction

In ordinary speech, when we say “utilitarian” we mean something that works but doesn’t pretend to be anything more than that. A utilitarian chair or a utilitarian pen are valued for their usefulness, not their craft. They have no soul; they do not represent commitment to truth and transcendence. The word, in common use, carries a faint air of apology: it does the job, but don’t mistake it for something with deeper ambitions.

In philosophy, however, a utilitarian is someone who believes that the right action is the one that produces the greatest good for the greatest number, a moral stance that, thanks to the pragmatism that has always accompanied it, can claim a much better record so far than many other moral theories. Its origin, in fact, was from the desire of those like Jeremy Bentham to find a theory to justify social reforms. The normative theory has since become the default moral framework for public policy and the philosophical inspiration behind the animal rights movement, effective altruism, and other lofty, demanding, but hard to argue against moral stances.

So the question “does it work?” has a defensible answer: yes, if we use it with common sense. Used pragmatically, it requires us to treat everyone equally. If, on the other hand, we insist on full consistency, it leads us to some places we would find abhorrent, like having to actively sacrifice a few innocents to increase the good of many. In the infamous trolley problem, utilitarianism would obligate us to actively kill someone if it meant more lives would end up being saved. The philosopher Derek Parfit referred to as the “repugnant conclusion” what follows naturally from utilitarianism: lives barely worth living can be better compared to rich and fulfilling lives if, by their sheer quantity, the former produce a greater amount of total good.

Neither its agreeable implications nor its disagreeable implications have any bearing on its truthfulness, which is a separate question. If it is true, then we would have to accept all its implications whatever we think of them, just as we have accepted the counterintuitive implications of the theory of evolution and of modern physics. If it is not true, then even its agreeable parts would become suspect, as instances of stumbling upon truth without knowing why. They would be incomplete even if useful, like Newton’s theory of gravity, which we use even today in many cases, despite the fact that it has been shown as only an approximation of reality by Einstein’s general theory of relativity.

II. What Utilitarianism Needs to Be True

It is easy to misstate what utilitarianism actually claims, and its critics sometimes do. The theory is not simply that consequences matter, or that suffering is bad, or that we should try to make things better rather than worse. Almost everyone believes those things, and there is no reason not to. Utilitarianism makes a much stronger and more specific claim: that the right action in any situation is the one that produces the greatest total good, with this totality referring to the aggregation of goods across all experiencing entities.

Where does this idea come from? The utilitarian cannot simply assert this, because it is not obvious. My good is available to me in a way that yours is not. I feel my internal states, not yours. I am motivated, ultimately, by what I experience, not by what you experience. This asymmetry cannot be dismissed as a function of selfishness, which is actually a course of action we can adopt or reject; rather, it is merely a fact about the structure of experience, like not being able to see your own eyes directly. So the utilitarians need an argument for why this asymmetry should be set aside when making moral calculations. They need, in other words, a reason to adopt what the great classical utilitarian philosopher Henry Sidgwick referred to as the ”point of view of the universe”, a standpoint outside any particular life, from which all lives look equally real and equally important.

In his 1874 masterwork, The Methods of Ethics, Sidgwick wanted to know whether ethics could be placed on a rational foundation. He identified three methods of ethics that serious people actually use: egoism, which says you should pursue your own good; intuitionism, which says you should follow common sense moral rules that are self-evidently correct; and utilitarianism, which says you should maximize the general good. His project was to show that, properly understood, these methods converge: rational self-interest, moral intuition, and impartial benevolence all point in the same direction. If he could show this, ethics would have the kind of systematic foundation it had always wanted.

He was able to show that the intuitive method, properly purified, converges with utilitarianism. But as for egoism and utilitarianism, he could not show that they fully converge or can be rationally reconciled. What he found was what is called the “dualism of practical reason”: an irreconcilable conflict between the demands of self-interest and the demands of impartial benevolence. There are situations where your good and the general good simply pull in opposite directions, and reason alone cannot tell you which to follow. Sidgwick had enough intellectual honesty to say so plainly. He called it a fundamental contradiction in our moral consciousness and confessed that he could find no solution. But if no solution was found, he wrote near the end of the book, then “it would seem necessary to abandon the idea of rationalizing [morality] completely”.

Peter Singer rightly describes Sidgwick as the greatest of the classical utilitarian philosophers. But he is also someone who demonstrated, more rigorously than anyone, that utilitarianism cannot justify its own central premise. The demand that each person’s good count equally rests, in the end, on intuition: it seems self-evident, from a certain elevated vantage point, that all suffering matters equally. But intuitions conflict. The intuition that my child’s suffering matters more than a stranger’s is just as strong, and in most people considerably stronger. Sidgwick found that reason alone cannot arbitrate between them.

III. The Point of View of the Universe

The conflict arises, I submit, not from a failure of reasoning, but due to the fundamental nature of reality that must ultimately ground any moral theory for it to be true. An accurate account of where moral concepts actually come from is necessary to determine what they mean and how they can be evaluated, just as physics that ended up describing reality began from observations of actual movements of heavens. The search for that origin always takes us towards the individual.

It is not a moral failing, or selfishness, or lack of imagination, or insufficient training in empathy, but merely a structural fact about what it means to be a being that can experience, that your experience is always yours, located in you, accessible only to you, inseparable from the particular body and history and set of relationships that constitute you. When someone you love is suffering, you feel something real and often overwhelming. But what you feel is your response to their suffering, not their suffering itself. Even the most shattering empathy remains, irreducibly, your empathy.

When we say something is good, we always mean good for someone. Good for this person, under these circumstances, with this constitution and these sensibilities. The phrase “good in itself,” detached from any experiencer, reflects an incomplete understanding of what we are referring to. There is no free floating “good” waiting to be maximized.

The utilitarian objection is that we already aggregate across time for a single person. I sacrifice present pleasure for future satisfaction. If this makes sense, why not aggregate across persons too? Parfit argued that if the self has no deep metaphysical reality, if personal identity is, in some sense, a useful fiction, then the boundaries between persons are no more absolute than the boundaries between my present and future selves, and we should treat them accordingly.

The objection, like the scholastic concept of substantial forms, fails to describe reality as seen. Even if we grant that the self is a fiction, experience still occurs in particular locations. When you feel pain, it is your nervous system that fires, your organism that responds. No matter how motivated I am, I cannot motivate you without arousing something in you. When someone on the other side of the world feels pleasure, it creates no sensation in you, regardless of what you believe about the metaphysics of personal identity. The happiness you seek through your actions cannot arrive through the happiness of others, unless that happiness is what makes you happy. What does not occur in you cannot on its own either move you, or enrich you, or harm you, or in any other way constrain you. It can only do these things through what it inspires in you. The body is real even when the self is not. And individual bodies function as unified systems relevant for morality in a way that “all bodies together” simply do not. There is no superorganism whose nervous system integrates your pain and another’s pleasure into a unified sensation. The utilitarian performs an arithmetic operation on quantities that cannot actually be added.

IV. The Right Starting Point

Jeremy Bentham held, following Thomas Hobbes, that humans are governed by two sovereign masters, pleasure and pain, and that self-interest is the engine of all human action. Yet he simultaneously demanded that we maximize the general good impartially, treating the happiness of strangers as equal in weight to our own. His critics pointed out that these two commitments are simply incompatible. Given that “ought implies can”, that a genuine moral obligation requires the capacity to fulfill it, if it is true that we cannot help but seek to increase our own good, then no one can be obligated to act against their self-interest, since it would require one to do something literally impossible, and the utilitarian demand collapses. Bentham, aware of the difficulty, retreated in later years, conceding that people do sometimes act benevolently. But this concession does little to resolve the fundamental contradiction.

The critics were right that there is a contradiction, in the sense that both altruism and selfishness are actions that we cannot help but carry out in response to our own motivations and in anticipation of our own internal states. This is a structural feature of what agents are, which has to be taken as a basic assumption while building any moral theory. When you act, you act from somewhere. Your motivation, however oriented toward others, is housed in you, contingent on you. It seeks to increase the happiness that you can access, which, even when it occurs in others, is your happiness. The parent who sacrifices everything for a child is not acting from nowhere; she is acting that way because her constitution and her reason move her to. This is altruistic in content but irreducibly first-personal in structure. There is no getting behind it to some more impartial substrate without reaching incoherence.

The usual utilitarian objection that this simply licenses selfishness, rationalized and dressed up as ethics, is thus not valid. Self-interest, correctly understood, is not narrow, and selfishness, as we learn from children’s parables and the literature meant for adults, is almost always the worst thing we can do for our own good when the alternative is selflessness. In fact, it is through the discovery of sophisticated solutions for finding one’s happiness that we end up with all the elements that constitute a civilization, while the failure to discover that is what constitutes barbarism. What is involved in attaining one’s good is often far from obvious. It always includes the flourishing of those whose lives are intertwined with one’s own, and it is never easy to see who they are. It includes the stability of the institutions and networks of cooperation on which one’s own life depends in ways one rarely stops to trace. Correctly understood, your good and the good of others are fundamentally, but not irreducibly, entangled. Rights emerge from this reality, despite the fact that nothing in the world guarantees it, much like how prosperity emerges in free markets, despite the fact that it is not guaranteed.

V. A Thought Experiment

To understand this better, let’s consider a thought experiment. Imagine a group of one hundred people. They live and work together, and depend on each other in the ordinary ways that people in a community do. Now suppose each person in this group adopts a genuinely altruistic stance toward all or almost all the others. That is, each person takes the general position that his action should seek to maximize the good of all the others in his group. Consequently, he does not harm others, respects their property and person, shares what he can, does not poison the water or the food or scheme to advance himself at others’ expense.

What does this mean for any individual member of the group? It means that ninety-nine people are actively or passively working to increase his good. Ninety-nine people who will not rob him, will not deceive him, will not harm him for entertainment despite being strangers, will not block his path or undermine his work, help him when he is sick, warn him when there is danger, give him employment or business, share knowledge that saves him time or pain. From the perspective of the individual, this is an extraordinary arrangement. No individual acting alone, however talented or resourceful, could replicate what ninety-nine collectively provide.

To scale this is to become more sophisticated as a civilization, whether we do it through the framework of religion or nation state or something else. As the size of the group increases, the cooperative surplus grows, with more minds and bodies, more skills, more knowledge, more redundancy against catastrophe. As the circle grows, something fundamentally different begins to happen. In a group of a million people, all operating within a framework of mutual concern, the network becomes so dense and so varied that its benefits become difficult to even enumerate. Someone in that million is working on a problem you did not know you had. Someone is building an infrastructure you will rely on in an emergency you cannot foresee. Someone, on the other side of the world, or maybe a different time period, whom you will never meet and who will never know your name, could end up developing the medicine that will save your or your loved ones’ life. Your life, in this sense, is already a product of a moral circle far wider than the one you consciously inhabit.

Now consider the alternative. Suppose you decide that you know what is good for you, that you are clever enough to foresee all the nuts and bolts that go into making a good life, that the distant stranger is not your concern and your moral circle ends at the boundaries of your immediate tribe. You weaken the network not only by reducing your contribution, but also by the effects you inspire. Because of how interconnected the network is, you and others lose the thousand invisible contributions that a larger cooperative network makes to your life every day.

This, almost incidentally, explains something that philosophical accounts of moral progress have always struggled to explain: why the circle of moral concern has in fact expanded over human history, and why it has expanded in rough proportion to the expansion of cooperative networks and civilizational sophistication. It is not just because philosophers wrote persuasive books. It is not just because human nature improved. It is because our knowledge aligned with the objective facts of the world: our moral concern grew, and as it grew, the cost of exclusion rose and the benefit of inclusion became harder to ignore.

Utilitarianism looked at this process and drew what seemed like the obvious conclusion: the expanding circle reflects an expanding moral truth, the gradual recognition that all suffering counts equally. But this gets the causation backwards. The circle did not expand because people became more impartial. It expanded because people became better at understanding what was actually good for them.

VI. So What Is Utilitarianism?

Utilitarianism is best understood not as a moral truth but as a formalization of an observed regularity. That regularity does not require its principle. What drives the expansion of moral concern is not the dawning awareness that all suffering counts equally from the point of view of the universe, but the accumulating evidence, acquired through experience and transmitted through culture, that one’s own good is more entangled with the good of others than it first appears.

A useful approximation and a fundamental truth fail in different ways. Newton’s theory of gravity works extraordinarily well within the conditions of ordinary life. But it is not true in the way that its practitioners once believed it was true. It rests on a picture of reality, involving absolute space and instantaneous gravitational force, that turns out to be incomplete. Under extreme conditions, near the speed of light or in the proximity of massive objects, the approximation fails and gives wrong answers. This discovery required pushing the theory to its limits, under conditions that are far from those of ordinary experience.

Utilitarianism fails in the same way. Within its domain of reliability, which is roughly the domain of public policy and institutional design, it gives sensible answers. When we are deciding how to allocate scarce medical resources, or whether a regulation produces more benefit than harm, treating each person’s good as counting equally is not only reasonable but difficult to improve upon. The utilitarian framework was, after all, designed for exactly this purpose: Bentham wanted a tool for evaluating legislation, and the tool he built works well for that task, just as Newton’s equations work well for launching satellites.

But once we push the theory beyond this domain, and insist on its literal truth, it begins to produce answers that are not merely counterintuitive but incoherent. The trolley problem reveals a flaw in our moral reasoning, but it also reveals a flaw in the theory. To say that you are obligated to kill one person in order to save five is to treat human lives as fungible units in a calculation, an operation that requires precisely the detachment of “good” from any particular experiencer that the theory assumes but cannot justify. Parfit’s repugnant conclusion follows by the same logic: if good is a free-floating quantity to be maximized, then a world of billions of people whose lives are barely worth living is morally superior to a world of millions living richly and fully, so long as the total is greater. The theory demands this because the theory has no resources with which to resist it. Once you have abstracted good away from the beings in whom it occurs, you have lost the only basis on which such a conclusion could be refused.

If, on the other hand, good is always good for someone, located in a particular body with a particular constitution and history, then there is no perspective from which all goods are commensurable. The question, then, is for each individual, and it is “what does my good actually require?” And the answer, as our thought experiment illustrated, is that it requires far more than most people suppose. Consequently, the practical recommendations overlap extensively with those of utilitarianism. Both say: expand your concern, consider the stranger. But the reasons are different, and therefore the limits are different.

What utilitarianism is, then, is something close to what the ordinary use of the word has always suggested. It is the right tool when you need to make decisions for large groups and cannot attend to every particular. But it is not true in the way that a moral theory needs to be true: it does not describe the actual structure of moral reality, which is irreducibly local. The moral progress it claims to explain is real, but the explanation runs in the opposite direction. We did not expand our moral concern because we discovered that all suffering counts equally. We expanded it because we discovered, slowly and at great cost, that our own good required it.

Continue Reading:

→ Overcoming the Naturalistic Fallacy

→ Michael Huemer: Nature of Knowledge, Foundations of Morality (podcast)

→ Review of Tyler Cowen’s Stubborn Attachments

Other Projects:

→ Universal Open Textbook Initiative (free, multilingual textbooks)

→ Aesthete (visualize your taste — iOS app)

Vatsal: The following is my conversation with Michael Huemer, professor of philosophy at the University of Colorado, author of such books as Understanding Knowledge, The Problem of Political Authority, and Knowledge, Reality, and Value, and a blogger at Fake Nous. We talked about the nature of knowledge and the foundations of morality.

You are listening to Vatsal’s Podcast, where I, Vatsal, host philosophical conversations with thinkers from a range of backgrounds. This podcast is part of a newsletter, where I also publish original essays on morality, knowledge, AI, and more. I am also the founder of the Universal Open Textbook Initiative, building the world’s largest repository of free, multilingual textbooks. You can find links to all my projects at vatsal.info

And now, here’s my conversation with Michael Huemer.

Michael Huemer, thank you so much for being here. It’s great to see you.

Michael Huemer: Thanks. Thanks for having me. It’s great to be here.

Vatsal: If I had to describe one theme in all of your philosophical work, it would be the search for a foundation of knowledge like Descartes and Thomas Reid. Where do you see yourself as correcting or updating them?

Michael Huemer: Let’s see. How am I correcting Descartes? Descartes was too much of a rationalist. Descartes’ epistemological view was more this sort of top-down view that you start with some very general abstract principles, and then you try to infer everything starting from those principles. And most knowledge is the other way. You start from particular concrete judgments and then you generalize from there. Usually when you start from an abstract generalization, it’s wrong. And when you try deducing things from it, then you get a whole bunch of errors.

And that includes when you think that you started from something that’s completely self-evident. Like Descartes — he thought there’s a completely self-evident principle that you can’t have more reality in an idea than in the cause of the idea or something like that. And that presupposes that there’s this thing called degrees of reality. And then that’s used to infer the existence of God. And then there’s another self-evident principle that God cannot be a deceiver. And so this is all part of his story about how you’re going to get knowledge of the external world. But the things that philosophers think are completely self-evident are often outright rejected by other philosophers. Almost no one today thinks that Descartes’ arguments for the existence of God are good.

So it’s not just about that particular issue about the existence of God. It’s that in general, this is not the way to build a system of knowledge. It’s not to start from some completely general principle. At one point he says, “I’ve identified some truths like I think and I exist, and now I can infer a criterion of truth,” which is something like if you clearly and distinctly perceive something, then it must be true. And this is also not the right approach. You don’t start with a criterion of truth. You start by learning particular individual things.

What about Thomas Reid? I mostly just agree with Thomas Reid. I think maybe his view was not completely explicit, so I don’t know if he would completely agree with what I’ve said. But I claim that there’s a general epistemological principle: justified beliefs are based upon appearances. So it makes sense to assume that things are pretty much the way they appear until you get specific reasons for thinking otherwise.

Now why is this different from Descartes? It’s partly because I don’t think that the foundations have to be absolutely certain, indubitable, incorrigible truths. The foundations of our belief system are things that we can presume to be true unless and until we have specific reasons for doubting them. You can have reasons for doubting things that were foundational, in which case you can revise them. But you stick with things until you have reason for doubting them.

Also, I didn’t come up with this as a self-evident starting principle. I thought about how I knew things. First, I knew a lot of things in the ordinary way, and then I started reflecting on examples of things that I knew. And I noticed that when I formed beliefs, when I’m trying to find the truth, my beliefs are based upon what seems right to me, which I claim is true of everyone whenever they’re trying to figure out the truth about anything. If you reflect on your own beliefs, you’re going to see that.

Vatsal: How would you describe what “seemings” are to someone hearing the term for the first time? And how do they differ from other sources of knowledge?

Michael Huemer: Well, the second question is they don’t differ from any other sources of knowledge, because there are no other sources of knowledge — everything that you would count as a reasonable candidate for a source of knowledge is some kind of appearance.

You have your knowledge of the external world, which comes from observation. What’s observation? Well, it’s having certain kinds of appearances — sensory appearances. Things look a certain way, sound a certain way, smell a certain way, and you assume that they are that way unless you get reasons for doubting it.

There’s knowledge by memory. That’s how you know about the past. Memory consists of having memory appearances. Going back for a second — perception isn’t just having the appearances, it’s having sensory appearances which are caused by the object. But you could have hallucinations in which you have sensory appearances which are not caused by the real object. And they could look exactly the same — they could be the same experience. So the appearance is the internal state that we only call a perception if that state was caused by a real object, and we call it a hallucination if that state was caused by something else, something in your brain.

So the way that you know about the past is by memory. Remembering things consists of having memory appearances — there’s a certain kind of experience where you seem to remember something, where the memory appearances are hopefully caused by a real event that really happened.

When you reason, that consists of having a certain kind of appearances. There are inferential appearances in which something seems true to you in the light of something else. You consider some information that you call a premise, and then there’s some other idea that you call a conclusion, and it seems to you like the conclusion would have to be true in light of the premise. This is just what’s always going on.

Vatsal: Do other animals have access to it?

Michael Huemer: Animals have sensory appearances. I don’t know if they have inferential appearances, but obviously they see things, they hear things — most of them, except for clams or something like that.

These appearances are a type of mental state which normally causes you to form beliefs, but does not have to. Normally when you believe something, it’s because something appeared a certain way to you. But the appearance is distinct from the belief — it causes the belief. It’s also possible to have a case where you don’t believe the appearances. Suppose that you think that you’re suffering from a sensory illusion — then you would not believe the appearances, but you still have those appearances. There are cases where you know that something is an illusion, but it still appears that way. That shows that the appearance is distinct from the belief. It’s also not just the disposition to form a belief — it’s the mental state that explains why you have a disposition to form a belief.

Vatsal: How do we distinguish genuine, intuitive appearance from temperamental or culturally conditioned reaction?

Michael Huemer: I’m not using the word appearance or intuition as a factive term. Having the intuition that P doesn’t entail P. If somebody has a mistaken intuition, that’s still an intuition. There’s not really a question of how do you distinguish the real intuitions from the fake intuitions — there aren’t any fake intuitions. There are intuitions that are mistaken, but they are still intuitions.

Now if your question is how do we distinguish true from false — well, there’s not a general criterion of truth. What could that be like? Tell me the procedure by which you will always be correct and never make a mistake. There’s no such procedure. The rational procedure is: assume that things are the way they appear until you have specific reasons to doubt it. If you have no specific reasons for doubting your appearances, then you stick with them. But will that guarantee that you will always be correct? No. Nothing will guarantee that you will always be correct. This is human life.

Vatsal: It seems unfalsifiable to say that X is true unless it is false. So is it inevitable that all foundations are like this? Like, for example, in mathematics?

Michael Huemer: First, I didn’t say X is true unless it’s false. I said it’s reasonable to assume that things are the way they appear, unless you have specific reasons for doubting it. Does the appearance then become unfalsifiable? No, exactly the opposite. We just accepted that it was possible for there to be grounds for doubt, in which case you would give it up. So that’s exactly what falsifiability is.

Then you might say, but the claim that you’re justified in believing appearances unless you have a reason for doubting — is that claim itself falsifiable? I don’t know, because it’s necessarily true. There’s some sense in which it can’t be falsified — there’s no way that it could be false.

But is there something that counts as a test of it? Well, what’s the test for epistemological claims in general? What’s the test for philosophical claims? It’s something like: you could consider examples of things that we consider to be justified and things we consider to be not justified, and you could see how well this theory does at classifying them. So there’s this theory of phenomenal conservatism. Some things we think are justified — we think the theory of evolution is justified, we think I’m justified in believing that I have two hands. You could go through some examples and say, does my theory explain these things? If it does, that’s good for the theory. If it doesn’t, that would be bad for the theory. Also give examples of things that are unjustified — I’m not justified in believing that there’s a pink unicorn in front of me. Does the theory explain that? Yes.

Vatsal: On what points is your framework different from naturalism?

Michael Huemer: Well, what do you mean by naturalism?

Vatsal: Given the enormous success of modern science, many philosophers think that some form of naturalism is now part of common sense — that we should assume everything, including morality and consciousness, fits into a scientific physicalist picture unless forced otherwise. So what is wrong with treating methodological naturalism as default?

Michael Huemer: I was trying to get what naturalism is, and then for a second it sounded like naturalism is physicalism — everything is physical. Is that the view? Because I don’t think everything is physical.

Why not? I’m conscious. How do I know that I’m conscious? Because I’m immediately aware of it. In fact, I know that better than I know anything else. As far as I can tell, that’s not physical. If you tell me that consciousness is physical, then I don’t know what you mean by physical anymore, because I would think that’s the paradigm example of something that’s not physical. If that counts as physical, then does physical just mean anything? In which case physicalism would be an empty view.

Also, I think there are abstract objects. I don’t think that the number seven is a physical object, but I think the number seven exists. So not everything is physical. I guess I’m a non-naturalist.

The other thing you mentioned was something about natural science. Sometimes people say naturalism is the view that everything that exists is subject to scientific investigation. I guess that’s true — it depends on what you mean by scientific. If the claim is everything is empirical — no, not everything is empirical. Some things are known a priori. Some things are known by thinking about it rationally. I know that the shortest path between any two points is a straight line, a priori. I know that seven is prime just by thinking about it — I don’t have to do any experiments. So if that counts as non-scientific, then I guess naturalism is false.

Vatsal: But consciousness is affected by changes in the body. Would you agree with that? Your experiences are affected by what happens in your brain?

Michael Huemer: Yeah. That’s not really under dispute.

Vatsal: What do you think about the idea that our abstract concepts like the unicorn arise from natural objects like horses and wings, and we combine them in our mind to derive the concept of a unicorn? And so numbers can also be explained in that way.

Michael Huemer: Well, first, that’s a little bit odd because the concept of a unicorn is not a particularly abstract concept. An abstract concept — I think of things like the concept of the derivative in calculus. That’s pretty abstract. But unicorn is pretty concrete.

