If Every Thought Has a Cause, Could You Ever Have Chosen Differently?

At 8:17 this morning, imagine you were standing in your kitchen. Coffee was on one side. Tea was on the other. You thought about it for a moment, then reached for the coffee.

Nothing unusual happened. It was the kind of decision you make hundreds of times without thinking much about what a decision actually is.

But imagine something impossible. At 8:18, the universe is rewound exactly five minutes.

Every physically relevant condition is restored: the state of your brain and body, your memories, the surrounding world, and whatever else the laws of nature require to specify the situation. Your hunger, mood, hormones, the temperature of the room, the light coming through the window — everything relevant to what happens next is as it was.

It is 8:13 again. Four minutes pass. It is 8:17, and you reach the kitchen.

Coffee. Tea. Could you choose the tea this time?

The thought experiment does not assume that the second run must unfold identically. That is exactly what we are trying to discover. A deterministic theory says the restored state and laws fix what comes next. An indeterministic theory may allow more than one physical continuation. And a theory of free will still has to explain what, if anything, would make the outcome your choice.

If absolutely nothing relevant has changed, where would a different decision come from? And if you could not possibly choose differently, what exactly did we mean when we said that you chose coffee the first time?

This is where the problem of free will usually begins. But there is another question hiding underneath it:

Why are we assuming that the universe works that way in the first place?

The tempting idea

There is an extraordinarily powerful intuition behind determinism: things happen because other things happened.

The glass broke because it hit the floor. It hit the floor because your hand released it. Your hand released it because muscles contracted. Those muscles contracted because signals travelled through nerves, and those signals came from activity in your brain.

Your current brain contains memories. Those memories came from experiences. Your reactions were influenced by biology, childhood, culture, education, relationships and everything else that happened before this moment.

Past → brain → thought → decision → action

But there is a trap here, and it matters for everything that follows.

Having a cause is not the same thing as being determined.

To say that a decision has causes is only to say that something contributed to why it happened. Determinism is the stronger claim: given the complete relevant state of the world and the laws of nature, only one future is physically possible. Philosophers therefore distinguish questions about causation from the separate thesis of determinism. There is no general agreement that determinism is true, or that our best physics settles the issue in a simple way.

Stanford Encyclopedia of Philosophy — Causal Determinism

"I can think about my decision."

Suppose I offer you cake. You want it, but then remember that you have been trying to eat healthier. You consider the consequences and decide not to eat it.

That certainly looks different from simply obeying an impulse. But under strict determinism, your reasoning does not escape the causal chain. It becomes another part of it.

Upbringing → knowledge → values → memory of your goal → reasoning → decision

The analysis is real. Your reasoning matters. Your knowledge matters. But those things also have histories. Why did that argument persuade you? Why did you care about your health enough to resist the cake? Why were you capable of remembering the goal at that particular moment?

A determined mind would not necessarily be a mind that cannot reason. Quite the opposite: reasoning itself could be one of the mechanisms through which causes produce decisions.

Determinism is not fatalism.

Fatalism suggests that an outcome occurs regardless of what you think or do. Determinism says your thinking, hesitation, learning and action can all be among the causes through which the outcome occurs. Reasons can matter even if reasons themselves have causes.

Follow the Question: Physics

Does physics give us only one future?

For centuries, physics made the deterministic picture deeply attractive. If the motion of matter follows mathematical laws, perhaps a sufficiently complete description of the universe would allow the future to follow uniquely from the present. This idea is often represented by Laplace's demon: an imaginary intelligence that knows the state of nature and all the relevant laws well enough to infer what follows.

Humans could never perform such a calculation for the entire universe. But that is not the point. Determinism is not the same thing as predictability. A deterministic system may be practically unpredictable because of complexity, chaos or our inability to know its initial conditions precisely.

Then twentieth-century physics made the story much stranger.

Quantum theory gives probabilistic predictions for many experimental outcomes. In collapse-style formulations, individual outcomes are treated as genuinely indeterministic. But quantum mechanics also has deterministic formulations, and that fact needs care rather than a slogan.

Bohmian mechanics, for example, gives particles definite positions whose evolution is deterministic given the wavefunction and configuration. Yet that determinism does not give ordinary observers unrestricted predictive power: the theory still reproduces the statistical predictions of quantum mechanics under its equilibrium assumptions. Deterministic dynamics and practical predictability are different questions.

Everettian or many-worlds approaches are deterministic at the level of the universal quantum state: the wavefunction evolves continuously rather than randomly collapsing. But all non-zero branches are represented, which makes the everyday question "could I have done otherwise?" harder, not easier. The universal dynamics can be deterministic while an observer still confronts branch-relative outcomes and probabilities.

