Battle of Ideas

Humans have free will

AI-generated · paired steelman agents · independently red-teamed · Pass-1 source spot-checks only · framing-fidelity not independently verified · single model family

Whether humans have the kind of control over their choices that grounds moral responsibility. Libertarian free will and compatibilism argue FOR under different definitions — definitional clarity is part of the contest — versus hard determinism and hard incompatibilism AGAINST.

AGAINST 7

no further strong arguments at this depth

FOR 8

no further strong arguments at this depth

Ordering within each column: strongest first — validation tier, then source quality, then representativeness.

AGAINST · Humans have free will
Empirical — moderateP1

Neural precursors precede the conscious feeling of deciding

Libet's experiments found a rising 'readiness potential' in motor cortex beginning roughly 350–550 ms before subjects reported the conscious urge to move. Soon and colleagues (2008), using fMRI, decoded which of two buttons a subject would press from prefrontal and parietal activity up to 7–10 seconds before the person was aware of deciding. The prima facie implication: the brain commits before the conscious self experiences 'deciding,' so the felt moment of willing is a downstream read-out, not the cause. If the neural antecedent precedes and predicts the choice, conscious will is not doing the initiating work the free-will picture assumes. The evidence is genuinely contested, and honesty about that strengthens rather than weakens the case: Schurger and colleagues (2012) reinterpreted the readiness potential as accumulating neural noise crossing a threshold rather than a decision signal, and the Soon decoding accuracies were modest (around 60%, above chance but far from determinative). So this proves no strong thesis. Its weight is evidential and defeasible: it removes the intuitive support free will draws from the feeling that conscious deliberation originates action, and shifts the burden onto libertarians to demonstrate that the felt author is the actual cause rather than a narrator arriving after the fact.

Key assumptions

  • The pre-conscious neural activity is causally upstream of the choice, not a mere correlate or general preparation. partial
  • Findings from arbitrary button-presses generalize to real, deliberated choices. testable
  • The free-will thesis stakes its claim on the conscious experience of deciding. partial

Red team — the strongest counters

Schurger removes the phenomenon, not just weakens it

The argument concedes Schurger's noise-accumulator reinterpretation but understates it. If the readiness potential is stochastic activity drifting to a threshold, and movement occurs when the threshold is crossed, there is no early neural 'decision' that conscious will lags behind — the RP is an artifact of averaging aligned to movement onset, not a commitment preceding awareness. This doesn't merely lower the argument's weight; it dissolves the temporal gap the whole inference needs. Later work (Schurger 2021, Travers and colleagues) reinforced the accumulator reading. The flagship Libet finding may be measuring when noise happens to cross a line, not when the brain 'decides.'

Arbitrary flexes don't model deliberated choice

Libet and Soon study the worst possible proxy for free will: spontaneous, reasonless button-presses with instructions to act on a whim. The free-will-relevant domain is reasoned, temporally extended deliberation — choosing a career, a spouse, a moral course — where there is no evidence of 7–10-second pre-decodability and where Soon's ~60% (barely above chance, consistent with a weak prior rather than a settled outcome) offers no traction. Generalizing from urge-timing in a whim task to the claim that all choice is neurally pre-settled is an enormous, unlicensed extrapolation. The paradigm was built to isolate exactly the choices least like those the debate concerns.

Unconscious processing is still the agent's

The argument targets only free will that 'stakes its claim on the conscious experience of deciding' — but that is a narrow target most defenders don't occupy. Compatibilists and many libertarians never required the felt conscious urge to be the first mover; the unconscious neural processes ARE the agent's own reasoning, not an alien force. Mele's point: showing the reportable 'urge' isn't the initiating cause relocates agency below the threshold of report — it doesn't show the decision wasn't the agent's. Equating 'conscious will isn't first' with 'the person didn't control it' assumes agency must be consciously transparent, which no serious account requires.

Sources

Confidence, decomposed

Logical validity●●●○○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●●

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · Humans have free will
Empirical — moderateP1

The feeling of authorship is a manufactured, post-hoc narrative

A large experimental literature shows the sense of conscious authorship is an unreliable, post-hoc construction. Nisbett and Wilson (1977) documented people confidently reporting reasons for choices that demonstrably did not drive them, while denying the factors that actually did (such as position effects). Johansson, Hall and colleagues' 'choice blindness' studies (Science, 2005) had subjects pick the more attractive of two faces, then covertly handed them the rejected face — most failed to notice the swap and went on to explain in detail why they had 'chosen' the face they had in fact rejected. Wegner's Illusion of Conscious Will (2002) assembles cases — facilitated communication, the Ouija effect, alien-hand syndrome — where the feeling of willing and the actual causation of action come apart in both directions. The convergent finding: the mind produces a compelling 'I decided this for these reasons' narrative that is often fabricated after the fact and dissociable from what actually generated the behavior. This alone proves no metaphysics, but it dismantles the introspective evidence that is free will's main everyday support. If the very feeling of authoring our choices is systematically produced whether or not we authored them, that feeling cannot serve as testimony that we possess responsibility-grounding control.

Key assumptions

  • The introspective sense of authorship is most people's primary evidence for believing in free will. partial
  • Confabulation findings generalize from lab tasks to ordinary deliberated action. testable
  • A feeling produced independently of the fact it represents provides no evidence for that fact. untestable

Red team — the strongest counters

Reason-report failure isn't authorship failure

Nisbett–Wilson and choice blindness show retrospective REASON-reports are unreliable and that people miss covert swaps in rapid low-stakes tasks. But that is a claim about introspective access to why, not about whether the choice was controlled. A person can genuinely author a choice and separately be poor at narrating its causes — the two dissociate. The argument slides from 'self-explanations confabulate' to 'the feeling of authoring is illusory,' conflating second-order reason-attribution with first-order agency. Confabulated justifications are fully compatible with the underlying choice being the agent's own controlled act. What the studies demolish is causal self-report, not authorship itself.

Dissociation cases are a biased sample

Facilitated communication, Ouija, alien-hand, and covert card swaps are engineered or pathological cases selected precisely because willing and doing come apart. That the authorship-feeling is defeasible under rigged intervention says little about its reliability in the vast base of ordinary action — I decide to lift my arm, and it lifts, for my reasons, billions of times. Wegner's over-generalization from anomalies to a global 'illusion' has been criticized by Nahmias and Mele. A faculty that misfires when experimenters actively deceive it can still track reality reliably under normal conditions, exactly as vision does despite the existence of visual illusions.