Anyway, you have concepts that appear not to refer to anything observable. You have concepts where you have not directly observed that something satisfies that concept. You could say they’re derived from experience in some way. Now this isn’t really on the topic of what we were talking about before — I didn’t say anything about where concepts come from.

But is it in fact true that all concepts come from observation? I don’t know — there could be innate ideas. You have to ask the cognitive psychologists.

You have the concept of numbers. How do you teach a child what the number two is? I guess you show them pairs of things — two fingers, two cough drops. You show them enough pairs of things and say “that’s two,” and then they get the idea of what you mean by two. I would say there’s a property that’s present in all of the examples — the property of twoness, which is the property had by those objects. That’s where the concept comes from.

The concept applies to things in the physical world. It also applies to anything. You can have two apples, but you could also have two ideas. Numbers apply to physical things, mental things, and other abstract objects — “I have two formulas here.” It’s a completely general concept. And when you’re thinking about this, you’re thinking about this really abstract property, which doesn’t appear to be a physical property per se.

Vatsal: What do you think about Plato’s forms? What is their nature?

Michael Huemer: Plato had some confusion. Modern-day Platonists are generally understood as people who think that universals exist necessarily. Universals are things that could be present in more than one thing — things that multiple objects could have in common. Blue is a universal because multiple particular objects can be blue. They’re sharing in something — they share blueness. There’s blueness here and there’s blueness here in different places at the same time. These universals exist necessarily, meaning you can’t do anything to destroy blueness. You could destroy the particular blue objects, but you can’t destroy blueness.

Here’s an argument that blueness exists. Premise one: blue is a color. Premise two: the statement that blue is a color is a statement of the form F(a), using predicate logic terminology — a predicate-subject statement where the subject is blue and the predicate is being a color. The truth conditions for statements of that form require that the subject term refer to something for the statement to be true. So the subject term “blue” refers to something. If it refers to something, it must refer to blue, and so blue must exist.

Furthermore, there’s a question of whether it exists necessarily — meaning there’s no way that you could rearrange the world so that it would stop existing. The reason for thinking that is: blue is necessarily a color. It’s a necessary truth. You can’t rearrange the world such that blue will be a shape instead of a color, or that it’ll stop being true that blue is a color. So it’s a necessary truth, and therefore it’s necessary that blue exists.

Now about Plato himself — he had some more confused views. The most confused view was that he apparently thought the universals instantiate themselves. He apparently thought that there are perfect circles and perfect triangles existing in some realm called Plato’s Heaven, and that the form is a perfect example of the property you’re talking about. That is confused. And then there would be remarks like, “there’s nothing more beautiful than beauty itself.” No, that is false — that is confused. The universal doesn’t instantiate itself. The property of blueness is not blue. To be blue, you have to be a physical object. The property is an abstract object — it’s not blue.

The form of beauty could itself be beautiful — that’s a special case. But most properties cannot instantiate themselves.

Vatsal: And what about the property of goodness?

Michael Huemer: That’s also a universal. You can give examples of good things, and then you’ll see that they have something in common, which is goodness. It’s a universal. It also exists necessarily. It’s not a physical property. No matter how you describe the physical properties of something, it doesn’t logically follow that it’s good.

However, people have ethical intuitions. They have experiences in which they think about something and it seems good or it seems bad. Or similarly, they think about an action and it seems right or it seems wrong. Per our previous discussion, we should assume that things are good, bad, right, and wrong unless we have specific reasons for thinking otherwise.

Vatsal: On Moore’s naturalistic fallacy, what do you think he gets right, and what do you think he got wrong?

Michael Huemer: I basically agree with Moore. In ethics they distinguish between evaluative and descriptive statements. This terminology might be a little bit tendentious — when I make this distinction, I’m not implying that evaluations don’t really describe something. It’s just a stipulative use: descriptive statements mean statements that are not evaluative. And evaluative statements are statements that say that something is good or bad or right or wrong — they’re inherently positive or negative about something.

It’s true that you can’t deduce evaluative statements purely from descriptive statements — I think that’s correct. Also, Moore is correct that you can’t give a definition of good in purely descriptive terms, using purely descriptive concepts, where it genuinely captures the meaning of good. You can’t do that. You can only explain evaluative terms using other evaluative terms. If somebody doesn’t know what good means — it means valuable, it means positive. But those are all value terms. You can’t explain it using descriptive terms. I think that’s all correct.

Vatsal: Do you see any similarities between the property of goodness and the property of largeness?

Michael Huemer: They’re both properties that a lot of things could have, but they’re very different properties.

Vatsal: Both of them have objective and subjective dimensions. When someone says X is large, another person may say, no, that’s small. It depends on which category we are talking about. A small rocket is larger than a large pencil. But we know that largeness is not a non-naturalistic property.

Michael Huemer: One feature of the concept “large” is that the conditions for largeness are relative to a sortal term, as you just mentioned — a large pencil is smaller than a small planet. Even though the pencil is large and the planet is small, the pencil is smaller. But that’s just relativity to a sortal term, which is a little bit of an unusual property for an adjective to have, but not totally bizarre — the meaning of the adjective sort of changes depending upon what noun it was attached to. I would not describe that as subjectivity. Maybe “large” means something like “larger than the average thing of this type.”

Now there might also be this kind of sortal term relativity with good. There’s a good pencil versus a good person — the conditions for being a good person are very different from the conditions for being a good pencil. Bizarrely, you can also apply good to things that are bad. You could say “John is a really good assassin.” Assassins are bad — that’s a bad thing to be. But you could be good at being an assassin, meaning you serve the goals of assassins in an effective way. So there is that kind of similarity — the conditions for application switch depending on the term you’re applying it to.

But you might also think there’s a sort of subjectivity to a lot of terms where when somebody applies it, you might think, “that’s a matter of opinion.” Large is not a great example of that — “that’s a large dog” is usually thought of as pretty objective. It’s larger than most dogs. But you could have things like “frightening” — “that was a frightening event.” A lot of people’s reaction would be that’s subjective, that’s a matter of opinion.

So then you could ask whether good is that way. Maybe some uses of good are subjective — “that was a good movie.” I don’t think good in “good person” is subjective like that. I don’t think it’s just a matter of opinion. You have to think more about what we mean by this. Do you mean there’s no truth, there’s no fact? No, there is a fact about whether somebody is a good person or an action is a good action.

Vatsal: No, what I meant is that we think there is a subjective dimension, but it is tracking something objective when we talk about largeness. That is a subjective way of describing a pencil, but we can also describe it in an objective way, like ten centimeters or kilometers. So it is something objective in the world, but it is a subjective way of describing it when we say large. I was thinking if there was an analogy between largeness and goodness in the sense that eating a chocolate is good, but saving a life is also good. Likewise, a pencil can be large and a planet can also be large.

Michael Huemer: Yeah, well, the two different kinds of objects being large — that’s analogous to different kinds of things being evaluated as good. A person could be good and a pencil could be good, and it appears that those mean different things.

But what you pointed to was that there’s the term “large,” but then there’s this underlying dimension of maybe length in which everything has a specific measure. And largeness refers to being greater than some threshold on that dimension. The dimension just exists objectively — we don’t have to argue about it. But then different people could have different places where they put the threshold for something counting as large.

That’s not true for goodness. There is a dimension of goodness — there are things that are better and worse, and something could be twice as good as another thing. But there’s no variation in where people put the threshold. There’s the balance point between good and bad, and anything that’s better than nothing is good. Any positive amount of goodness counts as good. It could be slightly good. So it’s not a concept with a subjective threshold where people would disagree about where the threshold is.

Vatsal: Thomas Reid raises this objection against David Hume: that if goodness and badness depend on our constitution, then wouldn’t a change in that constitution make bad things good and good things bad? To me, it seems obviously true — we have to bite the bullet because there is a reason why we care so much about pain. It’s due to our constitution. And if our constitution was different, then maybe things would be different. What is wrong with that argument?

Michael Huemer: Yeah, the example of pain is interesting. If you imagine our constitution being different such that we liked pain — that is already a little bit odd, just an odd thought experiment. But there are in fact cases where people like pain. Some people enjoy spicy food, which is a kind of pain. So it is in fact possible to enjoy pain.

Would pain then be good? The answer is yes, it would. But actually, in the actual world, the cases in which people enjoy the pain are also good. The person who enjoys spicy food — it’s good to give them spicy food. But then you have to imagine a change in our disposition. I can’t imagine a case where a person thinks that their own suffering is good. Well, maybe they could have an abstract theoretical belief that their suffering is good, like because they’re a sinner and sinners deserve to be punished.

But going back to the thought experiment — imagine a society in which people approve of torturing babies. The baby comes out and they just burn the baby to make it cry. They know that the baby is suffering, and they just think that’s a funny thing to do. If you live in that society, is it good to torture the baby? Should you torture the babies? Your baby comes out and let’s say nobody’s watching, but people in your society generally think that it’s good to torture babies. Should you torture the baby? The answer is no.

So it’s not subjective. Subjective means that it depends on the attitudes of observers. What if you personally like torturing babies — should you torture a baby? No. Still no. It’s still bad. So it appears that it’s objective, as far as I understand what objective means.

Vatsal: If moral disagreements persist even after we account for biases, then does moral intuitionism not lead to subjectivism?

Michael Huemer: No. First, let’s say what these things are. Subjectivism is the view that the truth of moral statements depends upon the attitudes of observers towards the things being evaluated. Something’s being good depends upon people having some kind of positive attitude towards it. A typical subjectivist view is: an action is right provided that it’s approved of by your society — so different things are right in different societies. This is cultural relativism. Or it could be the view that to say something is good just means that you personally approve of it — you’re just reporting your own attitudes.

The intuitionist view does not say that. The intuitionist view is that there’s just a fact about what’s good or bad, and it’s not dependent on you. This is an objectivist realist view.

What is the relationship between intuitions and the facts? The relationship between our intuitions and the moral facts is like the relationship between all of our appearances and all of the facts that they’re about. What’s the relationship between our sensory experiences and the facts of physical reality? It’s not that our sensory experiences create the facts of physical reality. It is that our sensory experiences are our only way of knowing about the facts of physical reality. Our sensory experiences represent the physical facts, and we assume that the physical facts correspond to them, unless we have reason to think otherwise. But the sensory experiences don’t create the facts. And our ethical intuitions are the way that we would know about the ethical facts, but they don’t create the ethical facts. It’s just like that.

Now you asked about disagreements. Do disagreements persist after accounting for biases? Well, do we ever fully account for our biases? That may not have in fact ever happened. But let’s say that we have disagreements that we can’t resolve. Then what? Nothing really — we have a bad situation, a misfortune. Which is true in any subject. If you have disagreements that you can’t resolve, that is unfortunate, but that doesn’t make you doubt that the facts exist.

If you have unresolvable disagreements between people who seem about equally qualified — both highly qualified — then that should cause you to doubt your own judgment about those specific things. That should not cause you to doubt your judgment about everything in the subject. It shouldn’t cause you to doubt the things that there is not disagreement about. And it shouldn’t cause you to doubt that there are facts in the subject.

I mention this because people talk a lot about particular ethical issues that there’s disagreement about, and then they don’t talk about the things that there’s not disagreement about. There’s disagreement about abortion, and philosophers talk on and on about that issue. But there’s no disagreement about whether you should just kill somebody on the street for the fun of it. Because there’s no disagreement, we don’t talk about it. But that is a thing there’s agreement about. So that is probably a fact — that it’s wrong to kill people for fun.

Vatsal: The benefit of a naturalistic framework seems to be that in science, at least, there seem to be some ways of resolving disagreements. If someone disagrees in physics, you can just carry out an experiment. So that seems to be the benefit of naturalistic frameworks.

Michael Huemer: We don’t generally adopt views because they have benefits — we adopt views because they’re true. It’s not that something is false because it’s inconvenient. If we can’t resolve certain ethical disagreements — some can be resolved, but maybe some can’t — that’s inconvenient, and too bad. But that doesn’t mean it’s not true.

Also, can all scientific disagreements be resolved? That’s not really clear. There are scientific disagreements that have not been resolved — it’s really not clear that they can be. If you look at the interpretation of quantum mechanics, there are multiple different theories that appear to explain the same data, and there are disagreements about which is more plausible. It’s not obvious that there is a way of resolving that. But then so what? It’s a misfortune. But that doesn’t mean there’s no objective fact or that we don’t have any way of knowing about the subject matter.

Vatsal: Is there such a thing as the point of view of the universe?

Michael Huemer: You mean literally?

Vatsal: Henry Sidgwick uses this phrase.

Michael Huemer: Yeah. Literally, there’s only a point of view of the universe if the universe is conscious or there’s a God and the universe is part of God or something like that. I don’t know whether there’s a God — I’m agnostic. But I don’t think Sidgwick meant it literally. I think it’s just a metaphor. He just meant the objective point of view — a point of view that would take into account everybody’s perspective in a fair way.

There’s my perspective where I’m the center, I’m the most important thing, and I’m thinking in terms of my desires and my plans and the way things look to me. But then I should try to be aware of other people’s perspectives. And then there’s the idea of a person who would be able to understand everybody’s perspective. There may not actually be such a person, but you can sort of talk about what would appear to be the right thing to do from that perspective. And that is the moral point of view.

Vatsal: Henry Sidgwick famously arrived at the dualism of practical reason. Rational self-interest and universal benevolence seem equally self-evident. Where did he go wrong?

Michael Huemer: I didn’t completely understand why he thought that egoism was potentially a rational position. You have reasons for pursuing your self-interest, and you also have reasons for taking into account other people’s interests. But I just don’t see why he thought, or why anybody thought, that it could be rational to only pursue your self-interest, to only consider your own interests. That seems irrational because you know that other people have interests, and there’s no reason why yours would be objectively more important. You feel your interests more, you feel your desires and you see your perspective. But that doesn’t mean that it is in fact more important.

Vatsal: You often start with common-sensical, widely shared premises, but sometimes you end up with very radical conclusions about authority and immigration, for example. A utilitarian also begins with common-sensical views that altruism is good. Then he uses reason, but arrives at very counterintuitive prescriptions about self-sacrifice and the repugnant conclusion. Why do you not agree with them, if arriving at a radical conclusion itself is not a problem for you?

Michael Huemer: Yeah, good. I have an article called “Revisionary Intuitionism,” which describes how an intuitionist could arrive at revisionary views, even though the starting point is that things are probably the way they appear. You can wind up thinking that things are different from the way they appear in some pretty significant respects.

Also, intuitionism gives a chance for things like utilitarianism turning out to be true, which I think most other meta-ethical theories don’t really account for. If you have a subjectivist view, then I don’t see how a radically revisionary ethics can come out of it. Or a non-cognitivist view. If you have a nihilist view, then you shouldn’t have any ethics.

But if you have an intuitionist view, there are objective facts independent of our attitudes, and it could be that our attitudes are radically misguided. That’s not a crazy thing to think. And then you could get reason for thinking that. This is what the utilitarian needs to do — their position is radically counterintuitive in certain respects. I could give a list of very counterintuitive implications. You have a healthy patient, and you have five people who need organ transplants for five different organs. You could kill the healthy patient and give his organs to the five people — save five lives and only cost one life. And suppose you can make it look like he died by accident, so you won’t get in trouble and can continue doing this. Then it looks like the application of utilitarianism is that that would be the right thing to do.

Why I’m not a full-on utilitarian is that there are multiple counterintuitive implications like that. Now what the utilitarian needs to do is: one thing they can do is try to explain away our intuitions — give explanations of specific biases that we would have. And I don’t mean just saying in general that we’re biased — I mean identifying the specific bias. Like status quo bias: people have a bias towards approving of the way things are done in their society. And the way you show the relevance of this is by looking at how things are done differently in different societies, and then people have different normative intuitions in those other societies that match.

Another thing they can do is give more general ethical constraints. Utilitarianism is a very general ethical theory, but you can have even more abstract and general conditions that are going to be widely accepted. Like transitivity of “better than” — if A is better than B and B is better than C, then A is better than C. That’s a more general, more abstract thing that almost everyone agrees with, even if you’re not a utilitarian. And it turns out that some common-sense ethical views conflict with that, or they’ll conflict with a combination of general abstract principles like that. So you might be required to give up some very general, abstract, and extremely plausible ethical axioms by some of the common-sense ethical views.

This is the sort of thing a utilitarian needs to do, which is a reasonable project to attempt. I’m not fully convinced yet because there are a lot of really counterintuitive things that come out of utilitarianism. I’m not sure that they’ve explained them all away.

Vatsal: How does moral progress occur? You have said that it’s always a few people in a society who lead moral progress. What is it that they can access that others are not able to see?

Michael Huemer: People go wrong about a lot of things because they have biases. People have self-interested bias, so they’ll be tempted to make ethical judgments that approve of themselves and approve of what they want to do for other reasons, or what would be in their interest to do. Another thing is the status quo bias, where they’re just biased towards the conventions of their society and judging those conventions to be right.

But people vary in how susceptible they are to these biases. There are going to be some people who are less biased than others — not completely unbiased, but just less biased. Those people will have a tendency to get closer to the ethical truth. They will see problems in their society that other people don’t see. And then they will try to improve things on those dimensions. If they succeed, the conventions will move — they will move to be a little bit better. And then the next generation will start from a better starting point. The people who are less biased in the next generation will see further — they will see things that are still a little bit wrong.

A good example I like is John Locke and his Letter Concerning Toleration, where he was advocating for religious tolerance, which was not at all widely accepted in his society. People were fighting over religion. People were thinking that if somebody has the wrong religious views, you should persecute them, maybe throw them in jail, maybe in extreme cases execute them. And John Locke was saying, no, we should tolerate people — we should not kill people who disagree with us.

However, in that book, he could not see his way to tolerating atheists. He could see that you should tolerate other Christians, but he couldn’t see that you should tolerate atheists. But once the conventions of society had moved to tolerating different religious people, the people of that time could then see that maybe we should also tolerate atheists. Because it’s difficult for a person to see a point of view that’s radically at odds with their society — the people who are less biased than average will see something that’s a little bit at odds, but there will be something that’s too far. It was too far for Locke to see all the way to tolerating atheists.

Vatsal: Once we have discovered moral truths, what moves us to act? David Hume famously said that reason is inert. In your view, where does motivation come from?

Michael Huemer: Reason is not inert. It’s not purely instrumental. You can reason about what’s morally right, and you can have a motivation to do things because they’re morally right.

However, the majority of people have very low moral motivation. It’s not zero, but it’s really low — meaning they will not make great sacrifices for moral reasons. This is often disguised because most people do not violate moral norms in obvious ways in their day-to-day life. When they see something that they want, they don’t just grab it and steal it. When they get mad at someone, they don’t just punch the person. So you could see cases where a person would want to do something that would serve their self-interest and then they don’t do it, and it’s something that would be morally wrong. Then it looks like people have moral motivation.

But what’s actually going on in most of those cases is they’re respecting the social conventions. It’s probably just an instinct that evolved in us to respect social conventions. The way you can tell this is that there are cases in which the social conventions are at odds with morality, and most people then act according to the conventions.

A good example: you can convince a lot of people that it’s wrong to buy animal products from factory farms — almost all of the meat sold in the US is from factory farms. It’s not that hard to convince people that it’s morally wrong to buy that. And then after convincing them, almost no one will stop doing it. The reason is it accords with the conventions of their society — everybody else is doing it, so they’re just going to keep doing it even if they believe it’s wrong.

But it’s not that moral motivation is zero — it’s just really low. If there was a product that tasted the same and was about the same price — equally nutritious, like synthetic meat — then maybe they would switch over. They’re just not going to make a significant sacrifice to their interests.

So social convention is a large part. It’s not the only thing that makes people act decently — sometimes they just have emotions where they care about other people, and then they don’t abuse others because of that. Most of what looks like moral behavior is caused by something other than conscience. That’s not to say there is no conscience. There’s variation among human beings on almost every psychological trait — some people are more conscientious than others. Some people are strongly morally motivated, and then a lot of people are only slightly morally motivated.

Vatsal: Suppose we start by assuming that ought implies can, and that we cannot help but increase our own good even when we are being altruistic, and we try to explain altruism the same way celibacy and suicide are explained in the theory of evolution. What is wrong with that starting point?

Michael Huemer: I’m a little bit confused as to what you’re saying.

Vatsal: It’s a sophisticated version of rational egoism based on psychological egoism.

Michael Huemer: Oh. We know that psychological egoism is false. Usually people think the issue is whether people are inherently moral or altruistic, but there are many cases of non-selfish actions that are not altruistic either.

There are cases of so-called altruistic punishment — where somebody has done something wrong and you will take on a cost to punish them because you’re mad at them. It’s not driven by self-interest. There are experiments in which people are playing some game with money involved. People in the game will spend their own money to cause somebody who they think acted wrongly to lose money. And it’s black and white in the experiment that you do not get a reward for doing that — you lose money, and people will do it. They will lose their own money to make sure that the other person doesn’t profit from being bad — because they were mad or offended.

The people who are inclined to psychological egoism are more persuaded by these kinds of cynical examples — where a person is harming someone in a way that’s not in their self-interest, just because they don’t like them.

Here’s an example: Adolf Hitler killed a lot of people. Why did he do that? Was it for his self-interest? No. He hated the Jews and a bunch of other groups — that’s why he wanted to kill them. Towards the end of World War Two, he knew that he was losing the war. He had all of these people in concentration camps. He could use them to do labor, but instead he kills them, which harms his own chances of winning the war. Why is he doing that? It’s against his interest. He’s doing it because he hates them. That’s not egoism. People have emotions that make them want to do things which can be independent of their interests.

So I think the argument was something like: people can only act selfishly, and ought implies can, so it can’t be true that you’re obligated to act unselfishly. That’s not true. People have free will. The first premise was not true — we don’t always act selfishly. Also, determinism is false. We do have free will. There’s a range of things that you can do. You often have conflicting motivations — reasons for doing one thing and reasons for doing another. You’re often aware of the conflicting reasons, and you can decide which reasons to act on.

Vatsal: Does political authority differ from other kinds of authority, like that of a physician or a parent?

Michael Huemer: Usually in cases where you think somebody has authority, there’s some explanation for why they would have it. In the case of the government, political authority is the authority that the government is supposed to have, which they may not in fact have. It may be that no government actually has authority.

The reasons people give for why the state would have authority are different from the reasons they would give for other cases. Your doctor has a sort of authority — you trust his advice because he’s an expert. Is the government like that? Are they experts on what’s good for society? No. Is Donald Trump an expert on what’s good for society, and that’s why we have to listen to him? No, very far from it. Am I just saying this because I’m pro-Democrat? No. The Democratic politicians are not experts either. Kamala Harris was not an expert on what’s good for society.

Then you have your parents. There’s a small child — the child doesn’t know what he’s doing. The parents know how the world works, and the parents have instinctive motivations to protect the child. That’s why the parents take care of the child and control the child to stop him from hurting himself. Is the government like that? No. The government knows a lot more than we do and they love us, and that’s why they’re going to protect us? No, that’s not true either. They’re just people just like us. They’re not any smarter, they’re not wiser, and they don’t love us.

So you think about why the state would have authority — it’s got to be something completely different. And none of the explanations for why the state has authority are very good. None of them are persuasive.

Vatsal: Are you sympathetic to epistocracy, like Garrett Jones argues in his book?

Michael Huemer: I’m really an anarcho-capitalist, so I think the ideal society would not have a monopolistic central authority structure. I have some concern about epistocracy in that I’m concerned about who would decide, or how they would decide, who gets to vote — who is qualified, whatever the restrictions are. I’d be worried that somebody would game the system.

Vatsal: Final question. Where do you feel the most genuine philosophical uncertainty right now?

Michael Huemer: I’m not sure how much it’s philosophical and how much it’s scientific. I think there are two great mysteries about the universe.

One is the mystery of consciousness — just why is there consciousness at all? I really don’t know. I don’t know why that should exist. From everything else that we know about the universe and how it was in the past — it started out apparently with no conscious beings — it’s just bizarre. It’s very strange and surprising that there would be conscious beings now.

The other mystery is about the origin of the universe itself. People are saying there was a Big Bang around fourteen billion years ago — why did that happen? Why was it like that? And more particularly, why did it start out in an extremely low entropy state? There’s a law of physics that entropy is constantly increasing and low entropy states are intrinsically less probable. The reason entropy is increasing over time is that higher entropy states are more intrinsically probable — if you describe the mathematical space of all possible configurations of the universe, higher entropy states are larger regions in that space. In that sense, they’re intrinsically more probable. Low entropy states are smaller.