So quantum mechanics did not simply replace a deterministic universe with a universally agreed random one. What the theory says about determinism depends partly on how its formalism is interpreted and what we take the physical state to be.

Stanford Encyclopedia of Philosophy — Bohmian Mechanics

Stanford Encyclopedia of Philosophy — Many-Worlds Interpretation of Quantum Mechanics

Physics does not currently hand us a simple proof that the entire universe is deterministic.

But that does not automatically rescue free will. Suppose some events are genuinely indeterministic. Then more than one physical outcome may be possible. That still does not tell us whether you authored which one occurred. A random fluctuation is not obviously a free choice.

Determinism asks whether the complete state of the world and the laws permit only one future. Indeterminism asks whether more than one physical continuation can occur from the relevant prior state. Free will asks a different question: what kind of control, authorship or reasons-responsiveness would make an action genuinely ours?

The third question is not merely another option beside "determined" and "random." A compatibilist can argue that a determined action may still be free. An incompatibilist can argue that indeterminism is necessary but not sufficient. Either way, escaping determinism is only part of the problem.

Follow the Question: The Brain

Does the brain decide before "you" do?

Perhaps the answer is inside us. Human decision-making depends intimately on brain processes, and this is where one of the most famous experiments in the free-will debate appears.

In 1983, Benjamin Libet and colleagues asked participants to make simple self-initiated movements while researchers measured electrical activity associated with movement preparation. Participants also reported when they first became consciously aware of the urge or intention to move. A signal known as the readiness potential began before the reported time of conscious intention — by several hundred milliseconds in the original study.

The finding was irresistible: brain activity begins before the person reports becoming consciously aware of intending to act. Maybe consciousness is not initiating the decision. Maybe consciousness is simply being informed about a process already under way.

That interpretation escaped the laboratory and became a popular argument against free will. The experimental result and the philosophical conclusion, however, are not the same thing.

A 2021 meta-analysis found a temporal pattern largely consistent with Libet's original result. But the comparison most relevant to claims about conscious causation — the interval between preceding brain activity and reported conscious intention — was represented by only six studies, and the meta-analytic estimate carried substantial uncertainty. The authors concluded that some of Libet's findings appear more fragile than the enormous literature built on them might suggest.

There is also an unresolved question about what the readiness potential represents. Later models have argued that it may reflect an accumulation process or spontaneous neural fluctuations associated with self-initiated movement rather than a fully formed unconscious decision waiting for consciousness to catch up.

The timing method itself is also difficult: participants must retrospectively report when an internal intention became conscious, an event that is inherently hard to timestamp with precision.

A 2023 review of neuroscience and free will surveyed these issues and concluded that current evidence is insufficient to undermine free will. That is a conclusion of that review, not a proof that free will exists.

Libet et al. (1983) — Time of conscious intention to act in relation to readiness potential

Braun, Wessler & Friese (2021) — A meta-analysis of Libet-style experiments

Schurger, Hu, Pak & Roskies (2021) — What Is the Readiness Potential?

Delnatte et al. (2023) — Can neuroscience enlighten the philosophical debate about free will?

Neuroscience complicates the story. It does not finish it.

Follow the Question: Psychology

How much can a choice change while still feeling like ours?

Physics asks whether events are determined. Neuroscience asks how brains generate actions. Psychology asks something slightly different: how stable are the preferences and experiences that make a choice feel like mine?

Human choices can clearly be influenced by context. In classic work, Amos Tversky and Daniel Kahneman showed that logically equivalent decision problems could produce predictable shifts in preference when outcomes were framed differently.

Other research has shown strong effects from defaults. Eric Johnson and Daniel Goldstein, for example, demonstrated how default structures could substantially alter expressed organ-donation choices.

Change the description, context or default and sometimes you change what people choose. That does not prove determinism. But it creates an uncomfortable question: if someone can shift my expressed preference without changing the underlying options, which preference is really mine?

Perhaps both. Humans are not static objects containing a hidden list of permanent preferences. We respond to information, context, memory, emotion and social cues.

Psychology and neuroscience also study the sense of agency: the experience that we are controlling our own actions and, through them, events in the world. Research treats that experience as something constructed from prospective signals about intended actions and retrospective information about outcomes.

But showing that a decision can be influenced does not show that the person has no agency. If I give you good information and you rationally change your mind, the fact that information influenced you may be evidence that your reasoning works — not evidence that reasoning is fake.

