Hume, Stoicism, and Whether Reason Can Motivate Moral Action

Bottom line

David Hume does not argue that reason has no connection to morality. His narrower claim is that reason alone cannot supply the motivational impulse to act. Reason discovers facts, causal relations, and effective means; moral sentiment supplies approval, aversion, concern, and motivational force.

The Stoic view is also subtler than “first learn what is good, then mechanically act on the information.” For the Stoics, practical knowledge is a stable organization of judgment, assent, desire, and impulse. Genuine knowledge is already embodied in action-guiding character.

The best modern scientific picture supports neither a pure Humean separation nor a naïve Stoic intellectualism. Human action emerges from recurrent interaction among representation, affective valuation, learned goals, bodily state, cognitive control, habit, and social context. Bare propositional knowledge is usually insufficient, but reasoning can reshape what a person values, attends to, anticipates, and is able to do.

Three claims that must not be conflated

1. Motivational claim: reason by itself cannot move the will.
2. Metaethical claim: moral distinctions are not conclusions of reason alone.
3. Is–ought claim: a normative conclusion cannot be introduced without explaining the normative bridge from descriptive premises.

Hume discusses all three, but they are not the same argument. See Hume Is-Ought Problem for the third.

Evidence boundary — updated 2026-08-10

The philosophical sections distinguish primary texts from interpretations. The scientific section reports evidence about how human motivation and moral judgment work, not about whether objective moral truths exist. Neuroscience cannot by itself decide moral realism, prove that sentiments justify moral claims, or derive an ought from a brain scan. Terms such as “reason,” “passion,” and “knowledge” do not map neatly onto single psychological variables or brain regions.

Three companion notes — added 2026-08-22

This note runs the comparison through motivation and the science. Three shorter notes carry the parts that sit outside that frame:

• Can a Feeling Be Wrong - Stoics vs Hume and Hutcheson — the metaethical divide (a Euthyphro question, not reason-vs-feeling), the agent-vs-spectator standpoint split, Hutcheson as the hard case, and what Seneca and Epictetus say to “morality depends on affections”.
• Hume's Corrections of Sentiment - The General Point of View — Hume’s own apparatus for correcting a sentiment, sourced to Treatise 3.3.1, and why it is neither a set of recommendations nor a technique for managing your own emotions.
• Graver Audit - Correcting the Stoic Side of the Hume Comparison — a working draft checked against Graver. Relevant here: Graver’s case that the divide is truth-aptness, not motivation, and that on her reading the Stoics do not sit on the far side of Hume’s fork — they deny the fork. Compare §5–§6 below.

---

1. Hume’s argument in A Treatise of Human Nature

The relevant argument is distributed across two places:

• Book II, Part III, §3: “Of the influencing motives of the will.”
• Book III, Part I, §1: “Moral distinctions not derived from reason.”

Book III relies on the motivational psychology developed in Book II. Reading Book III alone can make the conclusion sound more sweeping than Hume intends.

1.1 Book II: why reason alone does not move the will

Hume divides reasoning into two broad functions:

1. Demonstrative reasoning about relations of ideas—logic, mathematics, and conceptual relations.
2. Probable or causal reasoning about matters of fact—what exists, what will happen, and what means produce which effects.

Neither function supplies an ultimate end by itself.

• Arithmetic can tell a merchant what is owed, but does not by itself make the merchant care about paying.
• Causal knowledge can tell me which medicine will cure an illness, but does not by itself make health matter to me.
• Logic can reveal that two commitments conflict, but does not determine which commitment I should care enough to revise.

Reason changes action when it changes a belief relevant to an existing concern. If I want to eat a fruit but learn that it is poisonous, my behavior changes. Hume’s explanation is not that reason generated an entirely new motive from nothing. The new belief redirected aversion, self-concern, or the desire to avoid pain.¹

This yields the classic Humean structure:

Motivating state or end + factual belief about means and consequences → action

Belief without any concern has a “mind-to-world” role: it represents how things are. Desire or another conative state has a “world-to-mind” role: it disposes the agent to make the world different.²

1.2 “Reason is the slave of the passions” does not mean “be irrational”

The slogan is easy to caricature. Hume does not recommend impulsiveness or obedience to whatever feeling is loudest.