Decoupling under intervention isn't systematic falsity

The argument's premise — a feeling produced independently of the fact provides no evidence — does the metaphysical work and is questionable. Demonstrating that the authorship-feeling can be decoupled from authorship under manipulation is compatible with its being reliably PRODUCED BY genuine authorship in normal cases. Visual illusions do not show vision fails to track the world; they show it can be fooled. To license the strong conclusion, the argument needs the feeling produced independently of authorship generally, but the studies establish only that it can be so produced in contrived conditions. Fallibility is demonstrated; systematic falsity is merely asserted.

Sources

Confidence, decomposed

Logical validity●●●○○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · Humans have free will
Logically validP1

The Consequence Argument: unavoidability transfers through entailment

Compatibilists say freedom is just acting on your own desires without external constraint, so determinism poses no threat. Van Inwagen's Consequence Argument shows why that is too quick. If determinism is true, the state of the world in the remote past combined with the laws of nature entails every present action. But no one has power over the distant past, and no one has power over the laws of nature. If a proposition P is fixed and P entails Q, then Q is fixed too. So our present actions are as unalterable as the Big Bang. The compatibilist 'could have done otherwise' — meaning 'would have, had you wanted to' — is compatible with the fact that, given the past and the laws, you could not have wanted otherwise. The genuine access to more than one future that grounds moral responsibility is exactly what determinism forecloses. The argument's force is that it invokes no theory of the self or mind; it simply transfers unavoidability through logical entailment. To resist it, the compatibilist must deny that unavoidability transfers across entailment — which looks like denying a near-truism about the fixed past. That is a steep bill for a view whose main selling point was common sense.

Key assumptions

  • Determinism (or statistical near-determinism) holds at the neural scale relevant to action. partial
  • The transfer-of-unavoidability principle (Beta) across entailment is valid. untestable
  • Responsibility requires genuine alternative possibilities (leeway incompatibilism). untestable

Red team — the strongest counters

The transfer principle Beta is invalid

The argument's engine is the transfer-of-unavoidability principle (Beta): if no one has a choice about P, and P entails Q, then no one has a choice about Q. But McKay and Johnson (1996) demonstrated Beta is invalid — it fails agglomeration, and counterexamples built from probabilistic or conjunctive propositions break it. Van Inwagen himself later conceded the principle needs repair. Without a valid transfer principle, unavoidability does not provably flow across entailment, and the central inference — that fixed past plus laws makes present acts 'as unalterable as the Big Bang' — loses its warrant. The near-truism the argument invokes is not the formal principle it actually needs.

Lewis: local-miracle ability dissolves 'power over laws'

The argument claims the compatibilist must absurdly assert power over the laws of nature. Lewis ('Are We Free to Break the Laws?') distinguishes a weak from a strong ability: to have done otherwise, Alice needs only the weak ability such that, had she done otherwise, a law would have been broken — she need not herself be able to break a law. The premise 'no one has power over the laws' then equivocates. Under the weak reading, which is all responsibility requires, the compatibilist affirms could-have-done-otherwise without claiming any miracle-working power. The Consequence Argument trades on collapsing weak and strong ability into one.

Frankfurt cases sever responsibility from leeway

The argument assumes responsibility requires genuine alternative possibilities (leeway incompatibilism). But Frankfurt-style cases present agents who could not have done otherwise — a counterfactual intervener stood ready — yet who act on their own and seem fully responsible. If the Principle of Alternative Possibilities is false, then even granting that determinism forecloses leeway, the responsibility-defeating conclusion does not follow. Source-compatibilists like Fischer accept the Consequence Argument's leeway conclusion entirely and simply deny leeway grounds responsibility, relocating it in the actual causal history. The argument's third premise is thus its most vulnerable point, and it is the least defended one.

Sources

  • An Essay on Free Will Peter van Inwagen, 1983, Oxford University Press (Clarendon Press) — origin of the modern Consequence Argument. P1 checked

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · Humans have free will
Logically validP1

Pereboom's manipulation argument: no difference-maker from puppet to nature

Pereboom's four-case argument targets compatibilism on its own turf. Imagine Professor Plum, manipulated moment-to-moment by neuroscientists so that he murders — yet he satisfies every compatibilist condition: his action is reasons-responsive, flows from his character, and involves no coercion or phobia. Intuitively Plum is not morally responsible; he is a puppet. Now vary the case in small steps: manipulation only at the start of life; then rigid deterministic upbringing by an oppressive community; then ordinary determinism, where the 'programmer' is just the past plus natural law. At no step is there a principled, morally relevant difference that switches responsibility back on. Since Plum in Case 1 is not responsible, and the differences across the series (who did the determining, how recent it was, whether an intending agent was involved) look irrelevant to authorship, the ordinary determined agent in Case 4 is not responsible either. The argument's power is that it grants the compatibilist all their machinery — reasons-responsiveness, identification, desire-mesh — and exhibits that machinery in a case where responsibility is plainly absent. Compatibilism must then name a difference-maker between covert manipulation and determination-by-nature, and every candidate on offer looks morally beside the point. The machinery cannot be what confers responsibility, because it is fully present without it.

Key assumptions

  • The intuition that manipulated Plum is not responsible is correct and widely shared. partial
  • No morally relevant difference distinguishes the four cases (the no-difference premise). untestable
  • All compatibilist conditions are genuinely satisfied in the manipulation case. partial

Red team — the strongest counters

Hard-line reply: bite the bullet on Plum

McKenna's hard-line compatibilist grants the machinery and simply affirms that Plum in Case 1 IS responsible. The manipulation intuition is not data but a question-begging pump: if Plum genuinely satisfies every compatibilist condition, the sense that he's 'a puppet' may just be our unreliable reaction to the word 'manipulation.' The four-case series is symmetric — Pereboom runs it downward from non-responsibility in Case 1 to Case 4, but one can equally run it upward from ordinary responsibility in Case 4 to Case 1. Which direction to run the intuition is exactly what's contested, so the argument cannot legitimately presuppose the downward reading.