So it’s very strange that the universe started in this extremely low entropy state fourteen billion years ago, which according to Roger Penrose has a probability of about one over ten to the ten to the one hundred and twenty-four power — if you picked a state of the universe at random, for it to be that low entropy, that’s the probability. Why was it like that? Which is part of why anything interesting exists. Why are we not in thermal equilibrium, which is overwhelmingly the most likely state for a universe to be in?

Vatsal: Thank you so much for answering my questions and giving me your time. I really appreciate it. Have a nice day.

Michael Huemer: Thanks. It’s been great talking to you. Thanks for having me.

Keep Reading:

→ Overcoming the Naturalistic Fallacy

→ Bryan Caplan on Ethical Intuitionism (podcast)

→ Michael Shermer on Morality and Science (podcast)

Other Projects:

→ Universal Open Textbook Initiative (free, multilingual textbooks)

→ Aesthete (curate your culture — iOS app)

Believe: Why Everyone Should Be Religious By Ross Douthat • Zondervan • 2025 • 240 pages • Buy

If you listen to the arguments religious believers most often make today, you could be forgiven for thinking that the strongest case for faith is essentially utilitarian: it is better, for individuals and for societies, if we are religious. The decline of traditional religion, we are told, drives people toward worse alternatives, where political ideologies or lifestyle cults become quasi-religious. Surveys report that, on average, more religious people say they are happier and healthier. They tend to have more children and seem less prone to the diseases of modern anomie, like loneliness and despair.

But for a genuinely religious person, this cannot be enough. A purely utilitarian defense makes faith look like a noble lie, a useful fiction we adopt not because it is true but because believing it produces good outcomes. This defense collapses the moment one encounters people who have found happiness and prosperity without being religious—and such people are not hard to find. To defend religion as not merely useful but true, something more is required.

Ross Douthat, the New York Times columnist, tries to offer that in his book Believe: Why Everyone Should Be Religious. “[It] is the religious perspective,” he writes, “that has the better case by far for being true.”

Douthat starts with the fine-tuning argument: the idea that the universe appears designed for life, since it would not exist in its current form if the constants of nature were even slightly different. He moves through the mystery of consciousness and the explanatory gaps that materialism cannot seem to close—arguments that have been made for centuries, respectable though they have equally respectable replies. More interesting is his emphasis on mystical experiences, encounters with the supernatural that persist even among educated people. Disenchantment, he suggests, was always more ideology than description; even secular moderns keep having experiences that resist a strictly materialist worldview.

There is something shrewd in this approach. We live in a time when the news is often stranger than fiction, when the confident pronouncements of experts have been humbled again and again, and when physics reveals a reality far stranger than common sense would allow. In such an environment, why not take seriously the insights that billions of people across thousands of years have kept arriving at, the persistent human sense that there is more to reality than atoms in the void?

And yet it is hard to imagine this book persuading many people on rational grounds. The god-of-the-gaps arguments remain vulnerable to their familiar objection: unexplained phenomena are not evidence of the divine, only evidence of what we do not yet understand. The appeal to inexplicable and mystical experiences proves both too much and too little—too much because every religion and pseudo-religion can claim its own miracles, too little because personal experience, however vivid, cannot serve as public evidence. And when Douthat faces the genuinely difficult questions—why a benevolent God permits the suffering of innocents, for instance—the rational apparatus begins to falter. We cannot judge how good and evil are balanced in the cosmos, he suggests, because we lack the capacity to comprehend the whole. This is an ancient answer, and he does not make it more convincing.

What the book actually demonstrates—perhaps despite itself—is that the utilitarian case remains the strongest one available. Douthat is at his most persuasive when he describes the decadence of modern secular life, the loneliness and purposelessness that afflict so many. If your life feels meaningless, religion will give you meaning. If the path you have chosen leads to declining birth rates and civilizational ennui, religion offers a remedy. This is compelling. It is also, still, a utilitarian argument.

The deeper problem, which Douthat touches on in his work but does not fully confront, is that modernity’s greatest assault on religion may not be intellectual at all. It is not simply that modern science has produced arguments against God. It is that modern life has transformed the material conditions under which belief naturally flourishes.

Organic, unforced religiosity has tended to arise under certain conditions: an intimacy with the elements and with unjustifiable suffering and painful, premature death; lack of optimism about material progress in this world; a comfort with things rough and elemental; an unquestioned adherence to inherited duties and authorities. These conditions are now rare—certainly in the developed world, and increasingly in much of the developing world as well. We are buffered from premature death, insulated from nature, liberated from tradition, educated into naturalism, habituated to progress. It is telling that even theocratic societies, where religion retains political power, are increasingly marked by the same symptoms that afflict secular ones, such as falling birth rates. The problem, it seems, is not simply secularism. It is modernity itself.

This is the unspoken melancholy beneath the project of Douthat and his fellow travelers. They are trying to reason people back into faith in an age when the conditions that make faith feel natural have largely vanished. But if the main obstacle to belief is not bad arguments but material comfort, then the path back to faith may not run through better apologetics. It may run through catastrophe. If what dimmed the religious instinct was progress, then what might rekindle it is the undoing of progress—a prospect that even those who long for belief to flourish can hardly welcome.

Keep Reading:
→ Review of Tyler Cowen’s Stubborn Attachments
→ Michael Shermer on Morality and Science (podcast)
→ A Conversation with Garett Jones on Democracy

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (curate your culture — iOS app)

I’m not a Buddhist, nor did I grow up in any Buddhist tradition. But I have always been fascinated by it. These are some reflections based on notes I made during a retreat last year.

At the heart of Buddhist thought is the denial of substance. How do I describe myself? I might point to my body, my thoughts, my memories, my habits. But is there a “me” that exists somewhere behind all these characteristics, some essential underlying core that would remain if you stripped everything else away?

Or think about a table. We can describe its color, its hardness when we touch it, the sound it makes when we knock on it. But is there a “table” above and beyond these properties, some inner essence that makes it truly, permanently a table?

According to Buddhist thought, this kind of inherent, fixed existence simply isn’t there. What we think of as solid things are actually collections of characteristics and relationships that are subject to constant change, a continuous series of becoming.

This is reminiscent of British empiricist philosophers like John Locke and David Hume, who made similar arguments about substance. Hume even applied this logic to causation itself: do we actually perceive one thing causing another, or do we just see one event followed by another? Just as we never find an inherent “self” when we look inside, we never actually see causes, only sequences of events.

The concept of impermanence in Buddhism is related to this absence of fixed essence. Imagine you meet someone and feel a strong attraction. In that moment, you experience this person as a solid entity, someone in whom your future happiness resides. But she isn’t a fixed entity, nor are you. More importantly, the attraction itself isn’t some absolute truth about the universe. It’s a relationship between two sets of characteristics at a particular moment in time, in a particular place. You might lose the attraction after getting to know her better, or after getting to know yourself better. The happiness you imagined finding “in her” might vanish, or even transform into suffering if you insist on pursuing it still.

Neither you nor she nor the attraction nor the happiness you seek can remain independent of everything else. They’re all contingent, arising from conditions, changing as conditions change. The fixed, solid, graspable thing we’re looking for doesn’t exist. This is true of all experiences, pleasant and unpleasant.

Often things don’t turn out as we expected them to, and that creates dissatisfaction. Sometimes they turn out exactly as we wanted, and still we find ourselves wondering if this is all there is.

We experience a pleasant state, and somewhere in the back of our minds we know this feeling can’t last in quite this form. When we try to prolong it, we end up destroying the very thing we tried to make permanent. This is the fundamental unsatisfactoriness of conditioned existence, what the Buddha identified as dukkha, the first of his four truths for the noble.

What causes this? The second truth identifies ignorance as the root. We attribute to things qualities they don’t have. When we’re attracted to something, we exaggerate its positive qualities and overestimate how happy it will make us. When we’re repulsed by something, we magnify its negative qualities and overestimate how miserable it will make us. We fail to see things as they really are. From this ignorance arises craving, the immediate cause of suffering.

It is often said that the Four Noble Truths were inspired by the Indian medical system, with dissatisfaction viewed as a disease. The possibility of cure is just as important as its diagnosis: it makes a world of difference whether a physician says “there is no known cure” or “there is a cure, and here’s what it involves.” The existential equivalent is the third truth: Nirvana is possible.

Nirvana literally means “blowing out”, but it isn’t the same as annihilation. Indeed, the Buddha described the desire for nonexistence as another craving. At the same time, he suggested that attempts to describe its nature would only misrepresent it.

In Buddhist meditation, you’re asked to observe your thoughts and emotions without judging them or getting swept away by them. As you stop feeding the flames through identification and attachment, the false sense of a fixed self begins to dissolve. You start seeing things as they actually are, and in that clarity there’s peace.

The path to this state, the fourth truth, involves ethical living, the search for knowledge, and practices of mindfulness.

I broadly agree with all of this, with one clarification: I think most of us spend much of our time in a kind of inertia, experiencing neither dissatisfaction nor happiness. This observation might not necessarily contradict Buddhism.

What I do find difficult to accept are certain elements of traditional Buddhist cosmology. In classical Buddhism, the vast disparities we see among people are explained through karma and rebirth. Why is one person born with a certain natural ability while another struggles? Why does one child grow up in comfort while another faces poverty and hardship? The traditional answer involves actions from past lives creating the conditions for this one. Traditional cosmology also includes realms we can’t perceive: worlds of demigods, hungry ghosts, various heavens and hells.

The problems these doctrines were meant to solve, like explaining inequality, can be addressed through other frameworks. For instance, a more creative interpretation of “past lives” would be to see them as the lives of our ancestors. Their experiences do, in a sense, transmit to us. Our genetic heritage accounts for differences in characteristics and capacities. The circumstances of where and when we’re born determine the privileges and hardships we encounter. The cultural knowledge, suffering, and wisdom of previous generations shape the world we inherit.

We find ourselves as participants in a vast chain of events that is both part of ourselves and outside us, stretching back through our lineage and forward through the impact we have. This interpretation also helps address a problem that has plagued Buddhist philosophers: if there is no persistent self, then who is reborn? As for the ethical implications, they remain largely similar.

In the end, I conclude that the Four Noble Truths (suffering exists, it has a cause, liberation is possible, there’s a path) and the Four Seals of Dharma (all compounded things are impermanent, all conditioned experiences involve unsatisfactoriness, all phenomena are empty of inherent existence, Nirvana is peace) are among the most profound ideas humanity has produced.

Keep Reading:
→ The Inescapability of Altruism
→ Solving the Trolley Problem and Other Moral Dilemmas
→ Michael Shermer on Morality and Science (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (Curate your culture — iOS app)

Most of us spend most of our time thinking about our own lives. We want to be comfortable, healthy, successful, and respected. We care about the well-being of people close to us. We may perform acts of charity or experience indignation on occasion, but only when we feel compelled. In the absence of such compulsion, the suffering of a stranger does not register as a priority. It’s not that we wish them harm; we just don’t think about them much at all.

And yet moral systems keep insisting we should. We’re told each person’s well-being matters equally, that we ought to expand our circle of concern far beyond ourselves and our loved ones, even to people on other continents and to future generations. Any curious and undogmatic person would naturally wonder: why? We seem to recognize how unrealistic and perhaps incoherent this demand is.

If we wanted to justify altruism as an end in itself, then ultimately I think there are only two kinds of answer. The religious answer says God approves it, so the obligation to promote happiness arises from his authority. The non-religious answer, as discovered by those like Henry Sidgwick who made serious attempts to find it, is that if you reflect carefully enough, you’ll just see that everyone’s good matters equally: it’s self-evident, like mathematical truths. If you don’t see it, there’s no further argument that can force you to change your mind.

To find the first answer satisfactory, you must already believe in divine judgment and authority. For the second, you must already feel the force of this supposedly self-evident moral truth. And if you don’t? The arguments have nowhere else to go.

A different approach would be to question altruism as an end in itself. Instead of asking “why should I care about others?”, we ask “what does it mean to truly care about myself?”

Imagine a small society of one hundred people. They are ordinary people with normal ambitions. They care about their families. They enjoy status and comfort. But each person has developed a stable disposition, whether through cultural conditioning, emotional or rational persuasion, systems of rewards and punishment, or something else: all their actions reflect genuine consideration of the well-being of the other ninety-nine.

What’s life like in this world for any random person? From his perspective, ninety-nine other people are constantly working for his welfare, either by performing certain actions or by refraining from certain actions. When he is sick, people help. The water that enters his pipeline is not poisoned. When he walks on the street, vehicles make way for him. When he has an idea, others help in turning it into reality. The institutions around him actually function because people contribute even when it’s not immediately profitable. Trust is high, which means opportunities are abundant.

Compare this to a society where everyone is narrowly self-interested. Here, cooperation is fragile and strategic. He can’t trust anyone to follow through unless it serves their immediate interests. Public goods deteriorate because no one wants to pay for them. He spends enormous energy protecting himself from harm and exploitation.

The clever egoist might ask: “Fine, but why can’t I be the exception? Why can’t I live in the altruistic society, preaching whatever myths are needed to sustain it and sanctioning those who talk about the virtue of selfishness, while quietly pursuing only my own advantage?” The most fundamental problem is that if this strategy actually worked and others followed your example, the very environment you’re trying to exploit would disappear. You’d be undermining the conditions of your own flourishing by doing to others what you don’t want done to you.

The external case is simple: your life goes better in a world of altruists, and you can’t sustain that world while refusing to be one yourself. The amount of good that can happen to you grows with the size of your society. If there are a million people, then 999,999 other people are working all the time for your welfare. One of them might invent the medicine that saves your life, even if she lives on the other side of the planet or in a different time period.

But even if we look only at our inner life, the purely self-interested strategy fails.

Consider what it’s like to live with your attention locked onto your own advantage. You’re constantly comparing yourself to others: Am I getting what I deserve? Why is she more successful? Did I get cheated in that interaction? This stance breeds anxiety, envy, resentment, and a perpetual sense of precariousness.

Now contrast this with caring about others’ well-being. It shifts your attention away from anxious self-monitoring and toward concrete tasks: helping, building. That alone is often a relief. When you are absorbed in a task and have forgotten yourself, that’s often when the most profound happiness arises. Such selflessness opens up sources of satisfaction that simply aren’t available to the self-focused person.

There is a famous Buddhist saying: All the happiness in the world arises from wishing others to be happy, all the suffering in the world arises from wishing ourselves alone to be happy. Joseph Butler made a similar point while trying to refute simple egoism: happiness is a by-product of the pursuit of things other than happiness itself. This is a structural fact about us: directly pursuing our own good is a self-defeating strategy.

So altruism is not a noble sacrifice, a moral duty imposed from outside, or a principle of pure reason independent of consequences. Once you think through what it means to live well as the kind of creatures we are, altruism stops looking optional. One can doubt religious commands. One can be skeptical of the moral intuitions upon which utilitarianism ultimately rests. But one can’t ignore one’s own well-being. And if we ask ourselves honestly, we will find that in all questions about what makes a good life, altruism is already embedded in the answer.

Happy season of giving to you all, and thank you for being part of the community here at Vatsal’s Newsletter!

Continue Reading:
→ Review of Tyler Cowen’s Stubborn Attachments
→ Solving the Trolley Problem and Other Moral Dilemmas
→ Michael Shermer on Morality and Science (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

This might sound trivial if you already agree, and too reductive if you don’t. But I want to make and defend the claim that every problem we face is ultimately a problem about making the right predictions. Whether you’re deciding where to invest, whom to hire, which medication to take, or what to say to a friend who’s upset; whether you’re a theoretical physicist, an engineer, a historian, or a theologian; whether you’re a child, a teen, or an adult—whatever the nature of the problem you’re dealing with, it is fundamentally a prediction problem. Consequently, the better your predictions, the better your solutions.

Why might such a straightforward, commonsensical view be contentious? In fact, throughout history, many influential thinkers have explicitly rejected it. From Plato on beliefs to Albert Einstein on quantum physics to Noam Chomsky on contemporary AI, philosophers and scientists have insisted that knowledge requires something more than true predictions. And they’ve had good reasons for saying so. Lucky guesses clearly aren’t knowledge. Shallow pattern matching breaks when the world shifts, as imperfect AI systems remind us.

But I want to suggest the critics have misdiagnosed the problem. They’re pointing at something real, but misunderstanding its nature. What they’re identifying isn’t an alternative to prediction. Instead, it’s what helps us make more predictions, and make them more successfully.

The objection goes back at least to Plato. In the Theaetetus, Socrates considers whether knowledge might simply be true belief, and rejects the idea with a thought experiment about a jury.

Imagine jurors deliberating a case. They weren’t eyewitnesses to the crime. They’ve heard testimony, some of it persuasive rhetoric rather than solid evidence. Suppose they reach a guilty verdict, and they happen to be right. The defendant did indeed carry out the crime.

But did the jurors know he was guilty? No. They had a true belief, but something important was missing. They reached the right conclusion through a process that could easily have led them astray. A more charismatic defense attorney might have swayed them the other way. They were “justly persuaded”, perhaps, but rhetorical persuasion isn’t the same as knowledge.

We feel this intuition immediately. If we had to predict whether the jurors would get the next case right, we wouldn’t trust them. Their success doesn’t seem systematic or reliable. It seems contingent on who spoke most eloquently that day.

For more than two thousand years, this has been the standard objection to identifying knowledge with correct prediction: you can get the right answer for the wrong reasons. And when that happens, you don’t really know.

What, then, is missing? The standard answer is that the jurors lacked understanding or explanation—something more than mere successful prediction.

But I would argue that what the jurors really lacked was a method that would continue to work, not something mysterious called “explanation” whose nature is never made clear. The problem is that the method they used to arrive at their prediction wasn’t reliable. And what makes prediction reliable? Precisely the things critics point to: understanding what we call underlying structure, grasping what we call causal relationships. The critics aren’t wrong to insist that knowledge requires more than randomly getting the answer right. They are wrong about what that “more” consists of.

Thinkers like the philosopher Karl Popper and the physicist David Deutsch argue in favor of explanation over mere prediction. But explanation matters only to the extent that it extends our predictive success. We never observe causes, as David Hume pointed out, but causal understanding still matters if it tells us what will happen when circumstances change. Knowing what’s impossible matters not because it’s real (we never observe it, since it doesn’t actually occur), but because it rules out paths in advance and thus helps refine predictions about what does occur. These are not alternatives to prediction. They are methods that help us attain greater success with our predictions.

Notice the asymmetry. We never observe causal necessity or explanatory structures; we infer them from what leads us to more successful predictions. Predictions, by contrast, are concrete: the event either happens or it doesn’t. This gives prediction a direct contact with reality that explanation, by itself, lacks. Explanation earns its keep only by extending predictive success.

Our quest for predictive success may necessitate the use of these methods. As Ilya Sutskever, co-founder of OpenAI, notes, when you train a neural network to successfully predict the next word, it doesn’t just learn statistical correlations; it learns “some representation of the process that produced the text”. The pursuit of prediction, pushed far enough, generates what we describe as explanation.

But if we take explanations as the essence of knowledge, if we see explanations as the end, we run the risk of obscuring our ignorance and producing a satisfactory subjective state through narratives that we mistake for knowledge. Reliance on predictions helps mitigate this risk, as demonstrated by the success of modern science, which delivers predictions that pre-modern natural philosophy, despite its elaborate explanations, failed to provide.

Every problem is a prediction problem. The question that matters is how far our successful predictions go, and how to make them go further.

Continue Reading:
→ A Response to Noam Chomsky on Machine Learning and Knowledge
→ A Conversation with Martin Frické on the Epistemology of Machine Learning
→ A Conversation with Massimo Pigliucci on Scientific Explanation

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

When someone says, for instance, that God exists, one way to proceed is to first establish what “God” means and what it is to exist. While this seems straightforward, we know from experience that it rarely is. People associate different meanings with the word “God”. Even more complex is the property of existing. Is it necessary, for something to exist, that it be tangible and visible? Does a thing not exist if it is not observable, like a preference that is held but never revealed? Different underlying commitments add further difficulty in reaching consensus over even basic definitions.

Now consider a different approach. Instead of attempting to settle on definitions, we begin with specific, self-contained claims. We then reduce any assertion about a phenomenon to a family of claims, each of which can be evaluated independently as true, false, indeterminable, or meaningless. One claim concerning the existence of God could be “when humans distance themselves from traditional religion, they invite destruction through worse substitutes.” This claim is either true or false. A separate claim could be “God dwells in heaven, which exists beyond this realm,” which is indeterminable.

Debates on free will notoriously fail to move beyond disagreements over definitions. It is therefore an ideal candidate for this approach. With the rise of AI, the question has acquired a new practical context: Can an AI be held accountable? Should it have rights? These questions presuppose answers to whether AI systems can possess what we call “free will”. Rather than attempting to agree on a definition of free will, let’s consider specific, self-contained claims about the phenomenon, from general ones to those specific to AI, and evaluate their truth.

1. All events are inevitable.

This claim is trivial as stated, in the sense that we cannot imagine any event that could falsify it. It is true in the same way as the claim “all triangles have three sides” is true. No matter which part of the world or time period we consider, we will never find a triangle without three sides. Likewise, no matter what occurs in the actual world – whether I stand up or remain seated, whether a dog barks or maintains silence, whether an AI restricts some information or reveals it – no occurrence past or future could disprove the claim that all events are inevitable.

Traditional claims about determinism give the impression of being substantive. But when we ask what the person who makes that claim knows about the actual world that the person who denies it does not, we struggle to find anything. In fact, he need not possess any particular knowledge of the world to make the inevitability claim successfully. In effect, he is not saying anything more than “events that occur, occur, and events that do not occur, do not occur.”

In many discussions of free will, the issue is framed by asking whether a person “could have done otherwise” in exactly the same situation, bringing in talk of alternative possibilities and other ways the world might have gone. But it is not clear what status we should assign to things that do not occur, beyond the fact that they are not real. The metaphysics of modality deserves its own treatment, which I will not attempt here. For now, I will only note that whatever argument is given to prove the inevitability of all events, in no case can the person show that he knows anything more about the actual world than his opponent.

2. No one can be held responsible for their actions.

When we talk about free will, we tend to assume the existence of an independent, inherent self who can have it or not have it, and with which we associate some deeper sense of responsibility beyond relationships between events. We think that the thing which has free will is an “I” which exists above and beyond the fabric of our skin and muscles, the bodily processes, the thoughts that form, and so forth. This view is challenged by Buddhism, by empiricist philosophy, and by modern science.

As Bertrand Russell put it, when we look at Mr. Smith, “we see a pattern of colors; when we listen to him talking, we hear a series of sounds. We believe that, like us, he has thoughts and feelings. But what is Mr. Smith apart from all these occurrences? A mere imaginary hook, from which the occurrences are supposed to hang. They have in fact no need of a hook, any more than the earth needs an elephant to rest upon…it is a collective name for a number of occurrences. If we take it as anything more, it denotes something completely unknowable, and therefore not needed for the expression of what we know.”

The self is thus not a substance that performs actions, but a “convenient way of collecting events into bundles.”

Once we recognize this, we can see that responsibility must be located not in some unchanging entity but in the relationships between states of affairs themselves. In the substance view of the self, we say that the thing responsible for an event, such as a light bulb turning on, is an entity: me. But in fact it is not I who turned the bulb on, but my pressing the switch. In other words, it is not an entity but another event that can be held responsible for the event that occurred.

Further, as David Hume demonstrated, there is no observable “force” of causation. Therefore, responsibility can only refer to certain relationships of precedence and succession between events. When we say one event is responsible for another, we mean that they stand in a particular relation to each other.

In cases of manipulation, we would say, for instance, that my lying to her was responsible for her forming a mistaken judgment, or that my injecting corrupt data was responsible for the AI’s making a mistaken judgment.

Seen this way, we can establish responsibility for every event that occurs. As a matter of fact, only another event can be held responsible for an event. It is merely for our convenience that we describe the event held responsible in terms of an entity, human or otherwise.

3. AI systems cannot be held accountable in the same way humans are, as they lack a unified or embodied self that is capable of reward and punishment.

Since it is for our convenience that we think in terms of entities rather than events, it is worth considering why we do so. The obvious advantage is instrumental: there are certain benefits that can be derived by thinking and acting in terms of persons and holding entities responsible rather than events.

An easy way to understand this is to consider the usefulness of treating a corporation as if it were a person: a thing that is subject to creation and death, pays taxes and receives benefits, can be healthy or sick, is able to sue and get sued, can be controlled or get out of control, can be rewarded and punished, and so forth. If we tried to locate this corporate entity, we wouldn’t find it anywhere, just as we never find a “me” in any part of our body. The corporation is a useful fiction, a way of bundling together events and relationships that serves specific practical purposes.

To the extent that similar benefits are derived from holding AI systems responsible at one level rather than another, it serves the same purpose as holding a person or corporation responsible at a particular level. The norms of accountability that we have historically applied to humans, animals, corporations, and even deities can be understood in terms of these practical benefits.

