{"slug":"free-will-and-responsibility","title":"Free will and responsibility","date":"2026-04-22T08:55:49.296Z","excerpt":"","tags":["philosophy","psychology"],"order":72,"published":true,"content":"

Here's a very detailed explanation of Chapter 6: Free Will and Responsibility:

The Core Question

Do humans actually have free will? And if so, how much?

These are ancient philosophical questions, but this chapter approaches them from a new angle: what do psychology and neuroscience tell us about free will? The chapter's conclusion is carefully balanced — the science does NOT prove that free will is completely an illusion, but it DOES suggest we are less free than we typically assume.

Part 1: Setting Up the Problem

Why Free Will Matters

In everyday life, we assume we are free agents. We believe we consciously choose our actions, and we hold people responsible for what they do based on this assumption. Courts sentence criminals. We praise heroes. We blame people who behave badly. All of this assumes that people could have chosen differently — that their actions were freely made.

The problem is that science seems to challenge this assumption. Neuroscience shows that your brain does things before you're conscious of deciding to do them. Psychology shows that your behavior is heavily influenced by factors you're completely unaware of. Both seem to suggest that you, as a conscious deliberating agent, might not actually be in charge of what you do.

The Traditional Physics Debate

Before getting into psychology, the chapter briefly mentions the traditional physics-based debate about free will:

If determinism is true — if every event in the universe is caused by prior events according to fixed laws of nature — then your actions were already determined by the state of the universe before you were born. You couldn't have chosen otherwise. So free will is impossible.

If indeterminism is true — if events happen randomly at the quantum level — then your actions are random. Random actions aren't freely chosen either. You'd just be a dice roll.

Either way, physics seems to leave no room for free will.

But the chapter sets this aside. The philosophical premise in both arguments — that free will is incompatible with either determinism or indeterminism — is actually very controversial. Most contemporary philosophers think free will is perfectly compatible with a deterministic physical world. So the chapter focuses instead on psychological and neuroscientific challenges.

Part 2: The Structure of Scientific Arguments Against Free Will

Every scientific argument against free will follows the same logical template:

Premise 1 (psychological premise): Some empirical fact P is true, according to psychology or neuroscience.

Premise 2 (philosophical premise): Free will is incompatible with P.

Conclusion: Therefore, we don't have free will.

The chapter identifies three distinct versions of this argument:

Argument 1: From Materialism

The brain is a physical object. Psychology and neuroscience study it as a physical system. The soul — a non-physical, immaterial entity that some believe to be the seat of free will — doesn't appear anywhere in neuroscience. Everything mental can be explained in terms of physical brain processes.

If free will requires an immaterial soul that is separate from physical processes and can cause actions from \"outside\" the causal chain of nature, then materialism destroys free will.

Why the chapter dismisses this: The philosophical premise is weak. Most contemporary philosophers don't think free will requires an immaterial soul. You can believe that the mind is purely physical and still believe in free will. These are compatible views. Also, materialism is a commitment that typically comes from physics, not from specific psychological or neuroscientific studies. No experiment has ever \"proven\" that souls don't exist.

Argument 2: From Determinism

Brain processes follow physical laws. Neuroscience reveals that actions \"directly result from a set of law-like events in the person's brain\" (Haggard). If the action was entirely caused by prior brain states and laws of nature, the person couldn't have acted otherwise, and therefore isn't free.

Why the chapter dismisses this: Again, the philosophical premise is contested. Many \"compatibilist\" philosophers argue that acting according to causal laws is perfectly consistent with free will — what matters is whether the causal process runs through your own desires, reasons, and deliberations, not whether it could have been otherwise in some absolute physical sense. Also, determinism being \"true according to neuroscience\" is questionable — no neuroscience experiment has proven determinism.

Argument 3: From Epiphenomenalism — The Main Focus

This is the argument the chapter takes seriously, because its psychological premise is more directly supported by actual experiments.