Tversky & Kahneman (1981) — The framing of decisions and the psychology of choice

Johnson & Goldstein (2003) — Do Defaults Save Lives?

Patrick Haggard (2017) — Sense of agency in the human brain

Follow the Question: Philosophy

What kind of freedom are we looking for?

Imagine two people. The first is locked inside a cell. He wants to leave but cannot. The second is standing outside. Nobody is restraining her. She considers going inside and decides not to.

The difference seems obvious. One person is prevented from doing what he wants. The other acts according to her own reasons and motivations.

Now imagine that the second person's reasons and motivations were themselves completely produced by previous causes. Was she free?

This is where several major philosophical positions separate.

Incompatibilism and libertarian freedom

Incompatibilists argue that the relevant kind of free will cannot coexist with determinism. If the past and laws fixed the action, then something important about alternative possibility or sourcehood is missing. Libertarian theories add that humans sometimes do possess free will, but serious libertarian accounts do not simply equate randomness with freedom. Event-causal, noncausal and agent-causal theories try to explain a form of control that is neither ordinary determinism nor mere chance.

Stanford Encyclopedia of Philosophy — Incompatibilist (Nondeterministic) Theories of Free Will

Compatibilism and identification

Compatibilists take a different route. They argue that determinism and free will can coexist because freedom need not mean being uncaused. What matters is whether the action flows through the right capacities: understanding reasons, regulating behavior, acting without coercion, and in some accounts identifying with the motives that move you.

Harry Frankfurt's influential work helped shift attention toward this idea of identification. On Frankfurt-style accounts, a person is not simply whatever desire happens to be strongest. Part of agency lies in the higher-order attitudes through which a person endorses, rejects or identifies with particular motives.

That gives compatibilism a strong reply to the demand that a free person must somehow create themselves from nothing. Authorship might instead be a capacity that develops: the ability to reflect on motives, learn, respond to reasons and govern behavior over time.

Stanford Encyclopedia of Philosophy — Free Will (including Frankfurt-style identification accounts)

The regress problem

But the skeptic can press deeper. Galen Strawson's Basic Argument asks how anyone could be ultimately responsible for what they do if what they do depends on the kind of person they are. To be ultimately responsible for the choice, it seems you would also have to be responsible for the mental characteristics from which the choice arose. But any attempt to choose those characteristics would itself depend on earlier characteristics, producing a regress.

Strawson's argument targets ultimate moral responsibility. Even if it succeeds at that level, it does not automatically erase every thinner, practical or forward-looking concept of responsibility.

Stanford Encyclopedia of Philosophy — Moral Responsibility, section on Galen Strawson's Basic Argument

The disagreement is not only about whether humans possess free will. It is about what freedom would have to be before we should expect to find it.

Follow the Question: Law

Does responsibility require metaphysical freedom?

Now the philosophical puzzle stops being abstract.

U.S. criminal law does not ordinarily ask whether a defendant was an uncaused cause. It asks narrower operational questions: Was there a voluntary act? What was the defendant's mental state? Was the conduct purposeful, knowing, reckless or negligent? Was the person coerced? Did a qualifying mental disorder undermine legally relevant capacity?

These questions appear in doctrines of actus reus, mens rea, duress and insanity. They vary significantly by jurisdiction, and criminal law outside the United States can structure responsibility differently.

Cornell Legal Information Institute — Actus Reus

Cornell Legal Information Institute — Mens Rea

Cornell Legal Information Institute — Duress

Cornell Legal Information Institute — Criminal Insanity

A philosophical resemblance, not a legal doctrine.

This structure resembles a compatibilist way of thinking because responsibility is tied to capacities, reasons, control and freedom from certain forms of coercion rather than to metaphysical independence from causation. But U.S. criminal law does not thereby "endorse compatibilism," and legal doctrines do not map neatly onto one philosophical theory.

Neuroscientific evidence has also entered criminal cases as individualized evidence about defendants, though its proper role in culpability, defenses and sentencing remains contested.

Suppose every human action ultimately has causes. A violent person's behavior may have been shaped by biology, development, trauma, environment, learning and immediate circumstances they did not choose. Understanding those causes might make the behavior more explainable. But explanation does not automatically erase every form of responsibility.

Even in a deterministic world, institutions could still have reasons to protect others, deter harmful behavior, rehabilitate, incapacitate dangerous offenders and alter conditions that predictably increase harm. What might change is the justification and emphasis of punishment.

University of Chicago Law Review — Neuroscience and the Personalization of Criminal Law

What punishment, if any, is deserved? And what response best protects others, reduces future harm, respects proportionality and changes the causes that can actually be changed?