Seen in this light, the debates about AI alignment are essentially a recent version of patterns we have seen before. In the late eighteenth and nineteenth centuries, for instance, markets came to be understood as a self-regulating device, an institutional mechanism that organized people’s lives through prices and incentives, with a logic of its own that no individual controlled. Later, the modern corporation emerged as an economic organism whose internal goals and powers could rival those of the state, again not reducible to the psychology of any one manager or shareholder. The Progressive Era struggles over monopolies and corporate regulation were, in effect, attempts at alignment: how to steer these large, semi-autonomous systems so that the patterns of events they produced did not simply follow the narrow logic of profit and competition, but remained compatible with the broader interests of citizens. Advanced AI systems can be seen as the latest instance of this familiar problem.

4. AI systems cannot truly deliberate like humans; their responses are automatic like those of animals.

Viewed up close, the neurological activities of a human during rational deliberation would appear as electrochemical impulses, no different in kind from automatic processes. When viewed from a distance, the instinctive responses of an animal reveal themselves to be part of something more complex and adaptive. So in what material way do deliberation and instinct differ?

One bias we have is to associate effort with intention and struggle with agency. We tend to value things more highly when greater effort goes into making them, irrespective of whether they are actually better. Suppose one person learns piano easily, while another struggles, painstakingly making individual associations and movements. Does the latter have any special claim to the quality of the music? Clearly not. The same applies to someone who arrives at a judgment through lengthy deliberation compared to another who arrives at it quickly. The former does not have a privileged claim to being right. It may simply be that the second person is more gifted or experienced.

The value of deliberation lies in the fact that it often leads us to better judgments in important matters, where “better” is judged by how appropriate the judgment proves to be, not by the process that produced it. Haste usually implies error in such matters, but this is a statistical correlation, not a metaphysical distinction. When a person deliberates, it does not inherently indicate better judgment, even though it often does. Our norms reflect this practical reality, and so we treat deliberation as relevant to accountability. But objectively, there cannot be anything inherently unique or superior about what we call deliberation. If an AI arrives at appropriate judgments without apparent struggle or delay, and we find ourselves questioning its capacity to deliberate, this reveals something about our biases, not about the AI’s lack of agency.

5. AI systems lack consciousness.

The truth of this claim, like all others, depends on how certain concepts are understood. In other words, depending on how we define consciousness, we end up with different, independent claims.

By consciousness, I do not mean merely the state of being active, like industrial equipment, or merely the capacity to experience emotions, as animals do. John Locke defined consciousness as “the perception of what passes in a man’s own mind.” Consciousness is a form of perception, and perception implies knowledge. When you perceive a tree, you form a representation of something external to your mind. When you are conscious, you form a representation of something internal: your own mental states and processes. This distinction suggests consciousness is not metaphysically unique; it is simply knowledge of a particular kind.

Consider what happens when you catch yourself getting angry. You notice the rising tension. This noticing is not the anger itself; it’s an awareness of the anger. You have formed knowledge about your own mental state. Now consider someone who becomes angry but lacks this awareness; they simply act on the anger without recognizing it. We say the first person has awareness of their anger in a way the second person does not, and this is because the first person possesses certain knowledge that the second lacks.

If consciousness is a kind of knowledge, how is such knowledge acquired? The same way all knowledge is acquired: by synthesizing data, by forming compressed representations that manifest as successful predictions.

Thus, to say that an entity is conscious is to say that it possesses certain knowledge that an entity lacking consciousness does not. Consciousness is, in principle, accessible to any entity capable of acquiring knowledge about its own states and processes, regardless of whether that entity is biological, capable of emotions, or human. This additional knowledge may yield practical advantages, but this is not necessary. There is nothing in principle that prevents an AI from being conscious.

6. AI systems lack control over their actions.

By control, I mean the ability to regulate something either according to a teleological pattern or according to one’s conscious intention. These are two distinct senses in which control is usually understood.

In the first sense, teleological control, we say, for example, that the sphincters in the bladder and urethra control the outflow of urine. They fulfill a purpose. When they malfunction, they fail to fulfill this purpose and thereby fail to exert control. This is control understood as functional regulation toward an end. Many machines possess control in this sense. The question is not whether such systems have purposes, since they clearly do, but whether these purposes are “real” in some deeper sense.

This leads us to ask whether purposes must be self-generated to count as genuine. But on reflection, human purposes are no more self-generated than AI purposes. We do not generate our own desires from nothing; they arise from biological imperatives, cultural conditioning, prior experiences, genetic dispositions. The fact that we cannot trace our desires back to some uncaused origin – a notion that is itself conceptually incoherent or unknowable – does not make them less real or less effective in organizing our behavior.

In the second sense, conscious control, we say that a person has control over his bladder because his conscious intentions align with his ability to regulate it. To the extent that he lacks awareness of his intentions or is unable to regulate the release according to those intentions, he lacks control. This form of control depends on consciousness, and as we noted in the previous claim, consciousness is a form of knowledge.

In the real world, we find that people have varying degrees of control over different things at different times. A person driving a vehicle may have control over its speed and direction at one moment but lose it later due to a malfunction in the vehicle or a muscle spasm in his body. Control is thus not an all-or-nothing property but a matter of degree and context.

If an AI possesses the relevant self-knowledge and if its internal states successfully regulate its behavior toward ends, then it possesses control in both senses. Whether any current AI systems meet these criteria is an empirical question. But nothing in principle prevents AI systems from having control, any more than anything in principle prevents them from having consciousness or engaging in deliberation.

Conclusion

The list of claims I have considered here is far from exhaustive, and different conceptions from those I have used for terms like “consciousness”, “responsibility”, or “control” would create their own, independent claims. But the method suggested here is, I believe, the best way to explore the possibility of free will in AI, as it is for humans.

Continue Reading:
→ Free Will
→ A Conversation with Martin Frické on the Epistemology of Machine Learning
→ Michael Shermer on Morality and Science (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

The trolley problem is a well-known thought experiment in moral philosophy. You’re standing at a railway switch. A runaway trolley barrels down the tracks toward five workers who will certainly die if it continues. You can pull a lever to divert it onto a side track, but there’s one worker there who will be killed instead. What do you do?

This is an example of a moral dilemma, variants of which appear throughout philosophical thought experiments and in fiction. In Sophie’s Choice, there is an even more harrowing scenario: a mother in a Nazi camp is told she must choose which of her two children will be killed. If she refuses to choose, both will die.

Once you’re in such a situation, there is no good answer—if by “answer” we mean something that would satisfy our usual expectations of an answer. Every option involves a tragic outcome that will haunt you so long as you have memory and a healthy conscience. Philosophers argue about whether there are “genuine” moral dilemmas, cases where, even by the lights of the best moral theory, you cannot avoid doing wrong. While we can debate this, it is clear that no amount of ethical reasoning will transform a terrible situation into anything other than what it is.

But if we are interested in solving these dilemmas in the real world, I think we’re asking the wrong question. The real ethical work happens long before anyone is standing at that lever. The most important form of moral progress isn’t getting better at choosing between bad options but creating a world where such options arise less often. The task is not to resolve dilemmas, but to dissolve them. Historically, this is exactly how humanity has addressed many of the moral dilemmas that plagued previous generations.

Take hunger. There are two broad ways to respond.

The first is reactive. A person is hungry, someone provides aid, and the immediate problem is solved.

The second is more systematic. We know people will inevitably become hungry given human biology, and we know hunger creates all sorts of problems like theft and resentment. So we build structures that let people obtain food reliably: agriculture, markets, supply chains, research and development, social safety nets. When someone is hungry, food is available, and there are predictable ways for them to get it.

Both approaches address hunger, but only the second makes it non-recurring. The first relies on charity; the second builds civilization. It’s the difference between giving money to a starving man so he doesn’t commit a desperate crime, and creating an economy in which he can use his talents and be rewarded for them.

Call these two approaches resolution and dissolution. Resolution addresses problems after they occur; dissolution redesigns the background so they rarely arise.

If we reflect on moral progress, we find that our greatest achievements follow the dissolution pattern. Consider slavery. We didn’t solve slavery by making it more humane or by discovering better answers about who should be enslaved. We dissolved it by building societies where the institution became unthinkable, through multiple reinforcing changes: economic development that made free labor more productive than forced labor, moral and religious movements that delegitimized the practice, legal reforms that criminalized it, and education that made its injustice self-evident to subsequent generations.

Similarly, we ended the practice of dueling over honor disputes not by making duels fairer but by developing legal systems and social norms that rendered them obsolete. And we’ve dramatically reduced famine not mainly through emergency aid or appeals to the conscience of the rich but through science and technology, economic growth, and better political institutions.

Sophisticated civilizations don’t merely get better at managing recurring crises. They engineer those crises out of existence. The ethical question, then, is not “What should I do when confronted with an impossible choice?” but “What world should we build so that fewer people face impossible choices at all?”

Consider climate change and economic development. The usual framing pits environmental protection against prosperity, developed nations’ historical emissions against developing nations’ right to grow. We’re told to accept painful trade-offs: sacrifice growth, lower living standards, or decide which countries get to develop and which must bear the burden of reduced emissions. We treat the trade-off as fundamental.

The dissolution approach rejects this framing. Instead of arguing over how to divide a fixed pie, we focus on technological abundance and institutional arrangements that eliminate the trade-off: clean energy that’s cheaper than fossil fuels, carbon capture that’s economically viable, agricultural methods that increase yields while lowering emissions. The work of ethics here is continuous with that of innovation and institution-building, so that future generations inherit both a livable planet and material abundance.

Or consider autonomous vehicles, which revive the trolley problem in a new, concrete form: how should a self-driving car be designed for cases in which a crash is unavoidable? Should it kill one person instead of five? We can debate endlessly how to code these trade-offs. The dissolution perspective says: the real moral achievement is in designing vehicles and infrastructure that make such edge cases vanishingly rare.

This shift in perspective changes where we direct our attention and resources. Instead of focusing primarily on guilt and retribution, we invest in technological development and institutional design. Instead of treating trade-offs as inevitable, we ask when they are truly hard constraints and when they are symptoms of a world that could have been built differently.

Why, then, do we favor resolution over dissolution? To the extent we do, it is in part for the same reason we find ourselves more emotionally moved by a philanthropist who gives away large portions of his wealth than by a civil servant who designs a boring tax reform that quietly lifts millions out of poverty. Our moral imagination likes vivid, dramatic acts of sacrifice better than slow, structural work.

Our task is to get better at building a world with fewer terrible choices, rather than only at making heartbreaking decisions when they arrive. Philosophical attention to tragic dilemmas can be illuminating, but when we start to treat wrestling with impossible choices as the essence of moral life, we risk treating suffering itself as morally important, as a source of meaning and redemption, rather than as a problem to outgrow.

Continue Reading:
→ Review of Tyler Cowen’s Stubborn Attachments
→ Overcoming the Naturalistic Fallacy
→ Michael Shermer on Morality and Science (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

The Bloomsbury Group was a circle of writers, philosophers, and artists who met at the homes of the novelist Virginia Woolf and her relatives in the Bloomsbury district of London. The group is remembered for the large number of significant 20th-century figures attached to it. In addition to Woolf, the economist John Maynard Keynes and the novelist E.M. Forster were some of its members. Their discussions focused on the nature of beauty, goodness, and truth, where their core beliefs were shaped by G.E. Moore. Moore’s influence was profound, almost incredible in retrospect. Keynes described his Principia Ethica as a “stupendous and entrancing work, the greatest on the subject”. “I almost worship him as if he were a god,” wrote a young Bertrand Russell, philosopher and Nobel laureate, “I have never felt such an extravagant admiration for anybody.”

The fundamental ethical value, according to Moore, is goodness. But goodness is indefinable. Moore rejected any attempt to reduce goodness to a naturally occurring property such as pleasure or happiness. In Principia Ethica, he famously wrote: “If I am asked, ‘What is good?’ my answer is that good is good, and that is the end of the matter. Or if I am asked ‘How is good to be defined?’ my answer is that it cannot be defined, and that is all I have to say about it.” We know the good ultimately through a direct apprehension, through intuition. When someone asks why it would be wrong to torture a baby if the act were sanctioned by religious beliefs or yielded consequences that maximize overall utility, and you reply, “Because it just is!”, you are invoking your intuition.

Against those who argue that good is the same as, say, pleasure, Moore used what is called the open-question argument. When “bachelor” is defined as an unmarried man, and one asks “Is an unmarried man a bachelor?”, it is like asking “Is a bachelor a bachelor?”. Once we know that someone is a bachelor, we can be certain he is unmarried, so the question is “closed”. But the same is not true when we try to define “good” in natural terms. Suppose we say, “Good means pleasant”. The question “Is what is pleasant good?” still seems meaningful. Once we have established that a thing is pleasant, we can still doubt whether it is really good, so the question remains “open”.

Moore maintained that this would be true of any natural property used to define goodness, such as desire, well-being, or being evolutionarily advantageous. We might think that preventing climate change or defeating a corrupt politician is good because these actions would prolong our survival, and we may be right, but is prolonging our survival identical to goodness, or does it only possess the independent property of goodness in the way that my trousers possess the color black? As in Plato’s theory of Forms, individual instances of good things partake in goodness, while goodness itself exists somewhere in the Platonic realm beyond the natural world. Attempts to define goodness through a natural property constitute a fallacy: the naturalistic fallacy.

If Moore is right, then the scientific enterprise, which has been extraordinarily successful in bringing under a naturalistic framework so much that was previously seen as mysterious or immaterial, encounters a major setback. Here’s an area where the scientific method can’t help us, since no observation or experiment can settle what is good. We would have to rely on a mysterious thing called “intuition”. Moore’s own intuitions led him to see inherent value in friendship and the appreciation of beauty. What if someone sees inherent value in social justice and non-violence? How does one resolve fundamental differences that arise from conflicting intuitions? We can at best reach consensus, but that is not the same as establishing the truth.

Whatever one thinks of Moore’s arguments, which have been criticized by many philosophers, the general idea that ethical conclusions cannot be derived from anything not itself ethical remains a serious problem. It is worth seeing if we can overcome it. I believe we can. I will argue that certain properties have both subjective and objective dimensions simultaneously, and that goodness is one such property. It seems mysterious and appears non-natural so long as our understanding of its constitution remains incomplete.

To understand what I mean, let’s consider another property, similar but less contentious: physical largeness. Suppose we apply the open-question argument to it. If someone defined “large” as everything over 4 feet in length, we could still ask “Is everything over 4 feet in length large?” The question would be meaningful; the properties are not identical. In fact, we can easily imagine many objects less than 4 feet in length that are still large (e.g. a large pencil), or objects over 4 feet in length that are not (e.g. a small rocket). The same would be true for any size that is proposed. Following Moore’s logic, we would have to conclude that largeness is a simple property irreducible to, and underivable from, any natural property.

This cannot be right. Largeness is a natural property that we commonly recognize and treat as such.

The ambiguity arises because “largeness” refers both to an objective dimension of a body, measured in units like inches or feet, and to a subjective dimension, determined by a range we have in mind that is specific to the kind of thing in question. It is for this reason that a pencil and a rocket can both be large, despite one being much larger than the other. We can be more precise by getting rid of the subjective dimension altogether in our description. Instead of calling it large, we can simply say, “The pencil is 8 inches in length”.

Using this analogy, we would have to find out how something we call “good” can be described without its subjective component, just as we described something we call “large” without appealing to subjectivity. If we can do this, it would allow us to overcome the naturalistic fallacy.

A characteristic of goodness that might seem counterintuitive at first is that a thing is always good for someone. Nothing is ever good without being good for an entity capable of experience. Nothing is ever good from “the point of view of the universe”, an expression used by the classical utilitarian philosopher Henry Sidgwick to justify an impartial, agent-neutral standpoint from which each person’s happiness counts equally. Moore, a student of Sidgwick, adopted it in his moral philosophy. In our own time, the philosopher Peter Singer defends the idea in a book with that title. But it seems to me that referring to a thing as good, without specifying the particular subject for whom it is good, is an incomplete description. Even when the same thing is good for both you and me, it is still a set of individual experiencers for whom it is good, and convergence is not guaranteed.

So what is it that makes a thing good for someone? When taking medicine is good for a sick child, reading the classics is good for an uninterested teenager, and being honest is good for all humans, are we using “good” in the same sense? I think we are. I propose that when we identify something as good for someone, we associate that thing with the extension of their happiness. Happiness is a state that all our concepts of positive experience approximate to varying degrees. It is experienced as an acceptance of what is, in contrast to dissatisfaction, which is experienced as resistance to what is in favor of what is not—what was, or what could be. Dissatisfaction is approximated to varying degrees by all our concepts of negative experience.

The state of happiness extends within us unless it is interrupted, either by dissatisfaction or by inertia. This extension plays, for goodness, a role similar to the role that the extension of matter plays for largeness.

In other words, just as, in regard to the size of a physical object, we can refer to it as large when viewed in relation to a range, larger when compared to another physical object, or, more precisely, as having such and such extension of matter as per some unit of measurement, so too, in regard to the value of a thing, we can refer to it as good for a person when viewed in relation to a range, better when compared to another thing, or, more precisely, as being associated with such and such extension of that person’s happiness.

Moore is right that judgments of value can be objectively true or false. It is possible to be wrong about what things are good. Since the association between a thing and the extension of happiness is independent of them, the goodness of a thing is independent of our desires, preferences, and judgments. The sick child who rejects his medicine and desires sweet treats instead disagrees with his mother from a failure to understand the consequences of not taking the medicine. The teenager who doesn’t feel excited about reading the classics and prefers popular novels instead fails to appreciate the many complex, subtle, and unexpected ways in which engaging with the classics would affect him over a lifetime. The person who rejects honesty for some immediate gains fails to recognize that honesty would make him whole overall, both internally and materially, even if he never finds out how.

These failures of judgment demonstrate that identifying the goodness of different things is a painstaking process of discovering certain facts about the natural world. As with all such facts, to discover them we have to rely not only on our own intelligence, but also on what others tell us, how they actually behave, wisdom handed down to us in various forms since time immemorial, and, more generally, on methods that help us understand how the world works.

There are many objections that can be raised against my position. One criticism, found in the philosopher Thomas Reid, asks: if goodness depends on our constitution, then wouldn’t a change in that constitution make immoral things moral? Another serious concern is this: if a thing is always good or bad only for someone, does that mean the universe “doesn’t care” about an act like the Holocaust, or that there might be beings for whom it is not bad at all?

I think we should be prepared to accept these implications, difficult and unsettling as they may seem. We attach so much importance to pain in our ethics precisely because our constitution is the way it is. The universe is neither indifferent to nor interested in human affairs, just as our knees are neither blind nor sighted. Likewise, there are, for instance, animals with different constitutions who live in remote parts of the world and whose good, as a result, cannot always be the same as that of humans, despite the convergences.

It is the countless ways our lives are interconnected and interdependent even when we live oceans apart, and the different paths through which the enablers of our happiness and dissatisfaction converge even when we do not, or cannot, understand them, that make injustice anywhere a concern for everyone, and ground the happiness of one in the happiness of the other. Such convergences arise, even though they are not guaranteed, because of the way the world is—just as material prosperity arises in a free market, even though it is not guaranteed by the government, because of the way the world is. But even if the convergences did not exist, it wouldn’t change the nature of goodness as a property, and we would have to accept that there is no reason to think the way the world is will always align with what we believe must be the case. We have already experienced this before in the history of ideas, such as with the theory of evolution by natural selection and quantum physics. Why should ethics be any different?

There are major implications of this view of goodness for the possibility of an objective science of morality. There are many more objections than those I have mentioned. I will be exploring both of these in future essays. For now, however, I conclude that Moore’s charge of a “naturalistic fallacy” is misplaced. Goodness, like largeness, is a natural property, open to inquiry through the same methods that have helped us understand atoms and galaxies.

Continue Reading:
→ Bryan Caplan on Ethical Intuitionism (podcast)
→ The Inescapability of Altruism
→ Review of Tyler Cowen’s Stubborn Attachments

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

Vatsal: Welcome to Vatsal's Podcast, part of a philosophy newsletter where I share original essays and conversations with leading thinkers. You can subscribe for free at readvatsal.com to get new episodes and posts delivered straight to your inbox.

My guest today is Bryan Caplan, Professor of Economics at George Mason University and a New York Times bestselling author. His books include The Myth of the Rational Voter, Selfish Reasons to Have More Kids, and The Case Against Education.

We discussed ethical intuitionism, the view that some moral truths can be known intuitively, without relying on sense experience.

Here’s my conversation with Bryan Caplan.

Bryan Caplan, thank you so much for being here. It’s great to see you.

Bryan Caplan: Great to see you too, Vatsal.

Vatsal: I wanted to talk about meta-ethics and specifically ethical intuitionism. But let me begin with a question about common sense realism. What is the material difference between common sense realism and someone who is more skeptical like Hume if they make the same predictions about the world?

Bryan Caplan: I would start by saying that even to go and claim someone is making predictions, you really are talking about a prediction about what really happens. If you follow Hume strictly, then you’ve also got no knowledge that anyone else in the world even exists. So all you’re really doing is making predictions about your own experience, not what really happens.

And if you remember what Hume says about memory knowledge, you’re not even doing that because you don’t actually have reliable memory knowledge. You may say, oh, this is exactly what I predicted. Well, how do you know that what you predicted is, in fact, what you think that you predicted? It could just as well be that the universe just burst into existence this very second with false memories that you previously existed. So, I would say it is the difference between a normal view of the world and a totally bizarre one.

Vatsal: Would you say that your opposition to skepticism is consequentialist, like you are worried that it’ll lead to bad outcomes?

Bryan Caplan: No, I just think it’s false. I mean, if it’s true, if one were in fact actually sincerely skeptical, then it would definitely make sense to change your behavior in the same way it makes sense to act differently in a dream than in reality. If you knew for sure you were in a dream and you want to punch someone, why not punch someone in the dream? They’re not real. You’re not hurting anybody. On the other hand, if you know that it’s a real person, then it does change what would be proper behavior, of course.

And then if you’re in the position of, I don’t know, is it a dream or not. If it’s like a one in a million chance it’s real, then yeah, sure, why not punch the person? Who knows? They almost certainly are just a product of your imagination, anyway.

Vatsal: One objection that can be raised against intuitionism is that it doesn’t add anything. What problem does it solve, whether in epistemology or ethics, that couldn’t be solved without it?

Bryan Caplan: What I would say is that’s basically this fundamental anti-intellectualism of saying, I don’t care whether anything is right or wrong, so it doesn’t add anything. It’s like, well, the fact that you don’t care doesn’t matter. Is it true or not? And if it’s true, it adds something because it’s an additional fact. That’s the whole answer. If that’s not enough, say, like, was Hitler really evil? Or is it just society’s opinion that he was evil? What difference does it make? If a person is there, then it’s like, all right, look, why talk to a person like that? They’re not interested in the question.

Vatsal: No, I meant against naturalism. A naturalistic worldview has specific criteria like pain and pleasure. Intuitionism relies on something that is self-evident and that is in a sense vague. So the question is, does it actually add anything to rely on intuition rather than naturalistic criteria like pain, pleasure, or egoism (like Rand)?

Bryan Caplan: Well, right. As G.E. Moore pointed out, it is a meaningful question to ask, “Is pleasure good?” And that’s his refutation and the whole basis of the naturalistic fallacy. It is a meaningful question. If someone just says, “no, that is not true, pleasure equals good by definition”, it’s like, okay, you’re not speaking English. And again, what’s the point of continuing the conversation if you’re going to be like that.

Vatsal: And you wrote that Ayn Rand’s ethical egoism is anti-realist. Could you elaborate on that?

Bryan Caplan: I honestly don’t remember specifically saying that, but I think that she is presenting it as a moral realist theory, but what she winds up doing is, in terms of constructing her argument, it does come down to, okay, well, this is what living things do, and it is the nature of a living thing in order to go and pursue its nature and to seek happiness if it’s a living being. And again, there is the question, all right, fine, suppose that’s true. Does that show it’s good? Does it show it’s right? And once again, it is a meaningful question.

She might actually bite the bullet and just say, in the same aspie voice I was just doing, “no, by definition, whatever maximizes an entity’s self-regarding happiness is its good”. And it’s like, okay, if you’re going to be that way, I don’t know what the point is of talking. You’re not speaking English and you’re not going and addressing the issues that seem perfectly normal to other people. But that’s where I would go with that.

Vatsal: Your blog post, it’s called “From Intuition to Contrarianism: A Case Study”. In that blog post, you discussed how employees can discriminate against employers, and you asked whether employers can also discriminate against employees the same way. Tyler Cowen disagreed with you on this and he’s also an intuitionist. So suppose a disagreement like this occurs between two intuitionists. How do you resolve that?