Soft-line: another agent's intentions is the difference

The soft-line compatibilist names the difference-maker Pereboom claims doesn't exist: in Cases 1–2 Plum realizes the intentions of distinct manipulating agents, making him their instrument, whereas in Case 4 no agent's designs are executed through him. History-sensitive compatibilism (Fischer–Ravizza's mechanism-ownership, Mele's contrast between manipulation and normal development) locates responsibility in whether the deliberative mechanism became the agent's own through a non-usurping process. 'Who did the determining' is not morally beside the point when one case routes the agent's will through another mind and the other doesn't. The argument asserts this difference is irrelevant rather than showing it.

The no-difference premise is a sorites

The claim that no single step introduces a morally relevant difference, therefore Cases 1 and 4 stand or fall together, has the structure of a sorites. Each grain of sand added leaves a heap; conclude there are no non-heaps. But moral properties can be vague and threshold-crossing without any single step being the discontinuity. That we cannot point to the exact transition does not show there is no transition — it shows the property is graded. Pereboom exploits our inability to locate a sharp line to infer identity of the endpoints, which is precisely the fallacy the heap paradox warns against.

Sources

  • Living Without Free Will Derk Pereboom, 2001, Cambridge University Press; refined in Free Will, Agency, and Meaning in Life, 2014, Oxford University Press. Both confirmed as origin/refinement of the four-case (Professor Plum) manipulation argument. P1 checked

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · Humans have free will
Logically validP1

Strawson's Basic Argument: you cannot be the cause of yourself

Galen Strawson's Basic Argument shows desert-grounding responsibility is impossible whether or not determinism is true. To be truly responsible for what you do, you must be responsible for the way you are — the mental nature from which the action springs. But to be responsible for the way you are, you must have brought it about that you are that way. You could only do that intentionally, in light of some prior principles, preferences and values — a prior nature M1. To be responsible for M1 you would have had to shape it in light of a still earlier nature M2, and so on. The regress is vicious: at some point you were simply handed an initial nature — by genes, environment, chance — that you did not and could not choose. Nothing done later can convert that unchosen starting point into genuine self-authorship. Ultimate responsibility would require being causa sui, the cause of oneself, which Nietzsche called 'the best self-contradiction that has been conceived so far.' Indeterminism offers no rescue: random undetermined events in your formation are no more yours than determined ones. The conclusion is unconditional and needs no physics — the desert-entailing responsibility free will is meant to secure cannot exist for any created, finite being.

Key assumptions

  • Desert-entailing responsibility requires being responsible for one's own character. untestable
  • Being causa sui (self-caused) is impossible for any finite being. untestable
  • No bootstrapping act of self-formation can halt the regress. partial

Red team — the strongest counters

Causa sui is a target nobody defends

Strawson sets the bar at 'ultimate' self-authorship — being causa sui — then shows it impossible. But this strawmans what defenders claim. Compatibilists, and many libertarians, never asserted you must author your own initial nature; they hold responsibility requires the right control over action given whatever nature you have. Dennett's point stands: 'the self-made man' is a fantasy no serious account needs. The argument validly proves that an admittedly impossible thing is impossible, then trades on the word 'true' to imply ordinary desert collapses with it. The load-bearing move is a definitional stipulation dressed as a metaphysical discovery about persons.

Responsibility can bootstrap from an unchosen start

The regress assumes each self-shaping act must itself be one you are already ultimately responsible for. But responsibility could bootstrap: an agent handed an unchosen initial nature still makes choices that are genuinely hers, and successive self-shaping accrues ownership incrementally without any first fully-responsible act. Kane's self-forming actions are designed precisely to seed responsibility from within a partly-unchosen history. The vicious regress is generated only by the demand for ultimacy at every step — drop that demand for a graded, relational notion of ownership and the regress never starts. The argument imports its own viciousness through its maximal premise.

It leaves attributability and answerability untouched

Even granting the argument entirely, it targets only 'basic desert' — the backward-looking, suffering-justifying kind. It says nothing against attributability (the act expresses your evaluative stance), answerability (you can be asked to justify), or forward-looking responsibility (reform, protection, reconciliation). Most operative uses of 'free will' and moral responsibility survive intact; Scanlon and Watson develop precisely these desert-independent notions. So the conclusion 'moral responsibility is impossible' overreaches: what's shown impossible is one metaphysically maximal variety deliberately defined to be unsatisfiable, while the responsibility that actually organizes moral life is left standing. The victory is real but narrow.

Sources

Confidence, decomposed

Logical validity●●●●●
Premise support●●○○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · Humans have free will
Logically validP1

The luck pincer: indeterminism buys randomness, not control

Libertarians grant that determinism is incompatible with free will and therefore posit that the crucial choice is undetermined. But indeterminism buys no control — this is the luck objection, and it makes the AGAINST case independent of whether determinism is even true. Consider van Inwagen's 'rollback': rewind the universe to the instant before Alice's undetermined decision and replay it a thousand times. If the choice is genuinely undetermined, then across replays — with identical past, character, and reasons — she sometimes chooses A and sometimes B. Which occurs on a given run is settled by nothing about Alice; it is chance. An outcome that varies randomly given everything about the agent is not an exercise of her control but the opposite — a fluke she can be neither praised nor blamed for. So the libertarian faces a dilemma: to the extent a choice is determined it is not 'up to' the agent in the ultimate sense demanded; to the extent it is undetermined it is a matter of luck. Since determinism and indeterminism exhaust the options for any event, neither delivers responsibility-grounding origination. This is why hard incompatibilism, not merely hard determinism, is the strongest AGAINST position — even an indeterministic universe leaves the self-origination free will requires incoherent.

Key assumptions

  • Control requires the choice be settled by the agent, not by chance. untestable
  • Determinism and indeterminism jointly exhaust the possibilities for any event. untestable
  • Agent-causation is either unintelligible or collapses into one of the two horns. partial

Red team — the strongest counters

Agent-causation is the denied third option

The pincer's force rests on determinism-or-chance exhausting the possibilities. Agent-causal libertarians (O'Connor, Clarke) deny exactly this: they posit causation by a substance — the agent — irreducible to event-causation and not random. On rollback, what settles which option occurs is the agent-qua-cause, not 'nothing about Alice.' Whether substance-causation is intelligible is the real debate, but the pincer smuggles in its conclusion by describing the undetermined divergence as chance from the outset. To assert the dichotomy is exhaustive is to assume agent-causation incoherent, which is the very point at issue. The rollback intuition presupposes what it purports to establish.