Epiphenomenalism is the view that your conscious mental states — your conscious decisions, intentions, thoughts, deliberations — are causally irrelevant to your behavior. They are side effects, not causes. Your behavior is actually determined by unconscious brain processes. Your conscious experience of deciding is like the smoke from a fire: it occurs when the fire burns, but the smoke doesn't cause anything.

The philosophical premise here is hard to deny: if your conscious decisions play no causal role in your behavior, and everything you do is controlled by unconscious processes you have no access to, then it's hard to see how you could be a free and responsible agent in any meaningful sense.

The chapter focuses on the psychological premise — is epiphenomenalism actually supported by the evidence?

Two Types of Epiphenomenalism

Metaphysical epiphenomenalism: Based on the philosophical principle that because every action has a complete physical/neural cause, there is no additional causal role left for consciousness to play. The neural correlates of your conscious decision do all the causing; the conscious experience itself is causally excluded. This is a philosophical argument from the \"causal exclusion principle,\" not from experiments.

Scientific epiphenomenalism: Based on empirical findings that show consciousness doesn't seem to be doing any causal work. This is what the chapter actually examines.

For scientific epiphenomenalism to undermine free will completely, it needs to satisfy two conditions:

The Causal Irrelevance Condition: The evidence must show that conscious mental states play NO significant causal role in behavior — not just a reduced role, but essentially no role.

The Generalizability Condition: The evidence must apply to human behavior generally, not just to some narrow, unusual, or trivial class of behaviors.

Part 3: Three Lines of Evidence Examined

Evidence Line 1: Libet-Style Studies

The Original Libet Experiment

Benjamin Libet designed a now-famous experiment in the 1980s. The setup was simple:

Participants were asked to flick their wrist whenever they felt like it. A clock with a fast-moving dot was displayed in front of them. They had to note the clock position at the moment they first became aware of the urge or intention to move their wrist.

Meanwhile, electrodes on their scalp measured brain activity — specifically something called the readiness potential (RP), a pattern of electrical activity in the motor cortex that ramps up before voluntary movements.

The result:

The readiness potential began about 550 milliseconds before the actual wrist movement
The participants reported becoming consciously aware of their intention to move about 200 milliseconds before the movement

This means the brain started preparing the movement about 350 milliseconds before the person was consciously aware of any intention to move.

The brain \"decided\" before the person did — or at least before they were consciously aware of any decision.

Libet's interpretation: the brain initiates voluntary acts unconsciously. Conscious intention arrives too late to be the initiating cause.

More Recent Studies (Soon et al. 2008)

Researchers using fMRI (which gives more detailed spatial information about brain activity, though less precise timing) pushed this even further:

They found brain activity patterns that could predict which of two buttons a person would press up to 10 seconds before the person consciously decided
Not only did brain activity precede conscious intention (like Libet), but it actually predicted the specific content of the decision

This seems even more dramatic — your brain \"knows\" what you're going to do 10 seconds before you do.

Does This Prove Epiphenomenalism?

Libet's own interpretation — the veto:\nInterestingly, Libet himself did NOT think his results proved that consciousness is causally irrelevant. He proposed a different picture: unconscious brain processes generate \"bubbling up\" urges and initiatives for action, but conscious will can veto these. The role of free will isn't to initiate action, but to act as a gatekeeper — allowing some impulses to proceed and blocking others.

This is analogous to the default-intervention model from earlier chapters: unconscious processes set the default, but conscious processes can intervene.

This view preserves a significant role for consciousness — not as the initiator of action, but as the controller of whether actions proceed.

Most neuroscientists dismiss Libet's veto idea, but it does represent a coherent alternative to full epiphenomenalism.

Problem with the Causal Irrelevance Condition:

Even setting aside the veto idea, the Libet studies don't actually show that consciousness plays no causal role. They show that some brain activity precedes conscious awareness of intention. But this is entirely consistent with consciousness still being causally important — it just means that the causal chain leading to action begins with unconscious processes that then produce consciousness, which then contributes to the final action.