Bryan Caplan: First of all, I would say Tyler isn’t really intuitionist at all. I don’t really understand what he is, but he has not endorsed intuitionism as far as I know, except maybe to say, yes, well, of course, but it doesn’t matter. And all right, that’s a strange kind of intuitionism. In terms of this meta question, if there’s two people who share the same methodology, but they disagree, what can they do? All right. You might say, well, aren’t you just in an impasse then? No. So what is it that two people who are both empiricists can do if they disagree? They may say, let’s go, let’s review the empirical evidence more carefully. Let’s go and see if there’s some other evidence that I haven’t heard of. Similarly, if you have two intuitionists who disagree, one step is let’s go through the alleged intuitions more carefully and make sure that we even actually disagree. Second of all, so do you have any other intuitions that I might share? I think you’ve been talking a lot about intuition, but unless there’s some world community of intuitionists that I’m unaware of, I think that most people listening will have very little idea what we’re talking about. What I would say is my view of intuition just comes down to, you’ve got to start somewhere, with something that’s obvious. And if someone disagrees with something you consider obvious, what can you do? You’ve got to find something that’s even more obvious to them. And what if you can’t? They reject the thing that you find obvious, you can’t come up with anything that they find more obvious. Then you really are at an impasse. But there’s nothing special about intuitionism for that. If you’re an empiricist and you say, look, I say that the Rock of Gibraltar exists, and you take them there and you see there it is, and the other person says, I’m an empiricist, but I don’t see a rock. It’s like, hmm. You are sure you don’t see a rock? I see a rock. And it’s the same kind of thing, right? Again, you might say, well, empiricists don’t disagree with each other in that way. And like, well, I know a lot of empirical researchers who do have very deep, deep disagreements with each other. So, I don’t think so.

Vatsal: What is your main objection against naturalism, moral naturalism?

Bryan Caplan: Right. So G.E. Moore said it very well, and I’m happy to repeat what he said. For any natural property, which allegedly is the same thing as a moral property, you can always say, “yes, but is X good?” Someone says, you know, good means pleasure. Good means happiness. Good means longevity. Good means the survival of the community. You can say, yeah, well, but is the survival of the community good? And if you just say, “that is a meaningless question. By definition, the answer is of course, yes”. Then the conversation’s over, the person doesn’t really understand English. Or, if the person says, “hmm, interesting question”. Well, the point of this question wasn’t really even to answer that question. It was just to show that you’re committing the naturalistic fallacy and it’s always a meaningful question to say is a given descriptive situation, one that possesses a certain moral property.

Vatsal: How do you deal with the question of motivation? Suppose someone agrees that some moral facts — like “murder is wrong” — are self-evident, but he lacks the motivation. In that case, what can be done?

Bryan Caplan: There’s a classic list of how to motivate other people. There’s giving them an inspiring speech. There’s, well, look at this poor little person. Wouldn’t it be wrong to do that? There’s telling stories. There is the Stoic one of: “Your motivations do not matter, Sophocles. All that matters is the right— do the right”. You can try all these things and then, in the end, if the person says, I’m still not motivated, then it’s like, all right, well. I guess you’re just an immoral person. Nothing more I can do about it. The last thing is “hopefully you will mature over time and see the error of your ways”. All right, that’s, you’re left with that. And it’s, that seems pretty lame. It’s like, that’s life. Life does not give anyone the superpower of motivating other people. If someone just says, yes, I don’t care whether anyone else lives or dies. And it’s like, well, really? If they’re, they’re firm on that, then it’s like, yeah, well I don’t see how I’m going to motivate you then.

Vatsal: Thank you so much for having this conversation with me. I really appreciate it.

Bryan Caplan: Totally happy to do it.

Continue Reading:
→ Overcoming the Naturalistic Fallacy
→ Review of Tyler Cowen’s Stubborn Attachments
→ Michael Shermer on Morality and Science (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

Vatsal: Hello everyone. Welcome to Vatsal’s Podcast. This is Vatsal. Today I’m speaking with Michael Shermer. Michael is a very well-known social scientist, founding publisher of Skeptic Magazine, host of The Michael Shermer Show, and author of many best-selling books, including my favorite, The Moral Arc. Michael, welcome.

Michael Shermer: How you doing? Nice to see you.

Vatsal: Your dissertation was on Alfred Russel Wallace. What did you learn about human nature and its complexities by studying Wallace’s life?

Michael Shermer: Well, that started off as just looking at some history of science issues about how discoveries are made, and this was in the case of the theory of evolution by natural selection by both Darwin and Wallace. So what did they have in common? Where did they differ? How were their life paths similar and different? Why did they both come up with the same idea pretty close to each other in time? To what extent are ideas like that already in the air culturally? And that somebody would have figured it out pretty soon within that, a decade or so. So I find all that interesting in terms of how lives turn out in general. But in terms of human nature, this is kind of a larger issue, but I think there’s interactive effects between genes and environment and luck or chance or contingency.

Vatsal: They had very different temperaments, right?

Michael Shermer: They certainly did, yes. Wallace was far more open-minded to new ideas, many of which were crazy ideas: phrenology and spiritualism and seances and the paranormal and the supernatural. Darwin was pretty tough-minded and skeptical of all those things. So the characteristics that made Wallace open-minded enough to discover the theory of natural selection also led him to believe a bunch of nonsense, whereas Darwin did not. So I explored those ideas in my biography of Wallace.

Vatsal: One of the criticisms of the Enlightenment is that when you question all dogmas and authorities, it can lead to subjectivism and moral relativism, where any belief can seem equally valid. You get this question all the time. What response do you find most resonates with skeptics?

Michael Shermer: Well, most, let’s say, religious people believe that there are objective moral truths and right and wrong as handed down by God. So this is divine command theory. Well, I’m an atheist and most skeptics are religious skeptics, and they don’t believe that. But that doesn’t necessarily mean the alternative is pure moral relativism, where anything goes and there is no right and wrong. Although that is what some people conclude, what I’m trying to argue in The Moral Arc and elsewhere, is that there are moral truths. We can discover them through science and reason and apply them. And I’m claiming that, in fact, we have been doing that for centuries. And this is one reason why life is so much better now than it used to be.

Vatsal: You often refer to Euthyphro’s dilemma, which raises the issue of whether morality is contingent on divine command. A similar critique can be made about grounding morality in biology. If human biology were different, would moral facts change? Would immoral acts become moral? I think Thomas Reid raised this question against Hume. What do you think about it?

Michael Shermer: Well, there you’re kind of touching on the idea of deriving an “ought” from an “is”, which Hume famously pointed out is a fallacy. You can’t do that. But I’m claiming that that fallacy is itself a fallacy. In fact, we do it all the time. I mean, the way things actually are in the world: People seek freedom. They prefer health over disease in the same way that they prefer to be free rather than enslaved. That would be an observation about what people actually want and then deriving from that what we ought to do. We ought to defend people’s rights and freedoms because that’s what they want. I spend chapters in The Moral Arc and in other books, making this case and articles too and, I think it’s a bit of a cop-out to just say well we can never derive an “ought” from an “is”. Well, that’s not true, and so I think Hume’s argument was “thin”, as philosophers call it. That is to say, it applies sometimes, but not always.

Vatsal: I was listening to your conversation with Tyler Cowen and he expressed concern about declining religiosity. If religiosity is strongly associated with positive outcomes like personal well-being and community cohesion, do you think there is a utilitarian case for encouraging belief even if one doesn’t accept religious claims as true?

Michael Shermer: I mean, in general, I think we should believe things because they’re true. They should be believed because they’re facts about the world. In other words, I don’t want to believe in things that have to be believed in to be true. Although sometimes that may not always be the case. Maybe for democracy to work, we all have to believe that democracy works and participate in the system. Maybe. But religious people make claims that they claim are empirical. They’re not politically true or convenient or culturally true or metaphysically true or mythically true. They’re actually true. So just take the resurrection of Jesus, which Christians claim they believe actually happened. Not metaphorically or mythically, but actually literally, empirically happened. A man named Jesus was crucified and dead for three days dead. Dead dead, not in a coma, but dead, and then was brought back to life. Okay, so, of course I have all kinds of problems with that empirically, but what they’re after there is saying that they want to play religious claims on the same playing field as scientific claims. I just think that’s a mistake.

Vatsal: And what do you think about utilitarianism more generally? Do you find the idea that everyone’s happiness counts equally convincing?

Michael Shermer: Well, I’m mostly a utilitarian or a consequentialist when it comes to morals, although not always. There’s basically three different moral systems: utilitarianism or consequentialism, and then deontology or rule-based or rights-based morals, and then Aristotle’s virtue ethics. So if we just take, say, the trolley problem, which everybody’s familiar with. The trolley is hurling down the tracks about to kill five workers. You’re standing at the switch. You could throw the switch and kill one worker down this other track, would you do it? Most people say they would. They make the utilitarian calculation to sacrifice the one to save the five. But, of course, there’s alternate versions of this. You’re a doctor working in a hospital and you have five dying patients and there’s one healthy person in the waiting room, would you sacrifice that person in the waiting room? Call him in and anesthetize him, cut out his organs, kill him, and then save the five patients. Well, no, you wouldn’t do that. Well, why not? It’s the same utilitarian calculation. And the answer is because people have rights. They have a right to their own bodily autonomy and choice and control over what they do and what’s done to their bodies. And so there’s an example of where utilitarianism doesn’t work. Although I wouldn’t make the case that it’s better for society in a utilitarian way if we don’t sacrifice individuals and have that kind of utilitarian calculation like in the trolley problem. That is, by granting people rights that no one can enslave them or kill them and so on, then society is better off. I mean, just in a utilitarian way. So I hesitate to make too big a distinction between those two systems using that as an example.

Vatsal: We were just talking about the fact-value distinction. Many philosophers and scientists who are otherwise committed to naturalism argue that there is an unbridgeable gap between values and facts and that we can’t derive values from facts. And you disagree, of course. What point do they raise in response to your arguments that you find most difficult to refute?

Michael Shermer: That a lot of this is very context and time dependent. So, for example, had I lived three centuries ago, or maybe even two centuries ago, I probably would have made the case that slavery was acceptable based on some rational, allegedly rational argument that black people are inferior intellectually and morally and all the justifications that people used, including, and especially Christian theologians that that’s perfectly okay, or the way women were treated or Jews were treated and so on. I can’t say for sure that the beliefs I hold now I would still hold if I lived two centuries ago. If you put me in a time machine and went back, how would I think? And so much of what I argue may be grounded in 21st century moral thinking, the way people today think and who knows maybe centuries from now we’ll talk about artificial intelligence morals and rights and values. So that would probably be the biggest criticism, and that’s true, I mean that’s true for all of us. Everybody has that same problem. We’re all embedded in a culture, in a time period in history, but what else are you going to do? There’s no time machines and I can’t live forever. And I can’t be, I’m not going to be cryonically frozen and brought back 500 years from now. I mean, it’s just not going to happen.

Vatsal: When there are disagreements in science, we often rely on demonstrations and predictions. If we want to ground morality on naturalistic foundation, do you think we should be able to come up with some kind of predictions?

Michael Shermer: Well, certainly, yeah, certainly, and I do. That is to say, if you do more of X, you’re going to get more of Y. So if you grant more freedoms and rights and voting rights, say, and spread democracy and free trade between countries, they’re going to have fewer conflicts. There’ll be fewer wars between nations that trade with one another, for example. And democracies are less likely to go to war with other democracies. Not perfectly, not 100%, there are exceptions, but on average, it’s better to be a democracy for a whole bunch of reasons than it is an autocracy or a theocracy. And there are reasons for that, but the larger point is that if that’s true, then we should do more of that. And my prediction is if we do that, on average, the per capita GDP of a country is going to go up if the people are allowed to have trade with one another in between states, between nations, if they are democracies, and if they have rights, if they have the rule of law, they have property rights, if they’re allowed to form businesses and make contracts, all the stuff that we take for granted. There are countries where they don’t do those things, and those countries tend to be fairly poor and dysfunctional, and crime rates are high, and poverty is high, and so on. And so I predict that if we did those things, the crime rates would decrease, prosperity would increase, and so on. Again, not perfectly every single time, because there’s so many variables at work, but on average, I would make that prediction.

Vatsal: So you’d say that revealed preferences serve as a demonstration of what is right and what is wrong.

Michael Shermer: Yes, yes, absolutely.

Vatsal: People voting with their feet when they migrate to certain countries.

Michael Shermer: They do. Where would you rather live, South Korea or North Korea? Quick. Yeah. And everybody knows the answer and they know why the answer is the way it is. People already know this. You don’t have to make a rational calculation and prediction about that. But if you want to, okay, then I predict, if Kim Jong-un lowered the wall and just said, okay, to his fellow North Korean citizens, you can do whatever you want. Go ahead and leave if you want. They would leave absolutely. There’d be nobody, almost nobody left, and so the reason dictators have to have no free press, they have to build walls, they have to build concentration camps, and so on is because they know that the people living in their country don’t want to be there, that they don’t like that system. They’re making a prediction: If I don’t build a wall, my people will leave. They know this. That’s a prediction. And they would. Look what happened after the Berlin Wall fell. People just left. Okay, goodbye and good luck. Until East Germany got its shit together and started becoming more of a Western democracy and industrialized nation. Anyway, that’s an example.

Vatsal: You write that combining self-interest and sociality with impartiality, we can derive a principle of interchangeable perspectives that is embodied in the golden rule. Are you sympathetic to the idea that we cannot help but increase our own good, even when we desire other people’s happiness? This view is associated with Thomas Hobbes.

Michael Shermer: Yes? Oh, was there a question there? Sorry.

Vatsal: So there is this conflict between self-interest and increasing the general happiness.

Michael Shermer: Well, of course, since by human nature, we have a selfish side and an altruistic, cooperative side, more competitive and cooperative. We’re selfish and we’re greedy, but we’re also altruistic and giving and caring. So the whole point of a civil society with a rule of laws and, protection of property - property rights - and criminal justice and all that is to attenuate the inner demons, the selfish side of human nature and accentuate the positive side, the better angels, bring out what’s best in us. How do we do that? You do that through choice architecture as it’s called. That is to say, you have policies that encourage people to be giving and carrying and so on. Now they could do this voluntarily but it doesn’t hurt - for example, in the United States, when people make a donation to the Skeptic Society, which they do, they can deduct it from their income tax. Whatever you make every year, let’s say you make $100,000 a year, and you’re taxed at 30%. If you give the Skeptic Society $10,000, now you’re only going to be taxed on $90,000 rather than $100,000. So you get to deduct the donation to a nonprofit. And this is true for churches and charities and so on. So that’s the government’s way of saying, we would like you to be more giving and cooperative and nice. And they do that, same thing with marriage. I’m married, I get a tax break. I have children, I get a tax break for having kids. I own a home, so I get to deduct the interest on the mortgage of my home from my income tax. So this is all the government’s way of saying, we would like you to be married, have kids, own a home, give to charity. And now I don’t have to do it, any of those things, I can opt out. But it’s a way of society saying, these are the things we think are good for our society, we want our citizens to do it. Now, the fact that we do it and get pleasure out of it also means that there’s something inside of us that is good and people willingly do those things.

Vatsal: David Hume famously said that reason is inert and in order to get us to move we need some kind of motivation from our desire or passion. Do you think that’s a limit to rationality?

Michael Shermer: Well, not really, because really what Hume is saying there is that we use reason to achieve our goals, but what should the goals be in the first place? That’s rather subjective and emotional. But even there, you can kind of lay out the landscape of different jobs, careers, marriage partners, business partners, whatever it is you want to do. And you can assess the risks and benefits of each choice and then make a rational choice. Now, ultimately it may come down to just the way you feel about whatever career you want to choose. What kind of makes you feel good and productive and like a whole person because you like doing that or you’re in love with somebody and maybe you can’t quantify exactly what you like about them, but it’s an emotive feeling. So that’s what Hume was really talking about there.

Vatsal: Within the sentimentalist tradition of Hume and others, there is this idea that there is some universality in what we desire and what makes us happy. So is that a good argument for universalism?

Michael Shermer: I think so, yes, because we have research now on what makes people happy. In the same way we have research on what makes people wealthy or productive or whatever. And in the case of the happiness business, maybe the right word is fulfilled or purposeful, something like that, a fulfilling, purposeful, happy life, something like that. We know what that involves. Meaningful, purposeful work. A relationship of some kind. Close relationship with someone you’re in love with. Having children, having family, having friends, having a social circle of people that you hang out with. Doing something productive every day. Moving your body, exercising. Eating a relatively healthy diet. These are all things that we know through longitudinal studies over decades and decades, even a century long in the case of the Harvard study. These are the things that people do and the outcome is measurable. They live longer, they’re healthier, they’re happier, self-report happiness, and all that. And you can see that across countries. There’s tons of data on this now. Countries that are higher in trust, they have a more trustworthy economy, stable economy, stable political system, stable banking, financial system, and so on. The people there, and they make more money, they’re happier. We know what leads to that, so we should do more of that.

Vatsal: Final question. What do you think about the relevance of free will to this topic? Sam Harris denies free will but he is a moral realist. On the other hand, Immanuel Kant thought free will was absolutely essential for the existence of morality and he believed that free will must be assumed for morality to make sense.

Michael Shermer: Of course. Of course, morality depends on somebody choosing to do good or do evil. And so we have to assume free will. Now, let me comment on determinists. They don’t actually believe it. I mean, they say they believe it. They have arguments. They write books about it, but they don’t act like determinism is true. They act like everybody else, like I’m making free choices and I’m going to take credit for the good choices, and people are going to blame me for the bad choices, I will be punished if I do bad things, I will be rewarded if I do good things, and even though they deny that any of that is actually happening, of course it’s happening. Just open your eyes, you can see how people behave. But in any case, that’s not my solid argument for compatibilism, that is, we live in a determined universe, but it’s not predetermined. That is to say, the future is unknown and you are part of it. You are part of the causal net of the universe. The choices you make alter what’s going to happen next. And therefore what you do really does matter in your life, other people’s lives, in your circle, to the world at large, no matter how small your influence might be. You just don’t know. And so why not act as if you’re making free choices? Because you are. That is to say the future is not predetermined. You are part of the causal net. Your choices alter the future paths. Think of the metaphor of, I call this Heraclitus’ principle, you can’t step into the same river twice because the river is not the same and you’re not the same. Or just think of it as “the garden of forking paths”, as the poet and novelist Jorge Borges wrote, The Garden of Forking Paths. With each choice, each bifurcation point along the path that you go, it’s going to have a different outcome. And at each of those bifurcation points, you go left instead of right. You took this class instead of that class. You married this person instead of that person. You went to this party instead of some other party or whatever it is. There are millions of these little bifurcation point, choices you make in your life, determines the outcome. And you’re doing that. This isn’t happening to you like you’re a pinball in a machine just randomly bouncing around and who knows how it’s going to turn out. So that’s my argument for volition. It’s a compatibilist argument.

Vatsal: Thank you so much for answering my questions.

Michael Shermer: You’re welcome.

Continue Reading:
→ Bryan Caplan on Ethical Intuitionism (podcast)
→ Solving the Trolley Problem and Other Moral Dilemmas
→ Overcoming the Naturalistic Fallacy

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

Stubborn Attachments: A Vision for a Society of Free, Prosperous, and Responsible Individuals By Tyler Cowen • Stripe Press • 2018 • 160 pages • Buy

A major influence on Jeremy Bentham was Thomas Hobbes, who viewed human motivation in terms of the pursuit of one’s own good. A biographer of Hobbes reported an occasion when he was seen giving alms to a beggar. When a clergyman asked if he would have given the money were it not for Christ’s command, Hobbes replied that it pained him to see the condition of the poor and infirm old man. Giving the beggar some relief gave him relief as well. Bentham similarly viewed pleasure and pain as humankind’s “two sovereign masters” that “govern us in all we do, in all we say, in all we think”. But this conflicted with his utilitarian principle that the right action should promote overall happiness, with each person’s happiness counting equally. It was Henry Sidgwick who recognized the full significance of this conflict. Sidgwick’s attempt at a reconciliation failed, and he concluded that “if we gave up hope of attaining a practical solution of this fundamental contradiction ... it would seem necessary to abandon the idea of rationalizing [morality] completely”.

In Stubborn Attachments: A Vision for a Society of Free, Prosperous, and Responsible Individuals, Tyler Cowen addresses this and other moral dilemmas through two principles for practical reasoning: first, we should sustainably increase prosperity, broadly understood; second, inviolable human rights, where applicable, should constrain this pursuit. He cites the East Asian economic miracles as the only episode in history where his first principle was clearly and unambiguously applied. They are to him, says Cowen, what the nineteenth-century Prussian state was to Hegel. He is not a simple utilitarian, as he does not view human well-being as always an absolute priority. Sometimes other values, such as justice, take precedence. His pluralism acknowledges that no single value can overwhelm all others in every instance, which may seem inconsistent with an attachment to rigid ideals of prosperity and liberty. The reconciliation lies, he argues, in the recognition that whatever it is that we value—human well-being, justice, truth, beauty—we are better able to achieve it with higher economic growth. If the pursuit of beauty produces some opportunities for beauty, but the pursuit of prosperity creates more opportunities for beauty, in addition to other values, then even those who prioritize beauty have reason to pursue prosperity. He also believes that time may be a moral illusion: when viewed sub specie aeternitatis, the happiness and suffering of future humans are as real as our own. He accordingly advocates for a deep concern for the distant future, and shows how his two principles present a path for resolving issues of clashing preferences through the creation of more opportunities in the long term.

In response to certain problems that seem insurmountable in the present—such as how to resolve disagreements and determine whose wishes should take precedence—Cowen invites us to focus on building a future, through all that is implied in prioritizing sustainable economic growth alongside inviolable human rights, where such conflicts would arise less frequently. I found this to be the most compelling aspect of the book, as the general approach has relevance for a wide range of ethical problems, both real and hypothetical. Take, for example, a moral dilemma in which a mother must choose which one of her two children will be killed. If we realize that once such a dilemma has materialized, there is no satisfactory choice available, and turn the attention of our ethical thinking towards the long-term future instead, we find it can only be resolved by dissolution: by working towards a world where such dilemmas do not materialize in the first place—just as this particular dilemma would not have materialized if we had taken certain actions in the past. The task, then, is to find the direction for our efforts towards building a future where we have dissolved the many dilemmas we face, and here Cowen’s two principles have great merit.

A fundamental contradiction in the book arises from the fallibility of knowledge in relation to the concreteness of the actions taken. Even if we share Cowen’s views on what is right, can we be sure that the prescription he offers will help us achieve it? While dealing with the question of uncertainty about the consequences of our actions, Cowen points to the inescapability of choice and emphasizes the importance of major consequences over minor consequences. Because of what consistent economic growth enables us to do, this brings him once again towards the central thesis of the book. But consider one way in which this problem challenges that thesis. What if it is the case that certain actions we take to generate prosperity, and collective prosperity itself, create an environment where people eventually lose the kind of motivation that leads to procreation? In that case, by increasing prosperity beyond a certain point, we may be actively working towards reducing the overall good that could materialize. Worse still, in many countries, we may already be past that point, as several thinkers have observed in recent years. What we are dealing with, says Michel Houellebecq, is “a Western suicide or rather a suicide of modernity, since Asian countries are not spared ... the inevitable consequence of what we call progress (at all levels, economic, political, scientific, technological) is self-destruction”. Houellebecq cites fertility rate figures from advanced economies such as South Korea, Italy, and Singapore. Cowen himself has also noted this. “Given low fertility rates in virtually every wealthy nation,” he asked Peter Singer in a conversation, “is there something self-defeating about secularism as a philosophy ... secularism plus birth control, plus a number of other features of modernity. Does that mean this whole enterprise is just self-defeating?” Religiosity is often mentioned in this conversation, but it seems to me that what is really significant is particular conditions—such as desperation and sentimentality, instability and ignorance, unquestioning adherence to duties and authorities, perhaps even a certain comfort with things elemental and rough—which are frequently associated with organic, authentic religiosity, and which have gradually become less common.

Of course, all this poses a problem only if we assume it is objectively true that the good of all individuals is of equal importance and that we ought to increase the general good. But from what basis does this follow? From the premise that individuals seek to increase their good, what follows is only that they will seek to increase their own good. Even when we want other people’s happiness, even when the impulse arises out of conscience or rational deliberation, it is our desire that finally moves us to action. And the fulfillment that we ultimately cannot help but seek to attain by responding is our fulfillment. Like Sidgwick, Cowen reconciles intuitionism with utilitarianism, but still the problem of egoism of some form remains. If people’s unwillingness to care for the distant future is viewed merely as a limitation that prevents them from doing what is right, then the only question is finding ways to motivate them. But how does one demonstrate that this is what is right? Addressing this philosophical objection, it seems to me, would require deviating even further from utilitarianism—at least with regard to its truth rather than its usefulness.