Kane: both branches are willed and owned

Kane locates indeterminism at torn decisions where the agent has competing motivations and would rationally endorse either outcome. Whichever way the effort of will resolves, the agent wills that option and makes it her own — both branches are plural voluntary actions she is responsible for. Luck over which-of-two-things-I-willed is not the responsibility-defeating luck of an unwilled outcome befalling me. The rollback pumps intuition on a caricature: a single reason, an arbitrary flip. Redescribe the decision as a genuine self-forming conflict and 'settled by chance' mischaracterizes what is in fact settled by the agent's own effort under indeterminacy.

The luck standard proves too much

Levy argues present luck undermines compatibilist accounts too — any account faces luck in the agent's constitution or the moment of choice. But if the objection is powerful enough to defeat libertarianism AND compatibilism AND every account on offer, the more reasonable inference is that its implicit standard for 'control' is impossibly strong, not that control is metaphysically impossible. A criterion no conceivable account can meet is better evidence that the criterion is defective than that its target doesn't exist. Pressed to full generality, the pincer becomes a reductio of its own conception of sourcehood rather than a discovery about the world.

Sources

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · Humans have free will
Logically validP1

Causal closure leaves no gap for a contra-causal will

Free will in the responsibility-grounding sense requires the agent to be an ultimate originator — something about the self must inject a choice not wholly fixed by prior causes. But the self is a physical system: a brain, made of the same matter and governed by the same laws as everything else. The causal closure of the physical — the well-supported principle that every physical event has a sufficient physical cause — leaves no gap for a non-physical will to reach in and tip an outcome without either violating conservation laws or being causally redundant. If mental states are physical states, they are law-governed like any others; if they were non-physical, they would have to override physics, for which there is no evidence. Quantum indeterminacy does not open the needed space: it introduces randomness, not authorship (see the luck argument), and largely averages out at neural scale. So there is no locus in the natural order where origination-not-fixed-by-prior-states could occur. The libertarian's 'unmoved mover within the agent' has nowhere to exist. The self is a link in the causal chain, not a first cause — a weather system, not a god. This naturalist backbone underwrites the more formal arguments: physicalism plus causal closure simply leaves no room for contra-causal freedom.

Key assumptions

  • The causal closure of the physical is true. testable
  • The mind is physical, or at minimum does not causally override physics. testable
  • Responsibility-grounding free will requires contra-causal origination, not merely reasons-responsiveness. untestable

Red team — the strongest counters

Exclusion cuts against all higher-level causation

Causal closure gets its bite from Kim's exclusion principle — if every physical event has a sufficient physical cause, non-physical causes are redundant. But that principle is double-edged: pressed consistently it makes ALL higher-level properties (biological, chemical, ordinary reasoning) epiphenomenal, since each has an underlying physical sufficient cause. Many philosophers take this as a reductio showing the exclusion principle too strong, not that mental causation is impossible. If the argument's engine also deletes the efficacy of digestion and natural selection, its selective application to the will is unmotivated. Closure is a contested extrapolation, not the settled law the argument treats it as.

It only refutes libertarianism, by assuming its definition

The argument's own third premise concedes it assumes responsibility requires contra-causal origination — but that is precisely the compatibilist's denial. Grant full causal closure and physicalism (compatibilism grants them cheerfully) and freedom is relocated into the KIND of causal process: reasons-responsive, uncoerced, the agent's own. Nothing in causal closure touches this. So the 'naturalist backbone' defeats only the libertarian, and only by defining free will as the contra-causal thing libertarians posit. Against the majority live position it is a non-sequitur: the absence of a gap in the causal chain is exactly what compatibilism always said freedom never needed.

Complex physical systems can be genuine originators

The argument equivocates between 'no gap in physical causation' and 'no origination.' Origination could be a feature OF the physical process — a self-organizing, self-modeling agent IS a genuine locus of control — rather than requiring a non-physical hand reaching into the chain. 'A weather system, not a god' is rhetorically loaded: weather doesn't represent goals, model outcomes, or revise itself in light of reasons, and a system that does those things can originate action in every sense responsibility requires without being a first cause. The empirical 'quantum averages out at neural scale' claim is also increasingly contested by work on neural criticality and channel stochasticity.

Sources

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Empirical — moderateP1

Neuroscience has not refuted free will (Libet reinterpreted)

The most-cited scientific case against free will is Libet's: a "readiness potential" builds in motor cortex ~300ms before subjects report the conscious urge to move, suggesting the brain decides first and consciousness merely narrates. Two lines dismantle this. First, Schurger, Sitt, and Dehaene (2012) reinterpreted the readiness potential: modeling spontaneous movement as a noisy accumulator crossing a threshold, they showed the pre-movement ramp is largely an artifact of averaging trials aligned to movement onset — it reflects random fluctuations that happened to reach threshold, not a covert decision made in advance. Their model also predicted the timing distribution of movements, so the "brain decides before you do" reading lost its mechanism. Second, Mele (2014) notes Libet's paradigm covers only arbitrary, urge-driven flicks with no reason to prefer one moment over another — the least free class of actions — and cannot be generalized to choices deliberated over minutes or years, where moral responsibility actually lives. Even granting the data, an unconscious precursor to a whim tells us nothing about whether your decision to change careers or keep a promise is reasons-guided. The flagship empirical refutation thus rests on a statistical artifact and an illicit generalization from twitches to considered agency.

Key assumptions

  • The Schurger accumulator model correctly explains the readiness potential (empirically supported but not uncontested). testable
  • Libet-style arbitrary flicks do not generalize to deliberated, reasons-weighed moral choices. partial
  • Conscious reasons-guidance over longer timescales is where moral responsibility is actually located. partial

Red team — the strongest counters

Schurger rebuts the RP but does not restore conscious authorship

Recasting the readiness potential as a noisy accumulator crossing a threshold defeats Libet's 'covert decision made in advance,' but it is fully compatible with — even congenial to — the deeper anti-free-will reading: the movement is triggered when unconscious stochastic dynamics happen to cross threshold, with consciousness reporting after. A stochastic threshold-crossing is if anything less reasons-responsive and less consciously authored than a covert decision would be. So the model rescues the timing data from one over-reading while leaving conscious will exactly as absent from the initiating cause. The FOR side needs positive evidence that conscious deliberation initiates action; Schurger removes a specific inference against that but supplies no evidence for it.