He distinguishes:

• violent passions, such as acute fear, rage, craving, or resentment;
• calm passions, such as benevolence, concern for long-term interest, love of life, and a general orientation toward good.

What people commonly call “reason defeating passion” may actually be a calm, stable motive defeating a violent and short-sighted one. “Strength of mind” is the reliable predominance of the calm motives.¹

Reason still corrects passions indirectly:

• A passion may rest on a false factual supposition.
• An agent may choose ineffective means to an end.
• A person may miscalculate consequences.
• Attention may omit distant or socially distributed effects.

Correcting those errors can change desire and action. The point is that the correction matters only because something about the outcome already engages the agent’s concerns.

1.3 Book III: the motivational argument against moral rationalism

Hume begins Book III by targeting theories according to which virtue consists in conformity to reason and moral distinctions are discovered as rational relations.

His argument can be reconstructed:

1. Morality is practical: moral judgments influence passions and actions.
2. Reason alone cannot generate or oppose motivational impulses.
3. Therefore moral distinctions cannot be conclusions of reason alone.

Hume writes that morals “excite passions, and produce or prevent actions,” whereas reason by itself is motivationally inert.³

The conclusion contains the crucial qualifier alone. Hume is not denying that reasoning contributes to moral judgment. He is denying that a representation’s truth or logical form, considered without any evaluative response, explains moral motivation.

---

2. What reason still does in Hume’s morality

The claim “there is no connection between reason and morality” is therefore too strong. Hume gives reason several indispensable jobs.

2.1 Reason discovers morally relevant facts

Reason and experience tell us:

• whether an action was intentional;
• whether someone was deceived or coerced;
• who was harmed or benefited;
• what consequences an institution tends to produce;
• whether a character trait is useful or agreeable to its possessor or to others;
• whether a proposed rule can sustain cooperation over time.

Without accurate information, sympathy and moral sentiment can be misdirected.

2.2 Reason expands the temporal and causal horizon

Immediate feeling may favor a short-term benefit. Causal reasoning can show that the same act produces long-term injury, erodes trust, or creates dangerous incentives. The end is motivationally supplied, but reasoning profoundly changes what counts as an effective means to it.

2.3 Reason helps correct partiality

Humean moral judgment is not whatever emotion happens to occur first. Moral assessment uses a general or common point of view. We correct for physical distance, personal closeness, temporary mood, misinformation, and idiosyncratic interest so that approval tracks relatively stable features of character and social utility.⁴

This correction uses imagination, comparison, causal inference, and general rules. Moral sentiment remains the approving or disapproving response, but it is a sentiment informed and disciplined by reflection.

2.4 Reason systematizes rules and institutions

Hume’s “artificial virtues,” especially justice and fidelity to promises, depend on conventions whose benefits are often indirect. Reason helps explain how systems of property, promise, and government stabilize expectations. Raw sympathy with the nearest person would not by itself generate or consistently administer those conventions.

A more accurate Humean formula

Reason supplies the map; sentiment supplies the direction of travel. But the map can reveal that the road one wanted is impossible, self-defeating, or harmful to people one already cares about—and that discovery can reorganize motivation.

---

3. Humean moral sentiment is not mere subjective whim

Hume grounds moral distinctions in feelings of approval and disapproval, but his position is not simply:

“I dislike it, therefore it is wrong.”

Several filters separate ordinary preference from moral sentiment:

• Sympathy: the ability to be affected by the happiness and suffering of others.
• General point of view: correction for personal position and partiality.
• Stability: character traits are evaluated across recurring patterns, not only isolated reactions.
• Social utility: traits and institutions are evaluated partly by their tendencies to support human life and cooperation.
• Shared human nature: Hume expects substantial convergence because human beings have broadly similar needs and social capacities.

This does not automatically establish stance-independent moral truth. A critic can still ask whether corrected human approval makes an act right or merely explains why humans call it right. But Hume’s view is an intersubjective and disciplined sentimentalism, not an endorsement of arbitrary appetite.