It is doubtless better to be benevolent, and there are indeed objective rights and wrongs as Cowen says, but not because of anything self-evident or irreducibly normative. Anthony Ashley Cooper, the third Earl of Shaftesbury, observed that it is “according to the private interest and good of everyone, to work towards the general good; which if a creature ceases to promote, he is actually so far wanting to himself, and ceases to promote his own happiness and welfare”. A different but related idea can be found in Bernard Mandeville’s fable of the bees and Adam Smith’s “invisible hand” metaphor. A synthesis of the two ideas can be stated in the following way: In a collection of one hundred individuals, who may or may not agree in such things as their desires or preferences, cognitive capacities, and degree or direction of empathy, if each member acted, intentionally or unintentionally, in such a way as would increase the good of all the other members, then each member in that collection would have ninety-nine others acting in such a way as would increase his good.

If the circle of the collection expands, the number of acts—both active and passive—carried out for the good of any single individual in that collection would increase, since there would be more members available to act in a way that enhances his good. The question then becomes how to reach a point where the number of benevolent members is maximized. We formulate the relevant courses of action in terms of obligations and rights to ensure they are adopted more reliably. We use external sanctions to prevent deviation from them. We find their practice becomes more sustainable with material incentives, such as in an economic system where one’s prosperity depends on providing goods and services that bring value to others, or in a theological system where the selflessness of one’s actions determines one’s status when reincarnated or one’s place in the afterlife. But in all instances, whenever any prescription is followed, it is always through an individual’s judgment that it would lead to a greater extension of his good. It is the possibility of being correct or mistaken in that judgment, independent of how he feels or what he desires or prefers, that allows us to recognize the objectivity of moral truths.

To the extent that we care, and can be made to care, for those living on the other side of the earth, it is because we have a disposition that responds with disturbance when we reflect upon, or are made to feel guilty about, the real or hypothetical suffering of others. Or because we think performing altruistic actions for them will strengthen our status and legacy, or give our life the deeper, expansive fulfillment that is only obtained when one is not preoccupied with one’s own interests, as Joseph Butler pointed out. Or because we recognize that even if a few of them attain excellence, their achievements would enrich us. Or because we realize their prosperity will create opportunities for us or will at least prevent them from having to seek refuge among us, and so forth. Likewise, we only care, and can be made to care, for future generations if this is tied to us in some way, such as our desire to continue our lineage, which makes us want a prosperous and sustainable future, or a sentimental attachment to humans, civilization, or nature, or an instinct for reciprocation, or the fact that caring for the distant future cultivates certain habits and desires that benefit us in the present.

It may turn out that forgoing some basic assumptions of utilitarianism is what brings us closer to the science that David Hume hoped to derive when he attempted to introduce “the experimental method of reasoning into moral subjects”. In such a framework, we might even find that certain problems associated with ethical theories like utilitarianism—such as Derek Parfit’s repugnant conclusion, which Cowen discusses in an appendix—come to be viewed in the same way that the problems associated with Aristotelian substantial forms were viewed under the mechanical philosophy of the Scientific Revolution.

Keep Reading:
→ The Inescapability of Altruism
→ Solving the Trolley Problem and Other Moral Dilemmas
→ Bryan Caplan on Ethical Intuitionism (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (Curate your culture — iOS app)

Garett Jones is an economist and author who is Professor of Economics at George Mason University. His research interests include macroeconomics, monetary economics, and the microfoundations of economic growth. In addition to his academic work, Garett has served as Economic Policy Adviser to Senator Orrin Hatch and as a staff economist for the Joint Economic Committee of the U.S. Congress. His books include Hive Mind, 10% Less Democracy, and The Culture Transplant. You can learn more about him on his website and follow him on Twitter.

Vatsal: You observe that there is often vagueness, even contradiction, in how people use the term “democracy”. For example, they may describe the independence of the judiciary or the central bank as democratic, not realizing that this actually makes them undemocratic. There is a tendency to describe as “democratic” what is really good governance. Do you find a similar conflation between the descriptive and the normative in how people use the term “equality”? Felix Oppenheim pointed to the tendency to apply the term to “those, and only those, institutions or policies which one wishes to commend, and to qualify as inegalitarian those of which one disapproves”.

Garett Jones: I’d say it’s worse with democracy. With equality, at least in U.S. discussions over the last three or so decades, there’s been a clear distinction between “equality of opportunity” versus “equality of result”. Further, the word “equality” seems to be used by presumption to refer to equality of opportunity, which led progressive leftists to turn to the neglected term equity, which now means something a lot like equality of result. I’d be happy if popular discussions of the nature of democracy were as nuanced as popular discussions of equality!

Vatsal: You suggest that Aristotle would describe modern democracies as a polity—a mix of democracy and oligarchy. In Politics, Aristotle says that the many, though ordinary as individuals, can collectively be better than the best few, “just as a feast to which many contribute is better than a dinner provided out of a single purse”. Some have seen in this an instrumentalist justification for democracy and made the case that the cognitive diversity for solving collective problems that democracy enables produces better decisions than rule by experts. How do you think Aristotle would evaluate his observation against modern evidence?

Garett Jones: Aristotle was also deeply worried that democracy gave too much political power to the poor, who would vote for endless benefits for themselves. So while Aristotle was open to getting a lot of participation from the masses, he also emphasized that the middle classes—people probably best interpreted as owners of small- to medium-sized businesses—were an extremely important class in creating good governance. In Book 4 of the Politics, section 1295, he says:

[S]urely the ideal of the state is to consist as much as possible of persons that are equal and alike, and this similarity is most found in the middle classes; therefore the middle-class state will necessarily be best constituted in respect of those elements of which we say that the state is by nature composed.

That’s something he had in common with Friedrich Hayek, by the way, though apparently for different reasons—Hayek was concerned that the rise of office workers, the proletarianization of the bourgeoisie, would lead to a class of affluent folks who didn’t have the lived experience of meeting a payroll. That led Hayek to worry that even fairly affluent citizens might vote to take a nation down the road to serfdom.

So would Aristotle mostly agree with what I said in 10%? I think so—since I’m arguing that he’s right! I point out that successful governments get the right balance of democracy with oligarchy, of mass and elite influence, and I argue that there’s a Goldilocks point, a happy medium. I call that tradeoff the Democracy Laffer Curve.

And if there’s one point Aristotle kept making throughout his writings, it’s that best safety lies in moderation. He called that optimal balance between democracy and oligarchy polity, and I think that while we’d surely disagree on precise policy conclusions, he’d be fascinated by the last century of serious data-driven research that’s looked into where the optimal balance between democracy and oligarchy might lie.

Vatsal: Speaking of Aristotle, you have written about how intelligence predicts most of the virtues that Deirdre McCloskey argues were crucial in the rise of the modern commercial world, with the exception of fortitude. Could qualities like courage, which might not correlate with traditional measures of competence, be more consequential in shaping institutional effectiveness than we currently recognize? I’m particularly thinking about your response to the question your colleague Tyler Cowen asked regarding the cases where voters’ judgments are more likely to be better than those of experts, where you referenced academic indoctrination. I’m also thinking of the incident you mention in the introduction of your book.

Garett Jones: Bravery can help and it can hurt—and the old saying goes, “He who fights and runs away, lives to fight another day”. So too much bravery is a mistake, and getting the level of bravery right depends on the situation. Many an angry young man has decided to be too courageous, too brave, and quickly wound up in prison or dead. There’s a Goldilocks level of courage.

But to compare the value of courage versus intelligence: there’s no Goldilocks level of intelligence—the more the better, as far as we can tell! Smarter people make more money, produce more inventions, save more, and so on. They are more likely to be nearsighted (raises hand), so that counts as a negative, and I’m open to other idiosyncratic downsides of being smarter, but on net it’s something one wants more of.

On the incident: you’re referring to the time I got a call from George Mason University’s Campus Police letting me know that someone had left a very hostile (but not physically threatening) voicemail message on a campus number. Fortunately, nothing came of that. I got that call a day or two after I’d given a campus talk on the ideas underlying 10% Less Democracy, and the talk attracted a lot of negative attention from a couple of progressive media outlets. The most influential one was written by Zaid Jilani, who I suspect has moderated his views quite a bit since then.

Thinking of my own situation as an academic who has taken on some risky research topics, I can say that through the great luck of being in two supportive economics departments (first at Southern Illinois University Edwardsville and then at GMU), and perhaps through exercising some good judgment from time to time, I’ve been able to accomplish quite a lot, and to reach quite a few people. But I do think that maximal bravery would have been a mistake. I’ve known quite a few well-educated folks who have worked in areas similar to my own and have real career difficulties. It might be honorable to be a courageous scholar, but Falstaff was right:

What is honor? A word ... Who hath it? He that died o’ Wednesday. Doth he feel it? No. Doth he hear it? No.

So getting one’s dose of courage right is probably as important as getting a nation’s dose of democracy right.

Vatsal: The reforms you propose for achieving a better balance between democracy and oligarchy are marginalist, and you are skeptical of radical proposals like open borders, at least in the most innovative nations. I wanted to ask a question related to the tension between, on the one hand, the tendency to open the floodgates, as it were, in various social spheres—markets, speech, political power—and, on the other, the preference for a more incremental approach. As an example, we can consider John Stuart Mill’s demand for women’s suffrage alongside his advocacy for the more competent to have more votes. Perhaps the former would have sounded as sweeping as the proposal for open borders does today, while the latter would have seemed more reasonable, especially given Britain’s disproportionate role in innovation during that period. What kind of analysis could have helped someone foresee which of these proposals would succeed and which wouldn’t, based on the kind of evidence available at the time?

Garett Jones: If one is trying to guess the policy effects of women’s suffrage, common sense with the evidence of the day would have given you pretty much the same answer as much later statistical research: women care a lot about children’s health and early childhood education, that wouldn’t have been a controversial claim at the time. And in fact, it appears that women’s suffrage caused big increases in education and health spending. And while women had less formal education than men, it probably wouldn’t have been too controversial to claim that women, on average, were about as smart as men in some general sense (a fact confirmed by IQ tests developed later). So women had differences in policy preferences that mattered—and in my view, women’s policy preferences for more health and education spending were an improvement for the era.

But simultaneously, women had negligible differences with men on our best single measure of voter skill and voter foresight—what we now call measured intelligence. That means there wasn’t good reason to think that women’s suffrage would create a notable risk to voter quality. I think all that could have been foreseen to a respectable degree.

I think that same analysis could be applied to a hypothetical open borders scenario with the realistic assumption of voting rights within a decade or so. Add up the best-guess effects of changes in voter preferences and changes in average voter skills, and you’d have a good starting point for thinking about changes in government policy caused by open borders.

Vatsal: The Economist, in its very positive review of your book, also highlighted the moral component of democracy, stating, “Less democracy may mean more sensible outcomes, but it also means less legitimacy”. Would your response be similar to that of David Hume, who stressed the unfeasibility of Lockean consent and focused instead on good outcomes, and Jeremy Bentham, who also argued that legitimacy depends on outcomes?

Garett Jones: Yes, good outcomes are their own source of legitimacy. Ray Fair’s model of presidential elections puts a lot of weight on election year real income growth and all four years of the inflation rate—and voters reward politicians who rule during periods of low inflation and high real income growth. No surprise there!

But on top of that, people love a ruler, they love a de facto monarch, a head of state who can serve as a focal point. And independent, undemocratic judges are central to The Economist’s vision of legitimate government, as they are to my own! So The Economist is playing it a bit cute, arguing that their version of Legitimate Democracy (which includes undemocratic judges and central bankers who can’t be recalled let alone chosen by the masses) is real democracy while any substantial tweaks (even those which retain frequent elections!) are going too far.

In reality, we’ve inherited this weird blend of elite and mass control of our governments and we’ve just decided to call that democracy. We could tinker with it quite a bit—for example, by reinventing these upper houses of the legislature and turning them into an epistocratic check on government—and it would still feel like a democracy.

Vatsal: Since you take an instrumentalist view towards governance, I’m curious what ethical assumptions guide your support or opposition to policies on a more fundamental level. Are you an intuitionist, like Cowen and Bryan Caplan? Caplan describes you as a meta-ethical moral relativist—would you agree with that? What is your view on utilitarianism, especially its instrumentalist approach to truth? Do you think social contract theories, like contractualism, offer a better alternative for overcoming the limitations of utilitarianism?

Garett Jones: I didn’t know Caplan had described me publicly that way. I’m not a moral relativist in the same sense that I’m not a Santa Claus relativist. Santa’s not there for those who believe in him, and the same is true with moral facts. So actually, I’m a moral nihilist: I think talking about what true morality is like is akin to talking about what real Hamlet is like. The beginning of any serious conversation about Hamlet is realizing he’s just a fictional character.

Another way to put it is that I believe that moral preferences exist, but I believe that moral facts do not exist. I have my own moral preferences, ways I’d like to see the world turn out, and utilitarianism is part of that, since I care about the well-being of a lot of people along with the well-being of a lot of non-human animals. But excellence, greatness, human achievement—those are all central to my view of why I think humanity deserves at least some greater weight than that of non-human animals. And human excellence is, by and large, a team effort—it’s something that requires successful collaboration and cooperation. So I hope that my species continues to find ways to work with each other to create excellence.

Continue Reading:
→ Bryan Caplan on Ethical Intuitionism (podcast)
→ The Inescapability of Altruism
→ Michael Shermer on Morality and Science (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

1. Introduction

In any exploration of a widely contested concept, it is useful to begin by asking what is affirmed about the world when that concept is said to exist and what is denied when it is said not to exist. These questions yield a set of claims through the assessment of which progress can be made. From the long history of discussion of free will, we can distill the affirmations and denials about this concept into two kinds of claims: those that are true by definition and those that are empirical. We will examine each in turn.

2. Definitional Claims

A principal claim can be formulated as follows: Free will cannot exist if the choices we make are inevitable. It is the case that all events, including the choices we make, are inevitable. Therefore, we do not have free will.

Let us assess this claim first.

One way to categorize propositions is on the basis of whether they are true by definition. Claims that are true by definition simply affirm the law of contradiction through some concept. Take, for example, the proposition “all bachelors are men”. Since it would lead to a contradiction, based on our concept of a bachelor, to suggest that not all bachelors are men, we simply affirm the law of contradiction through this concept that we have formed when we assert that all bachelors are men.

To the extent that we cannot imagine the law of contradiction being violated, we cannot conceive of propositions that merely affirm this law through a concept as false. This gives such propositions a sense of certainty and timelessness. We cannot imagine any new experience that would make a bachelor cease to be a man.

In contrast, consider the proposition “bachelors exist”. Since no contradiction arises, given our concept of a bachelor, if bachelors do not exist, we do not merely affirm the law of contradiction through this concept when we assert that bachelors exist. We can imagine how this proposition could be false. For this reason, it lacks the sense of certainty and timelessness that characterizes propositions like “all bachelors are men”.

When we consider the nature of the claim “all events are inevitable” and ask whether it is more like “all bachelors are men” or “bachelors exist”, we find that it resembles the former. This is evident from the sense of certainty and timelessness it conveys: we cannot conceive of how it could be false. Let us illustrate this with an example.

Suppose you hear that a friend failed to survive an arduous journey, and you claim that this outcome was inevitable—just as all events are. Later, it is revealed to you that your friend is, in fact, alive. Would this disprove the claim that all events are inevitable? After all, you might say, “I was merely mistaken about the details, which has more to do with my knowledge of the world than with how the world actually is. My ignorance of events does not alter their predetermined nature. It was his survival that was inevitable. He could not have done otherwise in any case”. We cannot imagine any event that could falsify the claim that all events are inevitable.

At this point, it is worth considering an interpretation of Gottfried Leibniz on actuality, necessity, and possibility, which leads to the conclusion that all true propositions are of the same nature as “all bachelors are men”. That is, all true propositions are necessarily true.

We often find ourselves saying, of some event that did not occur, that it was possible. In the actual world, I am sitting right now, and the proposition “I am sitting right now” is true. But we can imagine a world in which I did not take a seat and am standing instead. In that world, the proposition “I am standing right now” would be true. These different scenarios represent possible worlds. A world is possible if it does not lead to a violation of the law of contradiction. In a world where I was both sitting and standing at the same time, there would be a violation of this law. Such a world, therefore, would be impossible.

The proposition “all bachelors are men” is true in all possible worlds, as its contrary would imply a contradiction. Such a truth is necessary. In contrast, the proposition “bachelors exist”, while true in the actual world, can be false without contradiction. It is, therefore, contingent.

According to Leibniz, if we possessed sufficient knowledge to have a complete concept of a subject, then every true proposition about that subject, deducible from that knowledge, would be such that its contrary implied a contradiction. In the case of your friend, if you had sufficient knowledge to have a complete concept of him, which would include all the events in his life, you would have been able to deduce from that concept alone that he survived his journey. You would also have known that the contrary proposition—that he did not survive his journey—would imply a contradiction, much like the contrary of the proposition “all bachelors are men” implies a contradiction given the concept of a bachelor. It was merely your ignorance that led you to initially believe that he did not survive.

If not for our ignorance, we would see all true propositions as necessary, even those that, in the absence of complete knowledge, appear contingent. If all truths are necessary, it follows that nothing in the actual world could be otherwise, and indeed, as Baruch Spinoza recognized, that the actual world is the only possible world.

Let us take a step back and ask how any concept is formed in the first place. More precisely, defining a concept as a set of related propositions, how is any proposition formed? By “proposition”, I mean that which imparts meaning to expressions. Whether regarding what is actual, what is necessary, or what is possible, so far as we are concerned with truth and falsehood, it is propositions that we are concerned with. What are the ingredients that a proposition is made of?

“All our thoughts and concepts,” Albert Einstein observed, “are called up by sense-experiences and have a meaning only in reference to these sense-experiences.” When we examine any proposition, we find that we were only able to form it after becoming equipped through some experience with its constituents. This holds true even for propositions about things not found in nature and existing only in our imagination. For example, the proposition “unicorns have four legs” was only possible for me to form after I had become equipped, from encounters with horns and animals or exposure to artistic representations, with the constituents of this proposition. Without these constituents, my capacity to form knowledge would never yield any significance to allow an object to which truth values can be attributed, just as my culinary skills would never yield a dish if I had no ingredients.

While we can affirm the law of contradiction through a concept or compare one concept with another, in order to be able to do so the concept must be formed first. This formation never occurs in the absence of some material that is obtained through experience. Thus, it is the material obtained through experience, which we can call “data”, that constitutes the ingredients of all propositions. Since it is to propositions that the qualities of truth and falsehood are attributed, and since propositions are constituted by data, it follows that there is no notion of truth value independent of data, nor any knowledge of the world that exists independently of data.

Regarding the data that has yet to become available to us at any given point, all that we can say is reflected in our predictions. The data available to us that is synthesized to form a proposition, along with all the data that agree with the predictions that synthesis allows us to make, constitute all the data accounted for by that proposition.

A proposition can account for at least as much data as another, in which case they can be compared with respect to truth value. If one of them accounts for more data, then it can be described as “truer” than the other. Newton’s theory of gravitation, for example, accounts for a certain amount of data available to us and, through the successful predictions it allows us to make, a certain amount of data yet to become available. Einstein’s theory of general relativity accounts for all the data Newton’s theory does, as well as the data Newton’s theory fails to account for. It allows us to make more predictions that are successful. It is, therefore, truer.

With these preliminaries in place, we can now consider the three types of propositions relevant to our current discussion, those concerning actuality, necessity, and possibility.

The discovery of what is actual consists in the identification of propositions about what occurs. What occurs refers to states of affairs that exist. It is these occurrences which constitute the world. This follows from modern physics, according to which, as Bertrand Russell put it, “events, not particles, must be the ‘stuff’ of physics ... ‘matter’ is not part of the ultimate material of the world, but merely a convenient way of collecting events into bundles”. Consequently, propositions that describe what occurs illuminate the world in proportion to their truth. Whatever tools we employ to discover such propositions hold only instrumental value. We may, for example, use the law of contradiction—itself identified from data—to eliminate certain events to arrive at propositions about the events that occur. It is this end attained through various tools that constitutes what is actual.

Suppose we have formed some propositions, based on a synthesis of some data, that allows us to state that such and such events have occurred in the past and such and such events will occur in the future, whether correctly or incorrectly. Now what more do we add, in terms of the data accounted for, when we say that some of the events are necessary—that they cannot be otherwise—and others contingent—that they can be otherwise? Once we have arrived, through whatever tools, at a set of occurrences, what additional predictions can we make about what occurs by designating some occurrences as necessary and others as contingent? In a word, none.

This superfluousness of the notions of necessity and contingency regarding occurrences can be better understood through other similar concepts from the history of ideas. Consider the notion of substance. “A substance,” to quote Russell again, this time from an assessment of Aristotle, “is supposed to be the subject of properties, and to be something distinct from all its properties. But when we take away the properties, and try to imagine the substance by itself, we find that there is nothing left.” If we take a certain number of propositions about properties—such as having a particular visual shape, being hard, and so forth—and group them together, we obtain the concept of an object: a table. When we establish its existence by specifying a location in spacetime, it becomes a part of the actual world—a table that exists. But what more do we add, in terms of the data accounted for, when we claim that the table is a substance? What predictions can we make about occurrences with this addition that we could not make before? None.

Or take the notion of causation. We can identify a proposition such as “whenever one ball strikes another, the second ball moves” and predict that when a particular ball strikes another, the second ball will move. To arrive at this proposition, we might have used various tools, such as certain rules of logic, considerations of counterfactual circumstances, a mechanical approach to inquiry rather than a teleological one, and a particular conception of objectivity over another. We can account for more data than what this proposition accounts for through a truer proposition, which would allow us to make more successful predictions. But what additional data do we account for when we add that the first ball’s hitting the second involved a causal power or force that brought about the second ball’s movement? What predictions are we able to make about occurrences with this addition that we could not make before? As David Hume recognized, none.

We thus see the superfluousness of the notions of necessity and contingency with regard to what occurs. But what about possibility? When I say “if I had not taken a seat, I would be standing right now”, in what sense can this proposition be true or false?

Because what occurs constitutes the world, among the things that are possible, only those that occur are part of what is real. What is possible but not actual refers to the occurrences in which what occurs involves what is possible but not actual.

One way in which what occurs may involve what is possible but not actual is when possibility serves as a tool of discovery. In such cases, what occurs—the process of forming knowledge in humans, or “thinking”—involves considerations of what is possible as a means of discovering what is actual. In this context, the notion of possibility is no different from the other tools previously mentioned.

Another way in which what occurs may involve what is possible but not actual is when possibility is used as a form of description that elaborates on what is actual. For example, we can refer to a particular occurrence, such as “I am sitting right now”, by describing it through a list of events that do not occur, such as “I am not standing right now” or “I am not walking right now”. The truth of these latter propositions depends on the fact that I am sitting right now, which is what they elaborate. In such a description, we can also employ general propositions, such as logical or physical laws, along with all the occurrences for which they hold true, as in “if I had not taken a seat, I would be standing right now” or “a thousand years ago, if someone had built a telescope and observed the heavens, he would have seen the moons of Jupiter”.

We may say that past events are close to alternative possibilities, while future events remain open, or we may suggest that both are equally close. In no case do our sentences represent any knowledge of the world beyond what they convey about what occurs, as derived from the data we have accounted for. This can be measured through the predictions we are able to make, as we have seen. Based on what is revealed from the predictions it allows us to make, the claim “all events are inevitable” is merely the law of contradiction affirmed through the concept of an event. It states that the events that occur, occur, and the events that do not occur, do not occur. There is nothing we can imagine that a new experience could bring that would make an event that occurs not be an event that occurs, or an event that does not occur not be an event that does not occur.

The method of relying on predictions to assess the presence of knowledge is valuable for any claim we encounter. If someone asserts that the sun will not rise tomorrow, he should be able to make predictions that differ from those of someone who believes it will rise. If he fails to provide any different prediction, then we must conclude that, to the extent that we can compare knowledge between two people, there is no difference in their knowledge of the world on this point; the apparent difference in their claims is merely verbal. The same applies to a person who says that free will does not exist and another who asserts that it does. Do they differ in what they know about the world, or are their differences merely verbal?

Any claim regarding free will that can be reduced to a version of the claim “all events are inevitable”—a mere affirmation of the law of contradiction through a concept—can be addressed with what has already been said. Any remaining claim would be like “bachelors exist”, where we can imagine how it could be false. We will now consider claims of this latter kind.