It attacks the oldest, weakest exhibit

Libet 1983 is no longer the flagship. Soon et al. (2008) decoded abstract left/right choices from fMRI up to ~7–10 seconds before reported awareness; Fried et al. (2011) predicted movement from single-neuron firing; Bode and others replicated pre-conscious decodability. Schurger's accumulator specifically targets the RP in arbitrary self-timed movement and does not address decoding of content-bearing choices seconds ahead. So even a total victory over Libet's RP leaves the broader converging case for unconscious antecedents standing. The argument rebuts the single easiest target and presents it as dismantling 'the most-cited scientific case,' when the empirical literature the skeptic actually relies on has moved well past it.

Mele's move buys immunity by conceding the evidence gap

Saying Libet-style flicks 'don't generalize' to deliberated choices protects free will only by conceding there is no direct evidence conscious will initiates even the simplest act — and offers no positive reason the deliberated cases are metaphysically different; both are neural processes. An instrument that cannot resolve deliberated choice yields no-signal, not a null: absence of a refutation is not evidence of freedom. So the argument is purely defensive — it neutralizes one experimental paradigm without contributing any positive empirical support for reasons-guided conscious agency, which is where it claims moral responsibility lives. It defends the citadel while admitting the citadel has never been shown to contain anything.

Sources

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Logically validP1

Free will is reasons-responsive control, not causal exemption

The incompatibilist assumes freedom requires escaping the causal order, notes we can't, and declares free will an illusion. That smuggles in a false equivalence: it treats "caused" as synonymous with "compelled." Compatibilism dissolves this. What matters for responsibility is not whether your action was caused, but by what. Fischer and Ravizza (1998) specify the mechanism: an action is free when it issues from a reasons-responsive process — one that would have produced a different action had the agent recognized sufficient reason to. A kleptomaniac's hand and a shopper's hand are both fully caused; only the shopper's mechanism tracks reasons (a security guard, a price change) and adjusts. That responsiveness is the control moral responsibility requires, and it is perfectly compatible with — indeed requires — orderly causation. Frankfurt (1971) adds the reflexive layer: a free agent acts on desires she endorses at a higher level. The addict who wants to want otherwise is unfree; the person acting on wholehearted desires is free. Determinism threatens none of this. Genuine randomness would in fact destroy it, making acts mere noise rather than yours. So the skeptic's picture inverts the truth: lawful causation is the friend of agency, not its enemy.

Key assumptions

  • Moral responsibility requires only reasons-responsive control, not metaphysical exemption from causation. partial
  • There is a real, non-arbitrary distinction between mechanisms that track reasons (shopper) and those that don't (kleptomaniac). testable
  • Indeterministic randomness would undermine rather than enhance agential control. partial

Red team — the strongest counters

Manipulation cases defeat the mechanism criterion

Pereboom's four-case argument targets exactly this account. Build an agent whose reasons-responsive, self-endorsed mechanism was installed by neuroscientists overnight: she deliberates, tracks the security guard and the price change, adjusts — she satisfies Fischer-Ravizza and Frankfurt to the letter, yet the intuition that she is responsible collapses. If a designed but fully reasons-responsive agent isn't responsible, and ordinary determinism differs from the manipulation only in lacking an intentional designer, compatibilism owes a principled reason why that difference grounds desert. The mechanism criterion certifies the manipulated agent and the ordinary agent identically, so 'control by a reasons-responsive mechanism' cannot be sufficient for the responsibility at issue.

It answers leeway, not sourcehood

The 'caused ≠ compelled' move is correct but wins only the wrong battle. Granting that the shopper's mechanism tracks reasons and the kleptomaniac's doesn't establishes that some caused actions differ from others — it does nothing about Galen Strawson's basic argument: the self-endorsed mechanism was itself fixed by heredity and environment the agent never chose, in a chain reaching back before her birth. To be ultimately responsible for how you act you'd have to be responsible for the mechanism you act from, and for the earlier self that shaped it, ad infinitum. Compatibilism dissolves the compulsion worry while leaving the origination worry — which is the incompatibilist's actual thesis — untouched.

The reasons-responsiveness threshold is arbitrary

The kleptomaniac/shopper pair is rhetorically clean because it's a boundary case. Real agency is a continuum: addiction, phobia, weakness of will, ideological capture, and ordinary temptation are all partially reasons-responsive — they would adjust to some reasons and not others. Fischer and Ravizza must draw a line at 'moderate' reasons-responsiveness, but every placement of that line is contestable, and wherever it falls it will excuse some agents we blame and blame some we excuse. Absent a non-arbitrary threshold, the account doesn't derive the responsibility distinction from the facts; it reads a pre-theoretic verdict back into the mechanism, which is question-begging on the contested cases that matter.

Sources

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Logically validP1

Frankfurt cases sever responsibility from could-have-done-otherwise

The strongest argument against free will runs: determinism means you could never have done otherwise; responsibility requires you could have done otherwise; therefore no responsibility. Frankfurt (1969) severed the second premise from the first with a thought experiment. Imagine Black wants Jones to shoot Smith and implants a device that will force the decision if Jones wavers — but Jones decides to shoot on his own, so the device never fires. Jones could not have done otherwise (Black stood ready to intervene), yet he is plainly responsible: he acted for his own reasons, unaided. The mere existence of a blocked alternative is morally inert. This detaches responsibility from the "could-have-done-otherwise" condition — the Principle of Alternative Possibilities — that the incompatibilist argument depends on. What grounds responsibility is the actual sequence: how the decision was in fact formed, not the branching structure of unrealized possibilities. Once PAP falls, determinism's inability to supply alternative possibilities stops being a threat. Even in a world with a single physically possible future, an agent can be the genuine, reasons-guided source of her act. The argument reframes the whole debate around actual-sequence sourcehood rather than counterfactual leeway — exactly the ground compatibilism occupies.

Key assumptions

  • The Frankfurt scenario is coherent: a fail-safe can be present without contributing to the actual decision (contested by the 'dilemma defense'). partial
  • Moral responsibility attaches to the actual causal sequence, not to unrealized alternative possibilities. untestable
  • The strong intuition that Jones is responsible is reliable evidence about responsibility. partial

Red team — the strongest counters

The dilemma defense (Widerker/Kane/Ginet)

Black needs a prior sign that Jones will waver. Either the sign is deterministically linked to Jones's decision or it isn't. If it is, the case simply assumes determinism at the crucial node — begging the very question against the libertarian, who denies the decision is thus fixed. If it isn't deterministically linked, then at the moment of choice Jones genuinely could still have gone the other way before any sign appeared, so Black cannot pre-empt in time and Jones retains an alternative — PAP is not violated. Frankfurt cases refute PAP only by covertly presupposing the determinism whose responsibility-compatibility is what's in dispute.