---

4. The Stoic position: virtue as practical knowledge

The apparent Stoic argument is:

1. The good can be known by reason.
2. Virtue is knowledge of good and evil.
3. A person who knows the good will act accordingly.
4. Therefore moral failure is a failure of knowledge or judgment.

That is a genuine form of ethical intellectualism. But “knowledge” and “reason” mean something much thicker here than possessing a sentence in memory.

4.1 The Stoic sequence: impression, assent, impulse, action

In Stoic psychology:

1. An impression presents something as the case or as appropriate to do.
2. The rational agent assents, rejects, or withholds assent.
3. Assent to a practical impression generates a rational impulse (hormē).
4. The impulse issues in action when not externally blocked.

The Stoics therefore do not place a motivationally dead belief on one side and an unrelated desire on the other. Practical assent is already action-guiding. The Stanford account describes this as a strong form of motivational cognitivism: human actions require and follow the judgment that an action is fitting to perform.⁵

4.2 Passions are evaluative judgments

Stoic passions are not irrational forces invading an otherwise rational mind. A passion is itself a mistaken and excessive evaluative impulse:

• fear assents to the appearance that a future external is a genuine evil;
• grief assents to the appearance that a present external loss is a genuine evil;
• excessive appetite assents to the appearance that an external object is a genuine good that must be obtained.

The Stoic remedy is cognitive because the disorder is cognitive. Correct the valuation and the disordered impulse disappears—not by suppressing a separate emotional substance, but by ceasing to endorse the false judgment. See Stoicism and Emotion - Margaret R Graver.

4.3 “Virtue is knowledge” means an achieved condition of character

Stoic knowledge is:

• stable rather than momentary;
• systematic rather than isolated;
• resistant to counterargument and pressure;
• integrated across judgments, desires, impulses, and action;
• expressed consistently in conduct.

A person who repeats “virtue is the only good” but panics, cheats, or betrays others when status is threatened does not yet possess Stoic knowledge in the strict sense. They possess an unstable opinion or verbal formula.

This is why the Stoic can deny ordinary cases of akrasia—acting against one’s better judgment. What looks like weakness of will is re-described as:

• rapid oscillation between incompatible judgments;
• incomplete or unstable cognition;
• assent to the immediate appearance that the tempting act is, right now, better or more fitting.

4.4 The Stoic position is not a one-time lesson

Stoic training includes:

• repeated examination of impressions;
• premeditation of adversity;
• attention to one’s impulses;
• rehearsal of principles;
• correction of value judgments;
• role-based practice and performance of appropriate actions;
• formation of stable habits of assent.

So the practical model is closer to cognitive-affective retraining than to “read the definition of good and become virtuous.”

---

5. Where Hume and the Stoics genuinely disagree

The deepest disagreement concerns the architecture of mind:

• Hume: representation and motivation are different types of mental state.
• Stoics: mature human motivation is itself rational and propositional; practical judgment culminates in impulse.

---

6. Where they may be talking past each other

Hume’s bare belief and the Stoic’s practical knowledge are not equivalent.

Suppose a person says:

“I know that returning the money is right, but I want to keep it.”

Three analyses are possible:

Humean analysis

• The person believes returning it is morally required.
• That belief activates some moral concern, but greed is stronger.
• The action reflects the winning motive, not a failure of logic.

Strict Stoic analysis

• The person’s verbal endorsement is not knowledge.
• At the moment of action, the person assents to “keeping the money is preferable for me now.”
• The conflict reveals unstable opinion and mistaken valuation.

Ordinary psychological analysis

• The person represents incompatible goals at different levels.
• Moral identity, reward expectation, habit, arousal, opportunity, executive control, and social monitoring interact.
• Which representation controls behavior depends on context and learned policy, not on one timeless faculty winning a duel.

If the Stoic defines knowledge partly by stable action, the thesis “knowledge guarantees action” becomes difficult to falsify: any failure of action proves that knowledge was absent. But if Hume defines every action-producing state as a passion, the thesis “passion motivates every action” risks a parallel circularity. Each system builds its conclusion into its taxonomy.