Consider an analogy: when you're driving a car, the engine starts running before you consciously experience the car moving. But that doesn't mean your conscious steering plays no causal role in where the car ends up.

Alfred Mele offers a useful reinterpretation: the readiness potential might represent an inclination or urge to act, not a decision. Inclinations influence but don't determine behavior. The conscious decision still does the work of converting the inclination into an actual committed action (or not). This is consistent with the data — the readiness potential predicts behavior, but only probabilistically, not with certainty.

Problem with the Generalizability Condition:

Libet's experiment involves one of the most trivial, meaningless actions imaginable: flicking your wrist at a random time with no consequences. There's no reason to prefer one moment over another. There are no stakes, no deliberation required, no competing considerations.

As philosopher Adina Roskies points out, this is a \"degenerate case of freedom of the will\" — the very case where free will is least relevant or interesting. Of course unconscious processes dominate when you're making a completely arbitrary, inconsequential, instantaneous choice.

The crucial question is: does this generalize to decisions that actually matter? When you choose which career to pursue, whether to end a relationship, how to vote, whether to blow the whistle on misconduct — you consciously deliberate, weigh options, consider values. It's very hard to believe that this deliberation does nothing.

Conclusion on Libet studies: They fail both the causal irrelevance condition and the generalizability condition.

Evidence Line 2: Situationist Studies

The Basic Idea

Situationist psychology argues that external situational factors have far more influence on behavior than people's internal character traits, conscious values, or deliberate choices. Our behavior is shaped much more by our circumstances than by who we are or what we consciously intend.

If this is right, then your behavior isn't really controlled by your conscious reasoning processes — it's controlled by the situation you happen to find yourself in.

The Good Samaritan Study

This is one of the most memorable and disturbing experiments in social psychology.

Darley and Batson (1973) recruited seminary students — people training to be ministers and clergy, people with deep religious commitments to helping others and doing good.

Setup:

Participants first filled out questionnaires about their religious beliefs
They were then told to go to another building to complete a task
The task they were heading to was either a talk about career options for seminary students (task-relevant, not about helping) OR a talk about the Parable of the Good Samaritan — the famous Biblical story about a man who stops to help a stranger in need while busy, important people pass by
Some participants were told they were running late and needed to hurry (high-hurry condition)
Others were told they were ahead of schedule (low-hurry condition)

On the way to the other building, each participant passed a man slumped in a doorway, head down, coughing and groaning — a person who clearly needed help.

Results:

People in the low-hurry condition helped significantly more than those in the high-hurry condition
Whether participants were thinking about the Parable of the Good Samaritan made no significant difference to whether they helped
Religious personality type made no significant difference either

The irony is striking: seminary students on their way to give a talk about the story of the Good Samaritan literally stepped over a suffering person because they were in a hurry.

What determined whether someone helped wasn't their values, their character, their moral beliefs, or even their current conscious thoughts about the moral importance of helping. It was purely whether they happened to be in a hurry.

Other Classic Situationist Studies

The Bystander Effect (Latané and Darley, 1968): People are much less likely to help in an emergency when other people are present. The more bystanders, the less likely anyone helps. The mere presence of others diffuses the sense of personal responsibility, and people look to others for cues about whether to act. This works entirely through unconscious social dynamics.

Milgram's Obedience Studies (1963): Perhaps the most disturbing studies in psychology. Participants were told to administer electric shocks to another person (actually a confederate, who only pretended to be in pain). An authority figure in a lab coat instructed them to continue even as the \"victim\" screamed and begged them to stop. A significant majority of participants administered what they believed were dangerous levels of shock, acting against their own stated moral beliefs because an authority figure told them to.

This suggests that even people who believe they would never harm an innocent person can be pushed into doing so by situational pressure — specifically, the presence of an authority figure.