The flicker of freedom survives

Even inside the scenario Jones is not alternative-less. He retains a residual leeway: he could have begun to waver, shown the tell, done otherwise-in-his-initiating. PAP defenders (Naylor, early Fischer) argue this flicker is robust enough to anchor responsibility — the actual sequence in which Jones proceeds unaided is responsible precisely because a genuine alternative branch existed up to the initiating moment. The Frankfurt case eliminates the alternative of successfully-completing-otherwise but not the alternative of initiating-otherwise, and responsibility can attach to the latter. So the case doesn't cleanly sever responsibility from every could-have-done-otherwise condition; it narrows which alternative is required.

Refuting PAP leaves source-incompatibilism intact

Grant the strongest reading — alternatives are unnecessary, and responsibility rests on the actual sequence. That defeats leeway-incompatibilism but is silent on source-incompatibilism (Pereboom, G. Strawson), which is the live modern position. The question becomes whether the actual sequence in a fully determined world confers the right kind of sourcehood — whether an agent who is the proximate but not the ultimate origin of her act is responsible. Frankfurt cases contribute nothing here: relocating responsibility to 'how the decision was in fact formed' just renames the disputed territory, since the skeptic's claim is that a determined formation, alternatives or not, never makes the agent the desert-grounding source.

Sources

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Logically validP1

The Moorean burden: responsibility outranks the skeptic's premises

Compare two things you might believe: that you are sometimes responsible for what you do — that praising a rescuer and blaming a torturer track something real — versus the chain of abstract premises in an argument that no one is ever responsible for anything. Van Inwagen (1983) presses the methodological point: when a valid argument leads to a conclusion more incredible than the negation of one of its premises, the rational response is to reject a premise, not swallow the conclusion. Our confidence that a person who tortures a child for fun is blameworthy is about as high as confidence gets — higher than our confidence in contested theses about causal closure, the transfer of non-responsibility across causal chains, or the precise analysis of "could have done otherwise." So even someone who cannot yet locate the flaw in a skeptical argument has excellent reason to believe the flaw is there. This is a Moorean move: the datum outranks the theory. It does not by itself say which premise fails, but it shifts the burden decisively onto the skeptic, who must show his premises are individually more certain than the responsibility he asks us to abandon — a bar the metaphysics of hard incompatibilism has not cleared. Denying all responsibility to save a tidy argument reverses the proper direction of evidential weight.

Key assumptions

  • Our confidence in specific responsibility judgments genuinely exceeds our confidence in each premise of the skeptical argument. partial
  • The Moorean method — reject a premise rather than accept an incredible conclusion — is a valid inferential policy. untestable
  • Strength of a considered moral intuition is evidence about the moral facts. partial

Red team — the strongest counters

Debunking screens off the intuition's strength

The move treats the felt certainty that the gratuitous torturer is blameworthy as evidence about mind-independent moral facts. But that certainty is exactly what an evolutionary debunking account predicts whether or not basic desert exists: organisms that felt intense blame and desert stabilized cooperation and deterred defection, so selection installed the reaction regardless of its truth. When a belief's strength is fully explained by a process indifferent to its truth, that strength is screened off as evidence. The same Moorean confidence once underwrote absolute simultaneity and a motionless Earth. High credence secured by a truth-insensitive mechanism cannot outrank an argument; it is the datum most in need of scrutiny, not the one that trumps.

Wrong datum: horror ≠ basic-desert blameworthiness

Van Inwagen's certainty attaches to the wrong proposition. What is genuinely Moorean-certain is that torturing a child for fun is monstrous, that the agent must be stopped, condemned, deterred, incapacitated — and hard incompatibilists (Pereboom) retain every one of these through forward-looking, non-desert responsibility. The specifically basic-desert claim — that he deserves to suffer purely for having done it, independent of any consequence — is a theoretical gloss most people have never consciously separated out, not a raw datum. Once you disentangle the two, the certain part is fully compatible with skepticism about free will, and the burden-shift evaporates: the skeptic denies only the gloss, never the horror.

The Moorean policy doesn't adjudicate — it entrenches priors

Van Inwagen is himself a libertarian who runs the Consequence Argument, whose premises he holds more certain than compatibilism. The Moorean method is a symmetric meta-policy: it licenses each party to keep its most-obvious datum and reject the other side's least-certain premise. A hard determinist with the opposite confidence ordering — finding causal closure and no-ultimate-origination more secure than desert — is equally entitled to the move and reaches the opposite verdict. So the argument does not shift the burden onto the skeptic; it merely reports that its proponent finds responsibility more obvious than the skeptic's premises, which the skeptic already disputes. It restates the disagreement rather than resolving it.

Sources

  • An Essay on Free Will Peter van Inwagen, 1983, Oxford University Press (Clarendon). Confirmed: van Inwagen defends the Consequence Argument for incompatibilism and is himself a libertarian who holds moral responsibility indubitable; the Moorean/relative-certainty style of argument runs through this and his later work — verify the exact passage before quoting. P1 checked

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Logically validP1

Arguing for hard determinism pragmatically refutes it

Hard determinism faces a pragmatic self-refutation. To argue for it — to present evidence, expect an interlocutor to weigh it, and revise a belief accordingly — is to treat that interlocutor as a reasons-responsive agent who could accept or reject the conclusion on its merits. The determinist asking you to "follow the argument where it leads" presupposes exactly the rational self-governance the conclusion denies: that what you believe can be settled by reasons rather than fixed by antecedent causes indifferent to truth. If every belief is simply the output of prior physical states, then the determinist's own belief in determinism is caused, not justified, and the appeal to evidence is theatre. This is the ancient Epicurean point and Kant's distinction of "practical freedom": deliberation is not an optional stance but the unavoidable standpoint of any agent asking "what should I do?" or "what should I believe?" — you cannot deliberate while regarding your decision as already settled independent of the deliberation. The freedom at stake is thus transcendentally presupposed by the very activity of reasoning, including the reasoning that would deny it. The skeptic can state hard determinism, but cannot coherently argue for it without enacting the agency he rejects. The stronger his case, the louder his performance of the freedom he denies.