---

7. The strongest case for Hume

7.1 The amoralist and the indifferent agent

A person can understand a moral argument, predict harm accurately, and use moral vocabulary while remaining indifferent. Psychopathy, severe callousness, depression, burnout, and ordinary selfishness provide different examples in which cognitive representation and motivation can dissociate.

This supports the idea that recognizing a proposition does not automatically energize conduct.

7.2 The same belief can support opposite actions

Two people may agree that an act will cause suffering:

• one avoids it from compassion;
• another pursues it from cruelty;
• a third remains indifferent;
• a fourth accepts it as a regrettable means to another goal.

The descriptive belief underdetermines action until it is integrated with valuation.

7.3 Instrumental rationality needs an end

Means–end reasoning can optimize a route only after some outcome is treated as worth pursuing or avoiding. Pure calculation does not choose its objective function.

7.4 The judgment–action gap

People often endorse a norm or intention yet fail to enact it. Moral-psychology reviews explicitly identify a judgment–action gap, while broader behavioral research documents large intention–behavior gaps.⁶ A 2023 meta-analysis in physical activity found that 47.6% of intenders failed to realize the intended behavior; this is not a moral-action study, but it is strong evidence against treating conscious intention as sufficient for action.⁷

7.5 Motivational disorders

People with depression or apathy can retain factual and moral understanding while losing initiative. This is difficult to explain if knowledge by itself necessarily produces action.

---

8. The strongest objections to Hume

8.1 Reason can transform ends, not merely select means

Deliberation can lead people to abandon a goal, acquire a commitment, or identify with a principle they did not previously treat as important. Examples include conversion, moral education, recognizing another person’s equal standing, or seeing that a prejudice is inconsistent with one’s other commitments.

The Humean can redescribe this as one desire reorganizing others, but the anti-Humean asks whether that explanation adds anything or merely labels every motivational change as desire-driven.

8.2 Some beliefs appear intrinsically action-guiding

Believing “this child is in immediate danger and I can easily help” can seem to generate motivation without a separately experienced desire. Rationalists argue that recognizing a genuine reason can itself move a rational agent.

Humeans reply that concern for the child, aversion to suffering, moral identity, or a standing disposition supplies the motive even when it is phenomenologically quiet.

8.3 Hume’s categories may be too coarse

Contemporary psychology does not treat beliefs and desires as two simple, isolated boxes. Goals, expected values, affective predictions, action policies, habits, and control signals are distributed and mutually modifying. A state can carry both representational and motivational information.

8.4 Moral reasons may apply without current desire

Even if desire is psychologically necessary for action, it does not follow that an agent has a normative reason only when it serves an existing desire. Otherwise, a cruel person’s lack of benevolent desire could imply that they lack a reason not to harm. This inference—from motivational Humeanism to reasons internalism—is heavily contested.⁸

8.5 The “slave” metaphor understates feedback

Reason is not merely a passive servant executing fixed preferences. Reflection can:

• change salience and attention;
• alter anticipated emotion;
• expose inconsistency;
• reframe identity;
• build rules and implementation plans;
• restructure the environment;
• train future habits;
• lower or raise the subjective value of an option.

Even if reasoning never acts without valuation, it helps constitute the valuations that later guide action.

---

9. The strongest case for the Stoics

9.1 Practical knowledge is skill, not information

Knowing how to act well resembles knowing a craft more than memorizing a theorem. The skilled musician’s perception, attention, motor preparation, and evaluation are trained together. Likewise, Stoic wisdom is an organized capacity to perceive situations correctly and respond appropriately.

9.2 Appraisals shape emotions

What a person believes an event means strongly affects fear, anger, shame, and grief. Reappraisal can change both feeling and behavior. This supports the Stoic claim that evaluative cognition is not merely downstream of passion.

9.3 Repeated assent creates character

Habits of attention and judgment make certain responses easier and more automatic. Stoic exercises correctly recognize that ethical change requires repeated practice before crisis, not abstract reflection during crisis alone.

9.4 “Reason” can include properly ordered motivation

The Stoic is not committed to a cold calculator commanding an alien emotional animal. Rationality is the coherent organization of the whole mature person. Under that interpretation, the modern discovery that cognition, valuation, and action are integrated is not a refutation; it resembles the Stoic monistic picture.