The Stanford Prison Experiment (Zimbardo, 1973): Participants were randomly assigned to play either \"guards\" or \"prisoners\" in a simulated prison. Within days, those assigned as guards began psychologically abusing those assigned as prisoners, and the prisoners began showing signs of genuine psychological distress. The arbitrarily assigned role — a purely situational factor — transformed how people behaved toward each other.

Does This Prove Epiphenomenalism?

Problem with the Causal Irrelevance Condition:

The situationist studies show that situational factors influence behavior — but they don't show that conscious processes play NO role. Not everyone in the high-hurry condition failed to help — some people still stopped. Not everyone in the Milgram experiment administered the maximum shock — some refused. Not everyone in the Stanford Prison Experiment became an abusive guard.

Individual differences in conscious values and deliberation still explain some of the variation in behavior. The studies show that situation matters more than we assumed — not that conscious choice matters not at all.

Also, the fact that conscious thinking about the Good Samaritan parable didn't increase helping doesn't mean conscious thinking never matters. It might just mean that this particular form of thinking (passive reflection on a parable) doesn't translate into action under time pressure. A conscious, explicit commitment to stop and help — actively formed before entering the situation — might matter more.

Problem with the Generalizability Condition:

The situationist studies tend to involve situations where people have to make quick decisions under time pressure, often in ambiguous situations, with little time for deliberate reflection. The high-hurry seminary students had seconds to decide whether to stop.

This is very different from the decisions where free will seems most relevant — important, considered choices about how to live your life, what kind of person to be, what career to pursue, what values to act on. These are deliberate choices made with time for reflection.

Of course situational pressures dominate when you have no time to think. But can we generalize from \"hurried decisions in ambiguous situations are situationally determined\" to \"all human decisions are situationally determined\"? That seems like a very large and unjustified leap.

Conclusion on Situationist Studies: They also fail both conditions. They are important and humbling findings — our behavior is influenced by situation more than we like to think — but they don't establish full-blown epiphenomenalism.

Evidence Line 3: Wegner-Style Studies

The Basic Idea

Where Libet-style studies try to show that unconscious processes determine behavior before consciousness gets involved, and situationist studies try to show that external factors control behavior without consciousness, Wegner-style studies attack from a different angle.

Wegner's target is the experience of conscious will — the feeling that you are consciously willing your action, that your conscious intention caused what just happened.

Wegner argues that this experience is an illusion. Just because you feel like you consciously chose to do something doesn't mean that your conscious choice actually caused it. The feeling of willing and the actual causing of action are two separate things, generated by two separate processes.

The I Spy Experiment

The setup: A participant sat at a computer screen filled with small images of various objects (from the children's book \"I Spy\"). The participant jointly moved a cursor around the screen with another person — but the \"other person\" was actually a secret confederate working with the researchers.

Every 30 seconds or so, they stopped the cursor on an object. After each stop, the participant rated how much they felt they had intentionally stopped the cursor at that location, on a scale from \"I allowed this to happen\" to \"I intended this.\"

The twist: Participants listened to words through headphones throughout the experiment, supposedly random distractors. But in the \"forced stop\" trials — where the confederate secretly controlled where the cursor stopped — a word corresponding to the object (e.g., \"swan\") would be played through the headphones just before the cursor was forced to stop on that object.

Results:

When the participant heard \"swan\" about 1-5 seconds before the cursor stopped on the swan (the period where the thought was fresh in their mind and consistent with the upcoming action), they rated themselves as having intentionally stopped on the swan — even though the confederate actually did it.

When the word was heard 30 seconds before the stop (too far back to feel relevant) or 1 second after (too late to have \"caused\" it), the illusion didn't occur.

This creates a completely illusory experience of conscious will: the participant feels like they chose to stop on the swan, but they didn't — the confederate made that choice. The feeling of willing was triggered by the thought-action pairing, not by any actual causal connection.

Wegner's interpretation: the experience of conscious will is created when a thought occurs just before an action, is consistent with the action, and has no obvious competing explanation. When these conditions are met, the mind creates a feeling of \"I did that\" — but this feeling doesn't track whether the conscious thought actually caused the action.