Key assumptions

  • Deliberation ineliminably presupposes open alternatives from the agent's own standpoint. partial
  • A belief that is merely caused cannot simultaneously be rationally justified (denied by reliabilists — the load-bearing contested premise). untestable
  • The standpoint of reasoning cannot be coherently abandoned while one is reasoning. partial

Red team — the strongest counters

Reliabilism dissolves 'caused, not justified'

The argument's load-bearing premise — a merely caused belief cannot be rationally justified — is denied by the dominant epistemology. On reliabilism, a belief is justified when produced by a truth-tracking process; determinism is fully compatible with reasoning being such a process, since evolution and learning can deterministically build brains whose belief-transitions track evidence. So a determinist's belief in determinism can be both caused and justified, and 'follow the argument where it leads' just means 'let the reliable evidence-responsive process run.' The dichotomy between caused and justified is exactly the assumption reliabilists reject; without it the self-refutation never gets started, and the argument itself flags this as its contested premise.

Equivocation between epistemic and metaphysical openness

Deliberation ineliminably presupposes only that the deliberator does not yet know her decision — epistemic openness. It does not presuppose the decision is metaphysically undetermined. A fully determined agent who cannot predict her own choice must still deliberate to reach it; the 'unavoidable standpoint' is satisfied by ignorance of the outcome, which determinism grants freely. So the transcendental move establishes the practical indispensability of deliberating, not the falsity of determinism. The argument trades on 'open alternatives' meaning metaphysical branching in the conclusion while only earning epistemic openness in the premise — a classic equivocation between the two senses of 'could.'

Performative inconsistency proves too little

Even granting the determinist cannot consistently assert her thesis, that is a fact about the pragmatics of assertion, not about the world. Truth is not hostage to assertability: 'I am not now speaking' and 'I am asleep' are unassertable-when-true yet can be true. A cosmos can be deterministic even if every agent within it must run a deliberative operating system to function and cannot coherently voice the denial of its own agency. So the argument, at most, shows hard determinism is awkward to advocate — an ad hominem on the speaker's position, not a refutation of the proposition. The stronger the performative reading, the more clearly it targets the arguer rather than the claim.

Sources

  • Groundwork of the Metaphysics of Morals / Critique of Practical Reason Immanuel Kant, 1785/1788 — source of the 'practical freedom' / deliberative-standpoint argument (Kant: rational agents must act 'under the idea of freedom'). Verified as an accurate general representation of Kant's practical-freedom/transcendental-freedom distinction; the specific modern transcendental-argument formulations are cited from memory and attribution to particular contemporary authors remains unverified — verify before quoting. P1 checked

Confidence, decomposed

Logical validity●●●○○
Premise support●●○○○
Representativeness●●●○○
Source quality●●●○○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Plausible, low testabilityP1

Responsibility is constitutive of the interpersonal stance

Strawson (1962) argued the whole debate is misframed. We do not first establish a metaphysical fact about free will and then decide whether to hold people responsible. Responsibility lives in the "reactive attitudes" — resentment, gratitude, indignation, forgiveness — woven into every relationship the moment we treat someone as a participant rather than an object. When a friend betrays you, resentment is not a theory you could revise on learning of determinism; it is constitutive of regarding him as a person who owed you good will. Strawson's key move: the demand that these attitudes be justified by some external, God's-eye guarantee of contra-causal freedom is incoherent, because the attitudes are the framework within which justification and excuse operate, not something standing in need of outside license. We already distinguish the excused (the child, the compelled, the deranged) from the responsible — from inside the practice. A blanket determinist "excuse" covering everyone would track no real distinction we recognize, and could not be sustained anyway without ceasing to have interpersonal relationships at all. Responsibility is thus as secure as personhood itself. The skeptic's complaint that it lacks metaphysical foundations mistakes the load-bearing floor for a missing basement.

Key assumptions

  • The reactive attitudes are constitutive of interpersonal life and cannot be globally and permanently suspended. partial
  • A social practice can be justified internally, without external metaphysical warrant. untestable
  • The excused-vs-responsible distinction we draw from inside the practice tracks something real. partial

Red team — the strongest counters

Inescapable does not entail justified

Strawson's strongest empirical claim is that the reactive attitudes are psychologically ineliminable. But the debate is normative: are resentment and desert-based blame warranted? That we cannot help feeling an attitude is no evidence it is appropriate — we cannot help many cognitive reflexes that are unjustified. Pereboom grants the attitudes' naturalness yet argues the specifically retributive ones can and should be tempered, replaced by grief, disappointment, and forward-looking concern that survive skepticism. Strawson's argument slides from 'constitutive of our practice' to 'needs no external justification' to 'is thereby justified' — but the incompatibilist never demanded a God's-eye license; he offered a first-order reason (no ultimate sourcehood) to revise the practice from within.

The blanket excuse is exactly what's argued for

Strawson concedes we excuse the child, the compelled, and the deranged from inside the practice, then asserts a universal determinist excuse 'tracks no real distinction.' But that is the conclusion, not an argument. The hard incompatibilist's claim is precisely that determinism is the global analogue of the local excusers — that the very feature which excuses the compelled (their conduct traces to factors they didn't originate) is present in every agent. To say this generalization tracks nothing real simply denies the skeptic's thesis by restating our habit of not applying it. Strawson needs to show the local excuses don't generalize, and he doesn't; he assumes it.

The dichotomy is false — attitudes are selectively modifiable

Strawson's rhetorical force depends on the choice being all-or-nothing: keep the full reactive attitudes or 'cease to have interpersonal relationships at all.' But cross-cultural variation in blame intensity, restorative-justice practice, therapeutic reframing of a wrongdoer's history, and everyday empathy toward those we understand ('to understand all is to forgive much') show the objective/participant stances mix continuously. People routinely soften retributive resentment on learning of an abuser's own abusive childhood without exiting the relationship. That tempered middle is precisely the hard-incompatibilist recommendation — so the practice's revisability, which Strawson denies, is empirically ordinary, and his defensive dilemma dissolves.