---

10. The strongest objections to the Stoics

10.1 Knowledge is redefined so that counterexamples disappear

If any person who acts wrongly is said not truly to know the good, the theory risks becoming unfalsifiable. Ordinary language allows that people sincerely know what they should do and still fail.

10.2 Human control is fragmented and state-dependent

Sleep loss, addiction, stress, hormonal state, cue exposure, depression, and learned habit can alter action without erasing explicit knowledge. A unitary rational faculty does not capture all of this variation.

10.3 Explicit judgment is often a weak predictor

What people say they value may diverge from behavior because of incentives, situational pressure, automaticity, or self-deception. Moral education that teaches propositions without changing practice and environment often fails.

10.4 Emotions are not merely false propositions

Modern affective science treats emotions as involving bodily regulation, action readiness, learning, attention, memory, and social communication. Appraisal matters, but it is not the whole phenomenon. Some affective responses occur before explicit conceptual assessment and cannot be eliminated simply by argument.

10.5 The sage standard is psychologically extreme

The Stoic equation of complete virtue with perfectly stable knowledge leaves almost all humans in the category of fools. It is normatively powerful but empirically less useful if it does not model degrees of competence and predictable failure.

---

11. Contemporary philosophical positions

No single philosophical label is forced by the science. Empirical findings can constrain psychological claims while leaving semantic, normative, and metaphysical questions open.

---

12. What modern science actually shows

Haidt, Chapter 2 — useful Humean evidence, but not the last word

Haidt uses Damasio’s patients to argue that intact IQ, rule knowledge, and verbal moral reasoning do not restore practical judgment when affective valuation is severely disrupted. He then replaces an emotion-versus-cognition dichotomy with intuition versus conscious reasoning: emotions contain appraisal, and many automatic intuitions are not full emotional episodes. This supports the note’s claim that abstract representation needs value and salience to guide action.

The chapter’s stronger slogan—that reasoning usually serves intuition—requires the qualifications developed below. Haidt’s own cognitive-load null results were unpublished, his dumbfounding interview involved thirty students, and he admitted having no daily-life frequency evidence for direct reasoned judgment or private reflection. Read Chapter 2 as a historically important model, not as proof of a brain divided into a passionate master and rational servant. Damasio and practical judgment, PDF pp. 52–53 (file not included in this release) · Intuition versus reasoning, PDF pp. 64–67 (file not included in this release) · Endnote 44, PDF p. 369 (file not included in this release) · Permanent model note

12.1 First caution: science does not find a “reason module” and a “passion module”

The brain is not divided into a rational cortex sitting above an emotional animal brain. Goal-directed behavior depends on interacting cortical and subcortical networks. Regions associated with valuation, memory, bodily state, social cognition, control, and motor preparation exchange information recurrently.

The useful scientific contrast is not reason versus emotion, but partially distinguishable operations such as:

• representing facts and rules;
• assigning expected value;
• learning from prediction error;
• selecting goals;
• estimating effort and opportunity cost;
• inhibiting or switching responses;
• planning sequences;
• forming habits;
• energizing action;
• monitoring social consequences.

12.2 Motivation requires value and expected consequences

Effort-based decision research implicates a distributed network:

• ventromedial prefrontal cortex (vmPFC): integrates reward- and value-related information;
• ventral striatum: represents expected subjective value and reward;
• dorsal anterior cingulate cortex (dACC): monitors costs, benefits, and control allocation;
• motor systems: prepare and energize effortful action.⁹

This is Hume-friendly: a representation influences action through systems that assign significance, cost, or reward. A fact treated as wholly indifferent produces no obvious reason for the organism to expend effort.

But these systems do not demonstrate that value must originate in a pre-existing, fixed passion. Value representations are learned and revised by cognition, social experience, language, and long-term planning.