Does This Prove Epiphenomenalism?

Wegner himself distinguishes four types of cases:

Normal voluntary action: You do something and you feel you intended it (most everyday behavior)
Normal inaction: You don't do something and you don't feel you intended it
Automatism: You do something but feel you didn't intend it (automatic behaviors)
Illusion of control: You feel you intended something that you actually didn't cause (the I Spy experiment)

The I Spy experiment demonstrates case 4 — but Wegner himself admits that most everyday behavior falls into case 1. The experience of conscious will seems accurate for most normal voluntary actions.

Problem with the Generalizability Condition:

The I Spy experiment is deliberately engineered to create an illusion. It exploits a specific mechanism of how the mind attributes authorship to actions. Demonstrating that this mechanism can be fooled in a specially designed laboratory scenario does not show that it's fooled in general.

The argument would be like demonstrating that optical illusions exist and then concluding that all visual perception is illusory. The existence of cases where perception goes wrong doesn't mean perception goes wrong in general.

Problem with the Causal Irrelevance Condition:

Wegner's broader theory involves two claims:

The matching model: The experience of conscious will is generated by a process that checks whether a thought preceded and is consistent with an action (this is what the I Spy experiment supports)
The automaticity thesis: The actual causes of action are unconscious and automatic, not conscious

The problem is that the I Spy experiment supports the matching model but doesn't support the automaticity thesis. These are separate claims. Just because you can feel you did something you didn't do doesn't mean that when you actually do do something, your consciousness didn't cause it.

Wegner relies on other studies (situationist and automaticity studies) to support the automaticity thesis — which means the Wegner-style studies aren't actually providing an independent, new line of evidence for epiphenomenalism.

Conclusion on Wegner-Style Studies: They also fail both conditions.

Part 4: The Moderate Conclusion

After examining all three lines of evidence, the chapter concludes that none of them establishes epiphenomenalism. The strong claim — that consciousness plays no significant causal role in behavior — is not supported by the available evidence.

But the chapter doesn't conclude that free will is completely safe. Instead, it argues for moderate scepticism: we have less free will than we naively assume, in two specific ways.

Dimension 1: We Are Free in Fewer Cases Than We Think

Libet-style studies show that for simple, arbitrary, instantaneous motor actions (like randomly flicking your wrist), unconscious processes do the initiating. You're not freely choosing in those moments in the way you might have thought.

This is a real, interesting finding. Before Libet, you might have assumed that even the most trivial voluntary actions are consciously initiated. Now we know that's not quite right.

Dimension 2: We Are Free to a Lesser Degree Than We Think

Even in cases where we are free, our freedom is reduced by factors we're unaware of. Situationist studies show that external circumstances influence behavior more than we realize. We like to think our values and character determine our actions, but being in a hurry can override those values without our noticing.

This doesn't mean we have no freedom — many people in the high-hurry condition still helped — but it means our freedom is less total and less robust than we assumed.

The Right Question

The chapter proposes that because of these findings, the right question about free will is not the binary \"do we have free will or not?\" but rather: In which specific situations do we have freedom and responsibility, and to what degree?

This needs to be answered case by case, with specific psychological research taken into account.

Part 5: Case Study — Implicit Bias and Responsibility

To illustrate moderate scepticism in action, the chapter uses a detailed case study.

What Is Implicit Bias?

The chapter introduces \"Elsa\" — a person who sincerely believes she is fair-minded and treats everyone equally regardless of race. But in practice, she ignores her Nigerian secretary, avoids sitting next to Asian passengers on public transport, and would rate a white job candidate's CV more highly than an identical CV from a Black candidate.

These are implicit biases — systematic patterns of behavior that operate below conscious awareness, conflict with the person's self-image, and deviate from how they ought to behave.