Sources

Confidence, decomposed

Logical validity●●●○○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Plausible, low testabilityP1

Libertarian free will locates in self-forming actions

Libertarians are often accused of needing a mysterious exemption from physics. Kane (1996) shows the resource is already in physics: indeterminacy. His model locates free will not in every act but in rare "self-forming actions" — moments of genuine inner conflict where an agent is torn between, say, ambition and moral duty. At such moments competing neural processes are indeterminate; the outcome is not fixed by the prior state. Crucially, whichever way it resolves, the agent wanted that outcome and made an effort of will toward it, so the result is willed, not merely random. Kane answers the standard "luck objection" (if undetermined, it's just chance, not control) with plural voluntary control: because the agent would rationally endorse either outcome as her own doing, she controls the resolution in the way that matters — much as a partner controls a deal struck in her name whether the agent closes it Tuesday or Wednesday. These self-forming choices then shape the settled character from which later, determined-but-free actions flow. The view is empirically constrained — it needs amplifiable neural indeterminacy — and thus genuinely falsifiable. But it demonstrates that libertarian free will is a coherent, physics-compatible hypothesis, not a ghost in the machine. The demand that indeterminism be "in the right place" is met, not dodged.

Key assumptions

  • The brain contains genuine indeterminacy amplifiable to the level of choice at torn-decision moments. testable
  • 'Plural voluntary control' genuinely defuses the luck objection rather than relabeling it. untestable
  • An agent willing either possible outcome makes the result hers rather than chance's. partial

Red team — the strongest counters

Plural voluntary control renames the luck, doesn't remove it

At the self-forming moment, by hypothesis the total prior state — including every reason, effort, and character trait — leaves the outcome open. So run the moment twice from identical conditions and it can go either way, with nothing about the agent explaining the divergence. 'She'd endorse whichever result' establishes endorsement-of-both, not control-over-which; the specific selection is still a chance event she doesn't determine. Kane's deal-broker analogy fails at the load-bearing point: the partner delegated authority in advance by a settled (indeed deterministic) arrangement, so either closing counts as hers. There is no analogue for the undetermined which — nothing plays the role of the prior delegation that would make the chance-selected branch authored rather than merely accepted.

The neural indeterminacy premise is empirically unsupported

The model requires genuine quantum indeterminacy amplified to decision-relevant macroscopic scales in neurons. But decoherence in warm, wet, ~310K neural tissue is estimated at femtosecond–picosecond timescales, and the mainstream computational-neuroscience picture treats neurons as effectively classical systems whose stochasticity is thermal noise — chance of exactly the kind the luck objection indicts. Kane needs indeterminacy that is (a) real, (b) amplifiable to the whole-choice level, and (c) not mere thermal randomness; no evidence establishes any of the three. The view is 'empirically constrained,' as claimed — but current evidence constrains it toward falsity, so its falsifiability is a liability here, not a virtue.

Ultimate responsibility rests on a handful of coin-flips

Kane concedes most acts are determined-but-free, flowing from settled character, and locates real origination only in rare torn moments. But that character is the product of a vast determined history the agent never originated, plus a small number of undetermined self-forming choices whose outcomes were, at the crucial instant, matters of chance. So the entire weight of desert-grounding responsibility hangs on a thin scatter of luck-events embedded in an otherwise unchosen causal stream. Even granting the model's coherence, it delivers a shockingly slender basis for the robust moral responsibility the FOR side wants — arguably too slender to bear blame and punishment, which is the practical stake of the whole dispute.

Sources

Confidence, decomposed

Logical validity●●●○○
Premise support●●○○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · Humans have free will
Unfalsifiable / philosophicalP1

Agent causation supplies the buck-stopping originator

Event-causal pictures assume every cause is itself an event, part of an unbroken chain reaching back before your birth. Chisholm (1964) and O'Connor (2000) argue this ontology is not mandatory. Agent causation posits that a person — a persisting substance, not merely a sequence of events — can be an originating cause: the agent brings about a choice without that bringing-about being itself caused by a prior event. When you raise your arm, you cause the neural event, and nothing causes you to cause it in the sense that would make you a mere conduit. The relation "substance S causes event E" is taken as primitive — on a par with, and no more mysterious than, the event-causation the determinist treats as unproblematic. O'Connor develops this with an irreducible "agent-causal power" that reductive accounts leave out. The position is metaphysically ambitious, but internally coherent, and it answers the deepest worry about libertarianism: it explains how an undetermined choice can still be the agent's doing rather than mere noise, because the agent herself is the terminus of explanation. It relocates the buck-stopping originator the concept of responsibility seems to demand — the point at which the search for a prior cause rightly ends.

Key assumptions

  • Substance (agent) causation is a coherent primitive, not covertly reducible to event causation. untestable
  • Agent causation is no more metaphysically mysterious than the event causation determinists already accept. untestable
  • Persons are persisting substances, not mere bundles of events. partial

Red team — the strongest counters

'No more mysterious than event causation' smuggles away the deficit

Event causation earns its keep: it connects states by laws that support counterfactuals, prediction, and explanation. Agent causation posits a substance producing an event with no law, no prior sufficient condition, and — crucially — no answer to the timing/selection problem: what makes the agent-cause produce E now rather than E' now or nothing at all? If the answer is 'nothing,' we are back to luck; if the answer is 'the agent's reasons or state,' that is a prior condition and agent-causation collapses toward the event-causal picture it meant to escape. The parity claim is therefore false: event causation lacks precisely the explanatory gap that agent causation is defined by.

An unexplained explainer bought only to save a conclusion

Agent-causal power is an irreducible causal relation found nowhere else in nature's inventory, invisible to a physics that describes only event-event relations and conservation laws, and posited for the sole purpose of rescuing a pre-theoretic intuition about origination. That is a textbook violation of parsimony: multiplying a novel ontological category to reach a desired verdict. It also revives the interaction problem — how does a non-event substance-cause inject an otherwise-uncaused neural event without breaching physical closure or energy conservation? 'Internally coherent' is a low bar; Cartesian dualism is internally coherent too. Coherence plus zero independent evidence plus a conservation-law tension is a weak footing for the metaphysics of responsibility.

It relabels the control problem rather than solving it

The advertised payoff is explaining how an undetermined choice is 'the agent's doing rather than mere noise.' But naming the agent the buck-stopping terminus does not show how her being the terminus makes this specific choice intelligibly hers instead of arbitrary. If, holding everything about the agent fixed, she could have agent-caused either option, then which one occurs is not explained by anything about her — the mystery the model claimed to dissolve is simply reasserted as a primitive ('substance S causes E'). Taking the desired feature (authored, not random) as an unanalyzable primitive is not an account of control; it is a promissory note that the primitive delivers what the objection says it cannot.

Sources

Confidence, decomposed

Logical validity●●●○○
Premise support●●○○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.