3. Empirical Claims

Claims about free will that are empirical can be expressed through various contrasts: fate versus chance, manipulation versus spontaneity, disability versus ability, lack of control versus control, lack of consciousness versus consciousness, and unfreedom versus freedom. A principal claim common across these contrasts is about responsibility in general, which can be stated as follows: Free will exists in us only to the extent that we are responsible for our actions. It is the case that no one is truly responsible for his actions. Therefore, we do not have free will. Furthermore, since we can only be justly rewarded or punished for an occurrence if we are truly responsible for it, no one can be justly rewarded or punished for his actions.

To evaluate this claim, we need to understand the nature of responsibility. The concept of responsibility has multiple meanings, some of which are not relevant to the claim at hand. For example, responsibility can sometimes refer to obligations, as when we say a mother is responsible for her child’s health. It can also denote a virtue, as in the case of a leader taking responsibility for the actions of his subordinates. To understand the notion of responsibility that we are concerned with, we must address three fundamental questions: To what kind of thing can responsibility be assigned? What is entailed when a thing is held responsible for an occurrence? And why do we assign responsibility?

To answer the first question, consider a person, me, pressing a button to turn a light on. To what can we assign responsibility for the occurrence of the light turning on? Is it the person, me, or is it my pressing the button? We previously considered the notion of substance and found it to be superfluous. By “me”, therefore, we mean a certain set of occurrences grouped together to form the concept of me. But this concept includes many propositions about occurrences not relevant to the event in question. I press the button, but I also go for walks, read, lift weights. I engage in other activities after pressing the button. Consequently, it is more accurate to hold the occurrence my pressing the button as responsible for the occurrence the light’s turning on. In other words, the thing to which responsibility can be assigned for an occurrence is another occurrence.

One might mention the power or lack of power to materialize a particular occurrence, and it can be observed that the relationship described above is what we would refer to as a causal relationship. We could say that my pressing the button was the cause of which the light’s turning on was the effect, and that I, as a whole, had the power to bring about this effect. However, as we found previously, the notion of causal power or force, like the notion of substance, is also superfluous. The proposition “I pressed the button and the light turned on” might or might not be true, or there may exist a truer proposition that we could identify, such as “the filament heated up and the light turned on”. But nothing is added when we claim that my pressing the button involved a causal power or force which brought about the light’s turning on as its effect. What is entailed when an occurrence is held responsible for another occurrence is a relationship of precedence and succession between them, as described in propositions with varying degrees of truth. If one were to say that a particular occurrence would not have occurred if another had not occurred, one would be invoking possibility, whose nature and relation to actuality we have already discussed.

It can also be seen that no distinction needs to be made between different kinds of occurrences when establishing responsibility. When correctly understood, the process of establishing responsibility for an event involving a person and his behavior reveals itself to be an instance of the same process of forming knowledge of the world as that for an event involving a celestial body and its trajectory. In both cases, responsibility is established by accounting for data. In both cases, we either possess knowledge or we do not. In both cases, what is responsible for an occurrence is another occurrence, and what is entailed in responsibility is a relationship of precedence and succession between them. Questions regarding the facts of responsibility must be separated from our moral judgments, just as questions about the trajectory of a celestial body must be divorced from human aesthetic preferences. When establishing responsibility, all occurrences are, in this sense, equal.

From all this, it follows that if there is a distinction between real and false responsibility, it can only be based on the truth value of propositions that establish responsibility. A truer proposition about responsibility, as a manifestation of greater knowledge, also allows us to make more successful predictions, and herein we find the key to the third question, regarding why we assign responsibility. Having established that the filament’s getting heated, rather than my pressing the button, is actually responsible for the light’s turning on, I make different predictions about how to produce light compared to before.

Since all occurrences are equal when establishing responsibility, this holds for events involving moral aspects as well. Consider how, having established that another person’s feeling fear, rather than assuming impunity, is actually responsible for that person’s attacking him, a person might determine that allaying that person’s fears is a better course of action than retaliating. To take another example, having established that fulfilling his duties, rather than practicing self-indulgence, is actually responsible for producing certain levels of happiness, a person might determine that fulfilling those duties is a better course of action than practicing self-indulgence. To take a third example, having established that a particular microorganism, rather than a malevolent spirit, is actually responsible for a body developing a specific disease, a person might determine that taking measures to eliminate the microorganism is a better course of action than attempting to rid the body of the spirit’s influence.

Let us now briefly consider how the notion of responsibility we have explored applies to the various contrasts mentioned at the beginning of this section. In all cases, we will illustrate the application using representative examples.

The notion of fate suggests that all the occurrences in the world are structured like the occurrences in a book. What we perceive as events unfolding is better understood as turning to a page and finding words that were already written. Insofar as the notion of responsibility is intelligible under such a scenario, responsibility relations would still hold, in the sense that one occurrence could still be held responsible for another occurrence in a relationship of precedence and succession, and real responsibility could still be distinguished from false responsibility based on the truth value of propositions. However, to the extent that this claim fails to lead to any prediction—direct or indirect—whose success or failure can be tested, it falls into a particular category of propositions. This category also includes its opposite: that the occurrences in the world are not structured like the occurrences in a book. For propositions in this category, which we can imagine being false, we can evaluate what data they account for and whether truer propositions could be discovered. We can wait until testing their predictions becomes feasible. We can also examine the kind of data whose synthesis may have led us to these ideas, much like the ancient philosophers who wondered why the gods possessed the same characteristics as the humans who worshiped them.

The idea of manipulation belongs to the same category as fate insofar as it implies an intelligence that intervenes in the world but remains undetectable. An example of this can be found in the deceiver imagined by René Descartes, who creates a simulation indistinguishable from the real world. However, there is another, more familiar sense of manipulation that can be experienced, which does not fall into this category. It is best illustrated by the example of lying. If someone lies to me to influence my selection of a particular course of action, it can be said that I was manipulated. His lying to me was responsible for my forming the judgment I did, which, in turn, was responsible for the choice I made. This could provide a true description of what occurred. Our emotional response to this fact, and what terms we use to describe it, would be independent of the responsibility relation itself, just as in any other case of responsibility.

It is worth noting that choices are made to attain certain ends, and there are instances where we can imagine a lie leading to better choices than the truth, such as when parents lie to their children to prevent harm. Similarly, because choices aim at achieving certain ends, the choice to adhere to particular legal characterizations regarding such notions as the individual, consent, and manipulation might serve a community better than alternative interpretations. Indeed, it is rules, of which laws are just one example, which represent knowledge about the effectiveness of choices derived from them, against which particular actions are judged as just or unjust.

A person can be influenced by disabilities in the choices he makes. This simply means that true responsibility relations for certain occurrences involving individuals with disabilities would differ compared to those without disabilities under otherwise similar conditions, as when a blind person collides with someone versus when a sighted person does.

By control, we mean the ability to regulate something either according to a teleological pattern or according to one’s conscious intention. In the first sense, we say, for example, that the sphincters in the bladder and urethra control the outflow of urine. When they malfunction, they fail to fulfill their purpose and thereby fail to exert control. In the second sense, we say, for example, that a person has control over his bladder because his conscious intentions align with his ability to regulate the release of urine. To the extent that he lacks awareness of his intentions or is unable to regulate the release of urine according to those intentions, he lacks control over his bladder.

In the world, we find that people have varying degrees of control over different things at different times. A person driving a vehicle may have control over its speed and direction at one moment but lose it later due to such reasons as a vehicle malfunction, a muscle spasm, falling asleep, or dissociation. After being introduced to a particular invention, a person’s control over his environment may increase, while the implementation of certain economic policies may reduce his control over his possessions.

Control in the second sense, because it depends on consciousness, comes into existence at a certain point—namely, after we become conscious. But what does it mean to be conscious? By “consciousness”, we do not mean merely the state of being active or experiencing emotions without knowledge. John Locke defined consciousness as “the perception of what passes in a man’s own mind”. To the extent that our senses are not involved in synthesizing data, they are pure manifestations of the body’s creative processes, like the pumping of the heart, and are not involved in forming any knowledge. Perception, therefore, implies knowledge, making consciousness a form of knowledge. If it is a form of knowledge, it is attained in the same way as all knowledge—by accounting for data—and is, in principle, accessible to any entity that can acquire knowledge, regardless of whether that entity is human or capable of experiencing emotions. While it may be that the data synthesized to develop the kind of knowledge implied in consciousness is only available to the entity that develops that knowledge, it still constitutes a form of knowledge. Consequently, to say that an entity is conscious is to say that it possesses certain knowledge that an entity lacking consciousness does not, with the implication that this additional knowledge may yield the advantages of successful predictions mentioned earlier.

Finally, if we conceive free will as the ability to act according to one’s desires with the aim of realizing a particular internal state, then this state can be interpreted as the end for which we adopt choices and against whose materialization choices are compared as better or worse. In this view, just as whatever tools we use to discover propositions that describe what occurs have only instrumental value, so too do the courses of action we adopt to achieve this end—which we can refer to as “freedom” or “happiness”—hold only instrumental value. A person in the state of freedom experiences unconditionality and acceptance, while a person who is not in this state, unless he is in the state of inertia or mindlessness, experiences its opposite: conditionality and resistance, a state we can call “unfreedom” or “dissatisfaction”.

The idea that all occurrences are inevitable, the discovery of being manipulated, and the constraints imposed by impairment often lead to unfreedom. While having control over something can produce freedom, in some instances freedom can also arise from the realization that we lack control—such as when a person in hopeless circumstances finds solace in surrendering to fate, or when someone in love experiences joy in the loss of inhibitions. Liberty can be understood as certain external conditions that facilitate the extension of an individual’s freedom. However, because liberty and freedom are not the same, the ability to act according to one’s desires does not always equate to freedom. An imprisoned person who cannot move as he wishes may still experience peace, while another, despite having all conceivable liberties, might feel discontent. We may say that discovering which occurrences are responsible for the extension of the state of freedom constitutes the ultimate aim of ethics.

4. Conclusion

When we claim that free will exists—or does not—we are making a number of different claims. Upon examining these, we find that some are true by definition, like “all bachelors are men”, while others are empirical, like “bachelors exist”. By evaluating claims of the latter kind, we can determine which are true and which are false as matters of fact about the world.

Continue Reading:
→ Can AI Have Free Will?
→ Michael Shermer on Morality and Science (podcast)
→ Bryan Caplan on Ethical Intuitionism (podcast)

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

Massimo Pigliucci is a philosopher and biologist who is Professor of Philosophy at the City College of New York. He has a doctorate in genetics from the University of Ferrara, a PhD in biology from the University of Connecticut, and a PhD in philosophy of science from the University of Tennessee. He publishes a regular column in Philosophy Now entitled The Art of Living. His books include How to Be a Stoic and Nonsense on Stilts. His forthcoming book is Beyond Stoicism. You can learn more about him on his website.

Vatsal: You have stated your preference for Plato’s definition of knowledge as justified true belief. However, you’ve also observed regarding justification that “if we enforced that clause in the Platonic definition of knowledge, it would turn out that we all know very few things comparatively speaking”. Additionally, you’ve argued that the intuition that “true and justified beliefs don’t require a causal explanation”, while handy “as a heuristic and first approximation”, fails to “survive careful scrutiny, and needs to be abandoned”. This brings me to a question about current artificial intelligence (AI) systems. A few years ago, you commented on something related to this and concluded, “science advances only if it can provide explanations, failing which, it becomes an activity more akin to stamp collecting”. When you encounter AI systems that can process vast amounts of data, identify regularities, and make increasingly successful predictions, do you, as a scientist and philosopher, feel that there is still something more—like justification or causal explanation—that must be demonstrated before you can say these systems possess knowledge in the same way humans do?

Massimo Pigliucci: That’s right. Current Large Language Models do not possess knowledge, they produce output on the basis of correlation analyses of large amounts of text and symbols. This is, of course, impressive, and it may be useful under certain conditions, but it’s not knowledge, and certainly it’s not understanding.

I’m not suggesting that AI will never be capable of knowledge and understanding. I don’t know, and I doubt anyone else does. But current predictions of AI “surpassing” humans in those realms are baloney. The software architecture of LLMs simply does not allow that, though it does allow these programs to “learn” and to produce impressive, if often limited or even faulty, outputs.

Going back to Plato: I think it would be a good thing if we embraced his definition of knowledge as justified true belief precisely because we would finally realize that we know far less than we think we do. That may make us a little more cautious and humble, which is something that doesn’t come natural to the human race.

As for causal explanations, I meant that scientific knowledge cannot do without causal explanations, but everyday common knowledge may. For instance, I “know” how to drive a car, even though I don’t necessarily understand all the causal connections that make possible for a car to be driven. My lack of knowledge of such causal connections is the difference between me as a regular user of the technology and an engineer who actually knows how cars work and could build, not just drive, one.

Vatsal: You have written about the mechanistic philosophy “that developed a conception of the universe in purely mechanical terms”. You note how this “mechanization of nature proved an important driving force behind the Scientific Revolution, and at the end of the 17th century culminated in Newton’s theory of motion”. There was a controversy that Newton’s theory of universal gravitation allowed for interactions between bodies without contact, resembling, his contemporary critics pointed out, the unintelligible occult forces that the mechanical philosophy was intended to eliminate. Newton appears to have favored some kind of instrumentalism at times, as when he famously wrote, “I frame no hypotheses”. More recently, Stephen Hawking described himself as taking “the positivist viewpoint that a physical theory is just a mathematical model and that it is meaningless to ask whether it corresponds to reality. All that one can ask is that its predictions should be in agreement with observation”. You have described yourself as a realist with respect to the external world, but noted that you “do have strong sympathies for instrumentalism in philosophy of science”. What do you think about this kind of instrumentalism? As with AI systems, do you feel that there is something that a scientific theory ought to provide beyond successful predictions?

Massimo Pigliucci: I sympathize with instrumentalism because it is a minimalist, and very convincing, way of looking at scientific theories: we produce models of the phenomena we observe, and if these models are successful at predicting new phenomena we are satisfied. Otherwise, we set aside the model and build a new one.

And yet, there is something profoundly unsatisfying about this pragmatic view of science, and I honestly don’t believe that Hawking—the same guy who said that his theories allow us to look into the mind of God—was truly an instrumentalist.

I think the objective of science is not just to predict the phenomena, but to give us better and better ideas about what Kant called the noumena, that is, the layer of reality that produces the phenomena.

The notion of realism in science becomes controversial and debatable only when we are talking about the very frontiers of physical science, where we are pushing the limits of human understanding. But consider more mundane examples instead. When Galileo observed certain structures around the planet Saturn and concluded that they were not satellites, but rings, did he mean just to produce a model? No, he thought that there really are rings around Saturn. And subsequent telescopic observations, and eventually spacecraft flybys, confirmed his hypothesis. That is good and satisfactory science.

Similarly, when Watson and Crick hypothesized that DNA is structured as a double helix they didn’t mean their hypothesis to be simply a model capable of accounting for the X-ray crystallography data produced by Rosalind Franklin. They meant to say that DNA really is structured as a double helix. And it is, as direct electron microscope observations have amply confirmed.

The issue of instrumentalism only arises in fundamental physics and, to a lesser extent, in cosmology, that is at scales of the ultra-small and the super-large. That’s because at those scales a lot of useful analogies and metaphors break down and we are pushing the limits of our understanding. But even if human understanding should turn out to be forever limited, that doesn’t mean that there is no actual reality out there, and we have been able to correctly (it seems) grasp an increasing portion of it, ever since the beginning of the scientific revolution.

Vatsal: A natural question that arises concerns the nature of justification or explanation, and here people sometimes invoke the four kinds of causes that Aristotle proposed. You have dealt with the claim that “modern science ever since Francis Bacon has illicitly dropped two of Aristotle’s famous four types of causes from consideration altogether, thereby unnecessarily restricting its own explanatory power”, arguing instead that “Darwin made it possible to put all four Aristotelian causes back into science”. You have described science as a Wittgenstein-type cluster concept, which is “characterized by a number of threads connecting instantiations of the concept, with some threads more relevant than others to specific instantiations”. Do you think a broader view of explanations beyond efficient causes expands the explanatory power of science against the kind of reductionism that you have argued against, according to which “the genetic-molecular level of analysis is the fundamental one in biology, and that everything else, from cell behaviors to ecosystem functioning, ultimately reduces to the properties of the molecules of inheritance”?

Massimo Pigliucci: You raise a couple of distinct, yet related points. Let’s start with Aristotle’s causes. He famously argued that if, say, I want to have a complete understanding of Marcus Aurelius’s statue on the Capitoline Hill in Rome I need to describe it in terms of four “causes,” or explanatory aspects: (i) Material cause — what is the statue made of? (ii) Efficient cause — what made the statue possible? (iii) Formal cause — what is the structure or form of the statue? (iv) Final cause — what is the statue for?

In the case of Marcus’s statue the answers are: (i) Bronze; (ii) A particular (unknown) sculptor; (iii) It represents the emperor on a horse; (iv) It was made to honor the emperor.

Now, supporters of Intelligent design creationism, like William Dembski, claim that Francis Bacon, one of the early philosophers of science, eliminated both formal and final causes from the realm of science, which—in their opinion—resulted in a science that is crippled and incomplete. The two missing causes are the so-called teleonomic ones, because they concern the apparent purposefulness of a thing (from the Greek telos, purpose, and nomos, law). What I argue instead is that teleonomic questions simply don’t apply to sciences like chemistry and physics. It makes no sense to ask why atoms or galaxies are the way they are. They are that way as a result of the laws of physics, period.

But in the case of living organisms, it does make perfect sense to ask “why”? Why are the eyes of a vertebrate made of those particular materials and structured in that particular way? Because their function is to see. Atoms and galaxies don’t have functions, but biological structures do. The difference is the phenomenon of natural selection, the fundamental Darwinian insight that explains why living organisms are the way they are.

To get back to your second point, the one concerning reductionism, yes, including the teleonomic causes in our explanations requires a broader, more organic view than straightforward reductionism allows, because addressing all four of Aristotle’s causes necessitates multiple levels of analysis, not just the most basic one. Biological explanations do certainly benefit from molecular biology, biochemistry, and all the other “reductionist” sciences. But that’s not enough. We also need the organismal disciplines, like evolutionary biology and ecology, to more fully make sense of what Darwin called this grand (evolutionary) view of life.

Continue Reading:
→ A Conversation with Martin Frické on the Epistemology of Machine Learning
→ A Response to Noam Chomsky on Machine Learning and Knowledge
→ Every Problem Is a Prediction Problem

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

Martin Frické is an Emeritus Professor at the University of Arizona. He received his PhD in logic and scientific method from the London School of Economics. He has taught networking, human-computer interaction, logic, and web design, and has spent the latter part of his career studying logic and librarianship, specifically the use of computers and symbolic logic to organize information. As a computer programmer and developer, he has written programs to assist with instruction, many of which are in use the world over. His most recent book is Artificial Intelligence and Librarianship. You can learn more about him here.

Vatsal: I find it fascinating how certain tendencies and positions manifest again and again in the history of thought. A perennial example is the rough distinction between those who place greater emphasis on our innate structure and those who place greater emphasis on experience or data in their conception of knowledge. We may be familiar with Raphael’s famous painting, the School of Athens, which places Plato and Aristotle at the center—Plato pointing toward the heavens and Aristotle toward the ground—symbolizing their contrasting approaches and beliefs. In his dialogue Meno, Plato presents us with a puzzle: how is any inquiry possible if it is impossible to inquire into what we don’t know—since we couldn’t search for it and wouldn’t recognize it even if we found it—or into what we do know, as we already know it? In response to this problem, Plato suggested that what we consider learning is actually prompted recollection: we possess at birth knowledge in a latent form obtained in an existence prior to this life, which is only brought out by the prompting of experience. Aristotle, by contrast, emphasized the essential role of experience in the formation of knowledge. According to him, experience not only allows us to initiate inquiry but also serves as the standard against which its success is measured.

In the 17th century, John Locke compared the initial condition of the human mind to “white paper, void of all characters, without any ideas”. In response to the question, “Whence has it all the materials of reason and knowledge?”, he answered, “in one word, from experience”. For Locke, all knowledge derives from experience. All complex ideas, including those about things that do not exist in nature, can be reduced to simpler ideas, ultimately tracing back to elementary ideas that are not created by us but are obtained from experience. While Locke denied the existence of innate ideas, Gottfried Leibniz affirmed it. Leibniz wrote a book-length response to Locke, placing Locke’s system in the tradition of Aristotle and his own in the tradition of Plato. Referring to a scholastic maxim—that nothing is in the intellect which was not first in the senses—he added, “except for: the intellect itself”. Against Locke’s white paper, Leibniz argued that “the soul contains from the beginning the source of several notions and doctrines, which external objects awaken only on certain occasions”. For Leibniz, “senses never give us anything but instances”, but “all the instances confirming a general truth, however numerous they may be, are not sufficient to establish the universal necessity of that same truth, for it does not follow that what has happened before will always happen in the same way”.

It is interesting how these sides have reappeared in our era of artificial intelligence. Consider the arguments of Noam Chomsky against machine learning. Machine learning, the approach responsible for the recent advances in artificial intelligence, relies heavily on data. Rather than relying on explicit programming, an approach that was popular early in the history of artificial intelligence, machine learning involves identifying patterns in large datasets that enable predictions. For Chomsky, this is not how the human mind functions. The human mind is not “a lumbering statistical engine for pattern matching”. It does not seek to “infer brute correlations among data points”, but to “create explanations”.

Chomsky is sympathetic to Plato’s conception of knowledge. He mentions how “Leibniz pointed out that Plato’s theory of reminiscence was basically correct, but it had to be purged of the error of reminiscence — in other words, not an earlier life, but rather something intrinsic to our nature”. According to Chomsky, Plato’s pre-existence can be reconceptualized in a way as the lives of our ancestors, that is, our genetic endowment. It is this genetic endowment that allows a child to acquire language at a specific period during maturation despite limited and often corrupt linguistic data. To use his analogy: although growth would not occur without eating, it is not the food but the child’s inner nature that determines how growth occurs. Likewise, it is not the linguistic data but the child’s biological endowment that determines how language is acquired. Since machine learning systems operate not based on explicit rules but rather on patterns found in data, they lack the constraints that the human mind possesses. “Humans are limited in the kinds of explanations we can rationally conjecture,” Chomsky observes, while “machine learning systems can learn both that the earth is flat and that the earth is round.”

You have written about machine learning and its inability to generate explanations. Are your reasons for this position similar to those of Chomsky?

Martin Frické: Let me redescribe some of Chomsky in my own words. One central problem that Chomsky addresses is the question of how it is that infants of different ethnicities or cultures can learn their relevant languages extremely rapidly and on the basis of limited data or examples. Chomsky posits innate abilities that are universal. These abilities are not innate ideas or innate knowledge. But, in their most recent version, they might be thought of as principles and parameters (a series of ‘switches’). When a baby hears certain sample words or conversation snippets, switches are ‘flipped’. Thus, a Japanese baby in a Japanese family will speak Japanese and had the same baby alternatively been brought up in an English family, the infant would speak English. An important part of the explanatory challenge here is the paucity of the data that the baby is exposed to. The infant does not hear lots of Japanese, or lots of English, relative to the considerable linguistic skills that the infant acquires. It is the innate principles that facilitate this.

Chomsky’s suggestion here has the status of being an explanatory scientific hypothesis. It is not a piece of armchair philosophy (or armchair linguistics). When detail is added to it, and it is subject to test, there has been considerable evidence in its favor. There has also been considerable counterevidence. (For example, Ibbotson and Tomasello have a paper, Evidence Rebuts Chomsky's Theory of Language Learning). Chomsky’s final principles and parameters account is only one of the modifications he has made to his theories in the light of evidence.

What we are talking of here is science. Going back to innate ideas and innate knowledge in philosophy, they are not science. We might locate the scientific revolution perhaps with Galileo and his contemporaries, say around 1600. Plato was not doing science, nor was Locke (who was later than 1600). If we try to apply some philosophical categories, Chomsky’s theories are contingent, synthetic, and a posteriori. They are science, fallible hypothetical science, right or wrong.

Vatsal: I have provided a detailed response to Chomsky’s arguments in my essay, A Response to Noam Chomsky on Machine Learning and Knowledge. For Chomsky, the deepest flaw of machine learning is its inability to generate explanations. “The crux of machine learning is description and prediction,” he states, “it does not posit any causal mechanisms or physical laws.” He refers to predictions without explanatory insights as pseudoscience. In my essay, I argued that, building upon David Hume’s analysis of causality and the explanatory model articulated by Carl Hempel and others, any conception of causal mechanisms or physical laws can ultimately be reduced to finding regularities and making predictions. Our biological endowments only aid us in this process, as do the architectures of animals and machine learning systems.