Four key features of implicit biases:

\n
Implicit psychologically: They are not accessible through introspection. Elsa cannot look inward and find her bias the way she can look inward and find her conscious beliefs and values.
\n
\n
Implicit methodologically: They are studied using indirect tests that don't rely on self-report, because self-report is unreliable for this purpose.
\n
\n
Biases descriptively: The behavior deviates from what the person believes themselves to be doing. Elsa thinks she treats people equally, but she doesn't.
\n
\n
Biases normatively: The behavior deviates from how the person ought to behave. Elsa should treat her Nigerian secretary the same as her white colleagues.
\n

The Implicit Association Test (IAT)

The primary tool for measuring implicit bias is the IAT, developed by Mahzarin Banaji and Tony Greenwald.

How the race version works:

Stage 1: Words appear on screen. Press right for pleasant words (\"joy,\" \"pleasure\"), left for unpleasant words (\"pain,\" \"failure\").

Stage 2: Faces appear. Press right for white faces, left for Black faces.

Stage 3 (combined): Words and faces appear. Press right if it's a pleasant word OR a white face. Press left if it's an unpleasant word OR a Black face.

Stage 4 (switched): Same, but now press right for pleasant word OR Black face, and left for unpleasant word OR white face.

For many people, Stage 3 (good/white, bad/black — GWBB) is easier and faster than Stage 4 (good/black, bad/white — GBBW). People make more errors and are slower when they have to associate Black faces with good things.

This suggests that many people have implicit associations linking white people with positive concepts and Black people with negative concepts — associations they were unaware of and didn't consciously choose.

Greenwald's own description of taking the test for the first time (with insects and flowers rather than racial groups) captures the experience: he couldn't improve his performance in the harder combination no matter how hard he tried, even repeating the task many times.

Limitations of the IAT:

It's not the only measure of implicit processes, and different measures sometimes give conflicting results
The IAT measures implicit mental associations (like associating Black people with bad words), but this doesn't automatically translate into biased behavior in real life
IAT scores correlate with some micro-behaviors (eye contact patterns, blinking) but the evidence for whether race IAT scores predict significant discriminatory behavior in real life is controversial

Are People Responsible for Their Implicit Biases?

This is where the chapter applies the free will framework directly to a morally important real-world question.

The intuitive answer might be: No, Elsa is not responsible. She didn't choose her bias, she wasn't aware of it, she can't directly control it. How can you be responsible for something you didn't know you had and can't directly stop?

Jennifer Saul articulates this view: people shouldn't be blamed for biases they're completely unaware of, and even once they become aware they're likely to have biases, they can't instantly control them.

The chapter identifies two main arguments for non-responsibility:

Argument from Awareness: You can't be responsible for something you're not aware of.

Argument from Control: You can't be responsible for something you don't have control over.

Each argument can be interpreted in a strong or weak version, depending on whether \"awareness\" and \"control\" mean direct (immediate, introspective) or indirect (via scientific knowledge and external interventions).

This generates four distinct arguments:

Argument from Direct Awareness: People lack direct/introspective awareness of their biases → therefore not responsible
Argument from Direct and Indirect Awareness: People lack both direct AND indirect/scientific awareness → therefore not responsible
Argument from Direct Control: People lack direct/immediate control → therefore not responsible
Argument from Direct and Indirect Control: People lack both direct AND indirect/scientific control → therefore not responsible

Evaluating Each Argument

Argument from Direct and Indirect Control — fails on the psychological premise:

Yes, people lack direct control over implicit biases. Elsa can't simply decide in the moment to be unbiased the way she can decide to raise her hand.

But do people lack indirect control? Probably not. Research shows various interventions can reduce implicit biases:

Looking at counter-stereotypical images before a decision (e.g., viewing photos of admired Black figures like Oprah Winfrey or Nelson Mandela can temporarily weaken implicit racial associations)
Using implementation intentions — specific if-then plans like \"If my Nigerian secretary is in a meeting, I will pay close attention to what she says\" — which have been shown to reduce the influence of automatic biases on behavior
Various training programs, mindfulness practices, and debiasing techniques

If indirect control exists, then the argument fails at its very first step.