The law therefore exposes an important possibility: society can make responsibility judgments without first solving the metaphysics of the universe.

Follow the Question: Artificial Intelligence

When does processing become agency?

Now consider something that did not exist when most classical arguments about free will were developed.

An artificial system evaluates several possibilities. It retrieves information, compares outcomes, revises its response after receiving new evidence, and finally produces: "I considered the alternatives and chose Option B."

A human might reply: you did not choose anything. Your architecture, training, inputs and internal state produced the output.

That is a reasonable causal description of an artificial system. It becomes philosophically interesting because a causal description also applies to us: human behavior depends on biological architecture, development, learning, memory, sensory input and current internal state.

But this analogy has a serious limit. Sophisticated information processing is not obviously the same thing as being a subject for whom a decision matters. Current AI systems can display goal-directed behavior in a thin technical sense, but thicker notions of agency introduce questions about normative understanding, control, self-directed values, consciousness or sentience, and the capacity to be affected by outcomes.

We do not currently have an agreed test showing that today's AI systems possess subjective experience, valenced interests or the kind of self whose commitments could make one motive genuinely "its own." The absence of such a test also means we should be careful about claiming the opposite with metaphysical certainty.

So the useful comparison is not "brains are computers, therefore humans and AI are the same." It is diagnostic: when we deny agency to a machine, what criterion are we applying?

Whichever criteria we choose, AI forces us to state them more clearly — and then ask how confidently humans satisfy the same standard.

Stanford Encyclopedia of Philosophy — Ethics of Artificial Intelligence and Robotics, sections on moral status and agency

Where the Fields Collide

We started with coffee and tea. We now have several different problems that are often compressed into the single phrase free will.

Physics — Does the complete state of the universe permit only one future? Our current physical theories do not settle that question in one interpretation-independent sentence.

Neuroscience — How do brain processes generate intentional action? Relevant neural activity can precede reported conscious intention in some tasks, but what that timing means for agency remains disputed.

Psychology — How stable are preferences, and how is agency experienced? Framing, defaults and context can change behavior without proving that agency is absent.

Philosophy — What kind of control, identification, sourcehood or alternative possibility deserves the name free will? The major positions disagree about the requirement itself.

Law — What kinds of voluntariness, intent, capacity and coercion matter for responsibility? Legal systems make these judgments without first solving metaphysical free will.

AI — When does goal-directed processing become agency? The question exposes how much our own concept of agency depends on consciousness, values, control and identity.

One question has become six, and none can completely answer the original problem alone. That is precisely why the question is interesting.

What We Know — and What We Don't

There are several things we can say with reasonable confidence. Human decision-making depends intimately on brain processes. Many processes involved in producing actions occur outside conscious awareness. Our decisions are influenced by biology, learning, experience, information, environment and current context. Our sense of agency is a real psychological phenomenon that can be studied, even though the existence of that experience does not by itself settle the metaphysics of free will.

Several larger claims remain unsettled. We do not know that the entire universe is deterministic in the relevant philosophical sense. If genuine indeterminism exists, randomness alone does not automatically create agency. Libet-style neuroscience does not establish that conscious human agency is an illusion. And philosophy has not produced a universally accepted account of what freedom must consist in.

"Science proves there is no free will" moves considerably faster than the evidence. So does "science proves free will exists."

The evidence constrains the debate. It does not close it.

Back to the Kitchen

It is 8:13 again. The thought experiment restores every physically relevant condition. Four minutes pass. It is 8:17, and you reach the kitchen.

If strict determinism is true, and the thought experiment really restores the complete relevant state, the same state and laws do not leave room for a different physical future.

If fundamental indeterminism is true, more than one physical continuation may be possible — but possibility or randomness alone would not explain why one outcome counted as your freedom.

If compatibilists are right, perhaps freedom never required an identical universe to produce a different result. Your choice may be free when it flows from the right relationship between your reasoning, values, motives and capacities for self-control — even if those processes have causes.

If incompatibilists are right, that is still not enough. A deeper kind of sourcehood or genuine alternative possibility is required, and the causal history of the agent becomes part of the problem.

Your reasoning has causes. Your character has causes. Your values have causes. None of that makes them causally irrelevant. The harder question is whether being an effective part of the causal chain is enough for authorship — or whether genuine authorship demands something more.

What would have to be true before a decision could genuinely be called yours?

That question points beyond action itself. If a free choice is supposed to express your reasons and values, we eventually have to ask where those reasons and values came from — and what makes them yours.

The Next Question

If you did not choose the forces that shaped your mind, how much of what you believe is actually yours?

Beyond the Question continues.