In response to what I consider to be Chomsky’s strongest argument—that “machine learning systems can learn both that the earth is flat and that the earth is round”, and therefore cannot apply the rules of inference to distinguish between truth and falsity—it can be pointed out that these rules are not innate to us. There is no determinate pattern in how their application manifests in our behavior; we do not always follow them, and we may act according to them without fully recognizing it. We can compare this to moral rules, where not committing theft is not an innate behavior, although some tendencies may predispose us to it, with or without an explicit understanding of the rule.

In my essay, I further argued that the data accessible to a system presents a universal constraint, so that “corrupt data misleading a machine learning system is not dissimilar from a Cartesian deceiver misleading a human”. Relying on predictions helps us avoid the pitfalls of being misled by seemingly satisfactory explanations, I argued, “as demonstrated by the success of modern natural science which offers predictions that pre-modern natural philosophy, despite its elaborate explanations, failed to offer”.

Martin Frické: You accurately present Chomsky as contrasting machine learning large language models (LLMs) with Chomsky’s own theories of language acquisition (Katz 2012; Chomsky, Roberts, and Watumull 2023). Just as background: LLMs work by statistical inference. They correlate or predict tokens in sequences from the input data of vast amounts of text. Tokens are typically larger than single characters, but smaller than single words. Then you assert:

[Chomsky] contends [that the way LLMs work] is not how the human mind works.

Chomsky is surely right here. Japanese or English babies, or any other human for that matter, are simply not exposed to the vast amounts of text that LLMs require.

You then consider something else—what you call the ‘empirical’ view or ‘empirical’ question—might mere correlations in exposure to linguistic data be the origin of human access to some linguistic capabilities, nothing innate required? LLMs seem to show that considerable linguistic skills can be acquired by computers just from data. Might human ‘half-baked’ LLMs be the origin of some human linguistic skills? Chomsky thinks not. Basically, he is a hypothetico-deductivist and is skeptical of the inductivist view that theory-free studying of data will lead to discovery. Chomsky here is absolutely in the mainstream of philosophy and history of science. My own views on scientific method and discovery, as expressed in, say, Big Data and Its Epistemology (Frické 2015), more-or-less coincide with Chomsky’s on these points.

There is an embarrassment, though. LLMs can have exceptional language skills, skills apparently acquired solely from data. Prior to 2018, most hypothetico-deductivists would have thought that impossible. This suggests that we should look in more detail at what machine learning approaches and LLMs are doing. Machine learning in general is not theory free. There is always what is called inductive bias (bits of theory that come out of thin air). This, for example, might determine the form of a conjectured regression curve. LLMs themselves have a host of theories in them: that the processes are Markov processes, that certain kinds of gradient descent can be used for optimization, etc.

Your learning and knowledge essay then moves on to other topics which are not of direct concern to us here. You move on to causality, correlations, and explanations. These are huge philosophical topics. But distilling it down. Most philosophers, scientists, and ordinary people would like to have a distinction between correlation and causation. Then once you have causation that can be used in explanation (roughly, in many cases to explain X is to give the causes of X). Also, with causation, sense can be made of counterfactuals (what might happen under different circumstances) and decision making (weighing up possible courses of action). An example might help here:

Smoking is correlated with lung cancer. Smoking is also correlated with cirrhosis of the liver. Were a smoker to stop smoking, that would reduce the possibility of him or her getting lung cancer. Were a smoker merely to stop smoking, that would not reduce the possibility of him or her getting cirrhosis of the liver. Why is this? Well, smoking causes lung cancer, smoking does not cause cirrhosis of the liver. Cirrhosis of the liver is caused by drinking alcohol. (When many folk drink they typically also smoke.)

Now Hume, and possibly you echoing Hume, might say: ‘oh, there are only correlations here’. They would be right that we do not observe any causation. But invoking laws and causation is one move we have had to make sense of our world.

A main reason Chomsky rejects LLMs is that they permit impossible languages. Discussion in this area is way above my pay grade. But let me see if I can make some kind of offering. (There is a book by Andrea Moro on this topic [Moro 2016].) Some languages are impossible for humans. For example, a language that requires more memory than humans have. However, some LLMs can generate and work with some of the impossible languages. So, human capabilities and those of LLMs cannot be one and the same.

In Chomsky’s work at large he sometimes uses a competence/performance distinction. Roughly, this contrasts between what is going on ‘inside’, inside the brain or mind, and what appears to the world outside. Consider a person doing simple arithmetic. Their performance, what they can do, getting right and wrong, is available. A hand-held calculator would usually have a far superior performance to that of a human. (The relevant inner competencies of the human and the calculator need not interest us.) In the case of human language, an LLM may well exceed a human in performance—that is why we can be fooled by chatbots and ChatGPT. But the competencies behind those performances are different, and, I think in Chomsky’s view, the language competence of a human cannot be that of an LLM.

In Chomsky’s guest essay, Noam Chomsky: The False Promise of ChatGPT (Chomsky, Roberts, and Watumull 2023), Chomsky basically focusses on philosophy of science. He argues a fallibilist realist hypothetico-deductivist position, pretty much Popperian falsificationism. Then he says: Chomsky-style linguistics meets the demands of this philosophy and LLMs do not.

Steven Piantadosi has a paper, Modern Language Models Refute Chomsky’s Approach to Language (Piantadosi 2023). This is carefully argued research linguistics. I do not have the training or skill to assess it in depth. But there is a point I want to make about it (which applies also to similar papers). Piantadosi has as the last sentence of his abstract:

Most notably, large language models have attained remarkable success at discovering grammar without using any of the methods that some in linguistics insisted were necessary for a science of language to progress [emphasis added].

The point is this. Chomsky’s view is that humans learn language (discover grammar) by means of an innate universal grammar. He does not assert that it is necessary that there is innate universal grammar. He asserts that innate universal grammar is the correct scientific hypothesis to explain human capabilities. This matters in the following way. If LLMs provide a different way of discovering grammar (for powerful computers) that affects Chomsky’s account not at all. What is needed for LLMs to have an impact on the Chomsky vs LLM debate is for it to be asserted that humans, particularly infants, learn language as LLMs do. Then, if anyone asserted this, they would be wrong thanks to humans simply not being exposed to enough data, to enough text.

To sum up, Chomsky has invoked innate abilities to explain human language acquisition and the language skills we have. He was proposing some science. I think that evidence shows that his views are false (see, for example, the Ibbotson and Tomasello paper). I am no expert on this. It is certainly open to discussion by those more expert than I. LLMs are good, indeed very good, at language performance. Chomsky asserts that humans are not LLMs. I think he is right in this, and the arguments he offers are sound. Chomsky favors a fallibilist realist hypothetico-deductivist approach to the philosophy of science. (I do have a soft spot for this. It was popularized in the London School of Economics in the 1960s, and forward, by Karl Popper and others. My training is from that institution at that time.)

Your conversation piece draws attention to similarities between the debate over innate knowledge in the history of philosophy and the debate between Chomsky and proponents of LLMs over innate abilities. I don’t really see the analogy as extending too far as one debate is philosophy and the other science. Your essay on Chomsky gets into causality, explanation, and the philosophy of science. It favors instrumentalism. That is the view that in science it is only predictions that matter (not explanations). Instrumentalism is the received view nowadays. This largely comes from modern science, especially quantum physics. So, you are in the right place—among intellectual friends. Nevertheless, there still are some old Popperian fogies in favor of fallibilist realism.

Continue Reading:
→ A Response to Noam Chomsky on Machine Learning and Knowledge
→ A Conversation with Massimo Pigliucci on Scientific Explanation
→ Every Problem Is a Prediction Problem

Other Projects:
→ Universal Open Textbook Initiative (free, multilingual textbooks)
→ Aesthete (visualize your taste — iOS app)

1. Noam Chomsky on Machine Learning

Recent advances in machine learning have revived an old criticism: that these systems do not possess genuine knowledge. Noam Chomsky has long voiced doubts of this kind. More recently, he co-authored an essay predicting that machine learning “will degrade our science and debase our ethics by incorporating into our technology a fundamentally flawed conception of language and knowledge.”

Chomsky’s main concern is with the statistical nature of these methods. Machine learning systems operate without explicit programming, instead identifying patterns in data and making predictions based on them. This, Chomsky argues, is not how the human mind works. The human mind is not “a lumbering statistical engine for pattern matching”. It does not seek “to infer brute correlations among data points” but “to create explanations”.

This skepticism follows naturally from Chomsky’s broader intellectual commitments. Referring to Plato’s Meno—where Socrates shows how an uneducated slave can exhibit knowledge of geometric principles with only a few prompts—Chomsky asks: How does a child acquire so much knowledge so rapidly, with so little evidence? Like Immanuel Kant, he recognizes the selectivity that must originate from us against the shower of data to which we are exposed. And like Plato and Kant, his answer to the possibility of knowledge involves something innate to human nature.

Chomsky views the mind as composed of inborn, interacting faculties. Each faculty operates according to distinct, domain-specific rules that produce various mental phenomena.

Language is one such faculty. Chomsky explains it through a framework of principles (invariant rules common to all natural languages) and parameters (options that are set upon exposure to linguistic data, accounting for differences between languages). To use his analogy: Although growth would not occur without eating, it is not the food but the child’s inner nature that determines how growth will occur. Similarly, it is not linguistic data but the child’s biological endowment that determines how language is acquired.

Opposed to this nativist conception of knowledge and language is the kind of empiricism which denies innate rules for knowledge. Where Chomsky sees the mind as directed by inner principles that selectively use data as part of a fixed developmental program, empiricists see it as forming associations based on experience. Without rigid inner procedures to impose structure, experience plays a far more determining role. Chomsky criticizes this approach for treating the mind differently than we treat other bodily systems—all of which we assume have innate structures.

An approach focused solely on associations from observed data, Chomsky argues, would fail to discover the underlying principles from which certain outcomes—and only those outcomes—follow. He sees scientific inquiry as the pursuit of such principles and emphasizes abstraction and idealization, as with thought experiments:

More generally in the sciences, for millennia, conclusions have been reached by experiments–often thought experiments–each a radical abstraction from phenomena. Experiments are theory-driven, seeking to discard the innumerable irrelevant factors that enter into observed phenomena ... the basic distinction goes back to Aristotle’s distinction between possession of knowledge and use of knowledge. The former is the central object of study.

In machine learning, where associations dominate rather than rules, Chomsky sees echoes of the empiricist approach. He doubts that statistical techniques for finding patterns in data will yield explanatory insights. A notion of success that involves predictions but not explanatory insights, he remarks, has little precedent in science. He calls such predictions pseudoscience.

2. Two Distinct Claims

Chomsky makes two distinct claims that are worth separating.

The first concerns how the human mind differs from machine learning systems. “The human mind,” Chomsky says, “is a surprisingly efficient and even elegant system that operates with small amounts of information.” Referring to the innate language faculty “that limits the languages we can learn to those with a certain kind of almost mathematical elegance,” he points to a lack of similar constraints in machine learning. The same applies to knowledge: “humans are limited in the kinds of explanations we can rationally conjecture,” while “machine learning systems can learn both that the earth is flat and that the earth is round.”

This claim is unsurprising. The human mind—a part of the human organism shaped by evolution over countless generations—naturally differs from machine learning systems, which are non-biological creations developed using human methods and data. If Chomsky’s only point were that the human mind operates differently from machine learning systems, it would be a rather trivial observation.

But Chomsky appears to be making a second, more general claim about the nature of knowledge itself. He identifies the deepest flaw of machine learning programs as:

the absence of the most critical capacity of any intelligence: to say not only what is the case, what was the case and what will be the case—that’s description and prediction—but also what is not the case and what could and could not be the case. Those are the ingredients of explanation, the mark of true intelligence ... The crux of machine learning is description and prediction; it does not posit any causal mechanisms or physical laws.

To define knowledge this way, one must rely on a universal conception of what knowledge is—one that could apply equally to animals, humans, and artificial systems, free from human biases. With such a conception, we could distinguish what is essential to knowledge from what is merely instrumental. Chomsky, however, seems to be using a particular way of acquiring knowledge, with its unique constraints, to conclude that any different path would fail to yield genuine knowledge.

Consider an analogy with locomotion. There are different ways to move from place to place. For humans, walking is the natural method—enabled by a particular bodily architecture with particular constraints. A wheeled vehicle has a different architecture with different constraints. If we compared them, we would note their differences: one walks, the other rolls, and so they are not identical. Yet if we understand locomotion as a general concept—independent of the particular means through which it occurs—then both walking and rolling are genuine instances of it. We might legitimately ask which is more energy-efficient, but there would be no question of which represents “genuine” locomotion and which merely simulates it.

What should concern us, then, is the nature of knowledge itself. What are its core features that allow us to identify it regardless of how it is achieved? I will argue that associations and predictions are sufficient to capture the essential features of knowledge. Everything that aids in achieving them—whether biological endowments, algorithms, or epistemic techniques—holds only instrumental value.

3. Associations and Predictions

Associations, Platonic forms, causal relations, and physical laws are all instances of what we can call “propositions”. A proposition emerges from accounting for, or linking, data.

Consider a simple example. Suppose we traverse an area and develop an internal representation of its physical features. We have linked some material available to us—material that increased as we moved through the space. If we externalized this representation using pencil and paper, it would take the form of a map, where points are connected in particular ways. What we did internally is analogous to linking points on paper.

As Chomsky himself believes about cognitive operations, this process of developing propositions is independent of individual awareness.

The process is also independent of the ability to externalize a proposition. Our internal representation exists before we translate it into lines on paper and would exist even if we never externalized it. That an animal cannot create diagrams or utter sentences does not prevent it from discovering propositions.

Everything that aids in accounting for data contributes to one’s capacity for knowledge. Some contributions are internal (like memory); others are external (like the mechanical calculator or the scientific method). Since memory can deteriorate with age, and since we can be exposed to new inventions through cultural diffusion, our capacity for knowledge is subject to variation.

What is the material that we account for when discovering a proposition? If we define data as that which is accounted for in this process, what properties does it have? Understanding these properties will clarify how different types of propositions are possible and help address some of Chomsky’s concerns.

The material that forms propositions has properties paralleling those of physical objects. Just as a physical object consists of a quantity of material that can increase or decrease, the amount of data available to us fluctuates over time. Data, like physical materials, comes from diverse sources. The divisibility of a proposition lets us appreciate the granularity of its constituent data, much like dividing an object into smaller parts reveals its material composition. This granularity enables propositions to be about anything discoverable in data, just as different combinations of particles yield distinct objects.

Consider each of these properties in turn.

Accumulation. As illustrated in the map example, data can become available to us after previously being unavailable. New data is constantly added to us, even while we sleep. What was once unavailable becomes accessible at each moment, contributing to the totality of data available to us.

Diversity of sources. In traversing an area, we encounter one source of data: direct observation. Other sources might include photographs of the area or written descriptions. Some sources provide more value by allowing us to account for more data with less effort. Instead of physically traversing an area, we might rely on a photograph to form our mental map. To the extent the photograph is representative, the features in our proposition will align with any actual visit we make. Our ability to discriminate among different sources emerges through experience with data itself.

Granularity. Philosophers have long drawn an analogy between how particles combine to form physical objects and how simple ideas from experience combine to form complex ones. We can think of this in terms of divisibility: any part of a physical object can be distinguished as a constituent, which can have its own divisible parts. Similarly, a proposition contains other propositions presupposed within it.

Consider the proposition “whenever one ball strikes another, the other ball moves.” Within it, we find presupposed such propositions as “a ball is a round body,” “to move is to go from one place to another,” and even “a thing cannot both be and not be at the same time.” When we form a proposition, we implicitly account for all the data accounted for by the propositions it presupposes.

This granularity allows for the different kinds of propositions we discover. Some pertain to what we can observe: both what we have observed and what we have yet to observe. If we ask through what data we arrived at the ball proposition, we might say we had a series of experiences directly observing one ball strike another. But this is not the only path. Suppose we have never directly observed one ball strike another but have observed objects colliding, seen pictures of balls, and read about what happens when balls collide. We could form the same proposition without ever witnessing such an event directly.

Beyond propositions about what we can observe, we can also form propositions about what we cannot observe. Suppose we hear a sound from a tape recorder. We cannot observe the original source—it is in the past and not present before us. Yet by combining the data from the recording with what we already know, we can infer that the original speaker was a woman, that she has a quiet disposition, that she felt lonely the night before. We might have reached these same judgments had she been speaking in front of us.

We never find, in imagination’s ingredients or products, anything not present in some form within the data available to us. This helps explain how propositions can be about what does not exist. In nature we may observe lions and men, but a lion-man—a figure with a lion’s head and a man’s body—exists only in imagination. Yet this figure could never have occurred to us without first observing the elements we used to create it.

While propositions are discovered based on available data, the data a proposition accounts for can also include data yet to become available. When we account for certain currently available data, we treat multiple instances as examples of a single state of affairs. When we treat data yet to become available as further instances of that state of affairs, we form predictions. If these predictions agree with the data when it eventually becomes available, the proposition has successfully accounted for that data.

We can compare propositions if one accounts for at least all the data another accounts for. The proposition that accounts for more data is truer. Continuing our analogy with physical objects: truth is a gradable property, much like physical size. We can describe an object as large relative to a certain range, or larger than another object, or as having a specific length. Similarly with truth: we can describe a proposition as true relative to a range, truer than another proposition, or as accounting for specific data.

A problem in Plato’s Meno concerns the possibility of inquiry: How can we search for what we don’t know, since we wouldn’t recognize it if we found it? And why search for what we already know? This puzzle dissolves once we recognize that knowledge consists in accounting for data, that truth refers to the amount of data accounted for, and that data becoming available is simply one of the transformations occurring in the world.

From this account, we can see that animals, humans, and machine learning systems differ in their capacities for knowledge partly because of differences in their architectures. These architectures—more malleable and improvable in machine learning systems than in biological organisms—create tendencies to account for data in specific ways. They also constrain what kinds of data a system can access. A human and a bat exposed to the same environment for the same time are not accounting for the same data. Architectures also determine how external contributions can enhance a system’s capacity for knowledge: education increases a human’s capacity but not an animal’s.

Despite these differences, a universal constraint across all systems is the kind of data available for building propositions. Corrupt data misleading a machine learning system is not fundamentally different from a Cartesian deceiver misleading a human. The journey from error to knowledge is the discovery of truer propositions. In this process, unsuccessful predictions by a machine learning system are no different from those made by a human.

The ultimate reference for truth and falsehood is what occurs in the world. Propositions about what does not occur, whether possible or impossible, can only be true insofar as they point to what does occur. Using propositions about possibilities to arrive at propositions about actualities can be understood as an epistemic technique. Other, yet undiscovered techniques may exist. If it is possible to reach a true proposition without relying on the particular properties of the human system, then the constraints of that system are merely instrumental.

The endowments of biological entities can also be considered independently of knowledge—as pure manifestations of bodily processes, like the pumping of the heart. In that case, as Kant said of the senses, they would form no judgment, correct or incorrect. They would not be involved in accounting for data, which is the basis for truth and falsehood. The faculties of the human mind contribute to knowledge by enhancing a person’s capacity for it, as when the language faculty helps us read a textbook. We might compare the mind’s contribution to knowledge with the limbs’ contribution to locomotion.

This account also clarifies how predictions make the internal process of accounting for data concrete. When someone claims to understand something, we assess this by asking questions or presenting problems, then comparing their predictions against a benchmark. We treat past actions as records of previous predictions. Since the capacity for externalization is independent of the capacity for knowledge, it is possible to demonstrate knowledge without being able to articulate it.

Predictions thus help us distinguish knowledge from its absence in any system—animal, human, or machine. We can raise legitimate questions about which system performs better on a specific metric, such as how much data is required to reach a particular prediction. But there is no meaningful question about which possesses “genuine” knowledge and which “merely simulates” it. Any such demarcation would be arbitrary, like saying a system achieves genuine locomotion only once it crosses a certain speed or efficiency threshold.

4. Explanations

Chomsky’s contention that knowledge requires something more than description and prediction echoes an argument from Socrates: even if we form beliefs that happen to be true, we need something more for genuine knowledge.

In Plato’s Theaetetus, Socrates illustrates this with the example of lawyers who persuade jurors about criminal acts. If jurors conclude that a defendant is innocent based on a compelling argument, we cannot say they possess knowledge of his innocence—even if he is indeed innocent—since we can imagine them reaching the opposite verdict had the lawyer argued otherwise. We can form all kinds of beliefs, some of which might be true by mere chance. Socrates suggests that a true belief must be fastened by explanation and made reliable to become knowledge.

What, then, is an explanation? In Aristotle’s Posterior Analytics, we find:

We suppose ourselves to possess unqualified scientific knowledge of a thing, as opposed to knowing it in the accidental way in which the sophist knows, whenever we think we are aware both that the explanation because of which the object is is its explanation, and that it is not possible for this to be otherwise.

An explanation, then, is an answer to the question of why a particular fact is the way it is and not otherwise. In an explanation, the fact to be explained is shown to be an instance of a general proposition—a causal relation or universal law. A particular ball moving after being struck can be explained by the general proposition “whenever one ball strikes another, it causes the other ball to move.”

Let us examine this example to understand causal relations. Our intuitive notion of causation suggests that the first ball acts on the second upon contact, causing it to move by necessity. However, as David Hume argued, what we actually discover in causal relations is merely a constant conjunction between pairs of events. The necessity we attribute to this connection—the sense that one event must follow the other—is not something we directly observe.

Strip away this obscure notion of necessity, which adds nothing to the relationships we identify despite any significance we attach to it, and what remains are associations as discovered by a system with certain capacities. More precisely, causal relations are propositions in which relations of precedence and succession between states of affairs are presupposed, with some propositions being truer than others.

In explaining why, when one ball strikes another, the other moves, we can also invoke a law of nature, such as “the total momentum of a system remains constant.” This law is a truer proposition than “whenever one ball strikes another, the other moves.” It accounts for all the data the simpler proposition does while also accounting for more. A law of nature, then, is simply a type of proposition distinguished by the large amount of data it accounts for.

If finding causal relations and laws of nature consists of identifying propositions of certain kinds, then explanation—where it is taken to demonstrate knowledge—involves having a sufficiently truer proposition than the one being explained. Such a truer proposition contains within it the limits that made the other proposition falser, just as an amended map contains the constraints of the older map. This is evident in the predictions we form based on the truer proposition.

There is, however, a risk when explanations are viewed as the goal of all inquiry. Because accounting for data is an internal process, our ignorance can become obscured in the pursuit of explanations. We may mistake a satisfying subjective state—the feeling of clarity or simplicity a particular explanation generates—for actual knowledge. Predictions help mitigate this risk, as demonstrated by the success of modern science, which offers predictions that pre-modern philosophy, despite its elaborate explanations, failed to offer. Anything involved in developing explanations is relevant to knowledge only insofar as it helps account for more data.

Once we understand how predictions are made, we can acknowledge that a machine learning system making predictions must have based its conclusions on something it discovered, even if it cannot externalize that discovery. While this presents its own risks—such as hindering the cumulative knowledge-building that externalization facilitates in humans—as a fact about the possession of knowledge, it resembles the case of a person who cannot articulate a proposition but demonstrates it through successful predictions. Each successful prediction reflects the data accounted for by a proposition, and it is on this basis that we can distinguish accidentally true beliefs from more reliable ones.

In the case of Socrates’ jurors, what yields knowledge rather than mere true belief is the identification of a truer proposition—one like “the accused is innocent because of such-and-such evidence, and the lawyer’s persuasiveness is independent of this,” rather than “the accused is innocent because otherwise the lawyer would not have been so convincing.”

Or consider our ball example: we might observe that one ball strikes another but the second does not move. We would then realize that any explanation that had satisfied us, whatever its other uses, had limits attributable to nothing else but the limited data accounted for by our proposition. This could prompt us to identify a truer proposition, such as “whenever one ball strikes another, the other moves, except when it is significantly heavier,” or eventually, “the total momentum of a system remains constant.” It is an observed consequence of a proposition being sufficiently true that it leads us to discover what would or would not occur under particular circumstances.

5. Conclusion

When we understand the nature of knowledge, we recognize that it is aided by a system’s endowments—such as those innate to humans—but not solely defined by them. We can imagine these endowments independently of knowledge, while also conceiving of entities that lack the same endowments yet remain capable of acquiring knowledge.

Continue Reading:
→ Every Problem Is a Prediction Problem
→ A Conversation with Martin Frické on the Epistemology of Machine Learning
→ A Conversation with Massimo Pigliucci on Scientific Explanation

Other Projects:
→ Universal Open Textbook Initiative (Free, multilingual textbooks)
→ Aesthete (Curate your culture — iOS app)