Argument from Direct Control — fails on the philosophical premise:

Even if Elsa can't directly control her bias in the moment, this doesn't settle the responsibility question. The question is whether she has any kind of control — direct or indirect.

Imagine Elsa knows about the interventions. She could look at counter-stereotypical images before evaluating candidates. She could use implementation intentions. She could take the IAT regularly to check her bias levels. But she chooses not to do any of this.

Is she really \"not responsible\" just because her control is indirect rather than direct? This seems counterintuitive. The degree of directness of control doesn't seem like it should be the thing that matters — what matters is whether control is available.

Argument from Direct Awareness — fails on the philosophical premise:

Suppose Elsa doesn't have direct introspective access to her biases (she can't look inward and find them). But suppose she does have indirect scientific awareness — she has read about implicit bias, perhaps taken the IAT, knows the research showing that most people have racial biases.

Should she be off the hook because she lacks direct introspective awareness? Again, this seems wrong. She knows in general that she probably has implicit racial biases. This indirect, scientific awareness seems like enough to ground some responsibility, even without direct introspective access.

Argument from Direct and Indirect Awareness — the most serious challenge:

What about the case where Elsa has neither direct nor indirect awareness? She's never heard of implicit bias, she doesn't know the research exists, she has no way of knowing she's biased.

If awareness of any kind is a prerequisite for responsibility, and Elsa genuinely has none, then she isn't responsible.

But the psychological premise is already questionable: implicit bias research is now widely discussed in media, in workplaces, in diversity training. Most educated people in wealthy countries have some indirect scientific awareness that implicit biases exist and likely affect them.

And the philosophical premise is also questionable: what if Elsa ought to have known? Her company circulates materials about racial bias. They run mandatory training seminars. She ignores all of it — skips the emails, sits in the seminar on her laptop pretending to listen.

In this case, Elsa lacks awareness through her own culpable neglect. This is similar to the legal concept of \"willful ignorance\" — you can be responsible for something you didn't know about if you chose not to find out when you had both the opportunity and the obligation to do so.

Overall conclusion on implicit bias: None of the arguments for non-responsibility fully succeed. People may have significant responsibility for their implicit biases — not because they consciously chose those biases, but because they have (or could develop) indirect awareness of them and indirect control over them. Responsibility doesn't require that you directly feel or directly stop a bias in the moment. It requires that you take reasonable steps to understand and manage your biases when this is possible and when the stakes (e.g., fair hiring, police use of force) are high enough to demand it.

Overall Summary of the Chapter

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n

Claim	Chapter's Verdict
Free will is impossible because the brain is physical (materialism)	Not proven — philosophically controversial, not supported by specific studies
Free will is impossible because everything is determined (determinism)	Not proven — same problems
Free will is an illusion because consciousness is causally irrelevant (epiphenomenalism)	Not proven by current evidence
Libet studies show consciousness doesn't cause action	Overinterpreted — fails both conditions
Situationist studies show behavior is externally controlled	Overinterpreted — fails both conditions
Wegner studies show the feeling of willing is illusory	Interesting but limited — fails both conditions
We have less free will than we naively assumed	Supported — moderate scepticism is warranted
People are never responsible for implicit biases	False — indirect awareness and control ground responsibility

The Final Position

Science doesn't prove free will is an illusion. But it teaches us important things:

1. We are free in fewer situations than we assumed. Simple, automatic, trivial actions may not involve genuine conscious choice.

2. We are free to a lesser degree than we assumed. Even in cases where we have freedom, external situational factors influence us in ways we don't notice, reducing but not eliminating our responsibility.

3. The right question is not \"do we have free will?\" but \"when and to what degree?\" This needs to be answered case by case, using both psychological evidence and philosophical analysis together.

4. Implicit biases are a fascinating borderline case. We don't choose them consciously, we can't directly control them — but we may still be responsible for them through indirect awareness and indirect control, especially when the consequences of acting on those biases are serious.