12.3 Affective valuation is causally important for prosocial action

A 2024 lesion study tested real effort for rewards benefiting oneself or another person. Participants with focal vmPFC damage were less willing to exert effort for another person and sometimes energized chosen prosocial actions less strongly, while self-benefiting willingness was comparatively preserved.¹⁰

This provides unusually strong causal evidence that integrated value/effort processing contributes to helping behavior. It supports Hume against the idea that abstract recognition alone guarantees prosocial motivation.

It does not prove:

• that vmPFC is a “moral sentiment center”;
• that emotion is the sole cause of morality;
• that people with lesions lack moral knowledge;
• that sentimentalism is normatively true.

12.4 Deliberative control also causally changes choice

A 2024 systematic review and meta-analysis of 33 non-invasive brain-stimulation studies found that changing dorsolateral prefrontal cortex activity altered delay discounting, risk-taking, and fairness-related choices. Excitatory stimulation of left DLPFC was associated with more self-controlled intertemporal and risk decisions, while inhibition of right DLPFC increased self-interested responses to unfair offers.¹¹

This is Stoic- and rationalist-friendly in a limited sense: control-related processes can causally alter action rather than merely rationalize a decision after the fact.

It does not show that DLPFC activity detects moral truth or motivates independently of value. Cognitive control is itself allocated partly according to expected benefit and effort cost.

12.5 The judgment–action gap is real, but undermeasured in morality research

An analysis of 1,278 empirical morality studies found that the literature disproportionately studies judgments, reasoning, and self-reports while providing much less evidence about actual moral behavior. The authors call for models connecting cognition, emotion, identity, group context, and action rather than isolating them.⁶

This warrants two conclusions:

1. Provisional Humean support: knowing or endorsing a norm is not reliably sufficient for enactment.
2. Epistemic humility: moral psychology has often inferred too much about behavior from hypothetical dilemmas and verbal judgment.

12.6 Moral judgment uses both intuitive and deliberative processing

Early dual-process models associated deontological judgments with emotion and utilitarian judgments with controlled reasoning. Later work shows that this mapping is too simple:

• consequence sensitivity and norm sensitivity can vary independently;
• action versus inaction bias can masquerade as a moral principle;
• intuitions can incorporate learned rules;
• deliberation can defend deontological as well as consequentialist conclusions;
• emotions can encode information rather than merely disrupt reasoning;
• reasoning is frequently recruited to interpret intention, causation, and context.

The better model is interactive and plural rather than “emotion votes deontology; reason votes utility.”¹²

12.7 The somatic-marker evidence is suggestive, not decisive

The Iowa Gambling Task helped popularize the idea that bodily and affective signals guide decision-making under uncertainty. Patients with vmPFC damage often perform poorly and may fail to develop normal anticipatory physiological responses.

However:

• later studies found more conscious knowledge of deck quality than early reports detected;
• task structure permits multiple cognitive explanations;
• skin-conductance effects are real but generally modest and methodologically variable;
• executive control, reversal learning, working memory, and explicit knowledge also contribute.¹³

The defensible conclusion is that affective signals contribute to decision-making, not that unconscious emotion independently decides while reason merely narrates.

12.8 Dopamine is not simply desire or pleasure

Dopamine systems contribute to reward prediction error, learning, salience, effort, and incentive motivation. They help update expectations when outcomes differ from predictions.¹⁴

This again favors an integrated model:

• cognition supplies predictions and models of outcomes;
• learning systems compare prediction with result;
• valuation changes;
• future attention and action policies are updated.

Calling dopamine “passion” would be as misleading as calling prefrontal cortex “reason.”

---

13. The best current scientific model

The most defensible synthesis is a recurrent value–control–action model:

flowchart LR
    A["Perception and factual beliefs"] --> B["Appraisal and value assignment"]
    B --> C["Goal and action selection"]
    C --> D["Planning and cognitive control"]
    D --> E["Motor preparation and action"]
    E --> F["Outcome and social feedback"]
    F --> G["Learning and habit updating"]
    G --> A
    G --> B
    D --> B
    B --> D
    H["Bodily state and emotion"] --> B
    H --> D
    I["Identity, norms, and relationships"] --> A
    I --> B
    I --> C

The arrows run in both directions because:

• beliefs change valuation;
• valuation controls attention and reasoning effort;
• planning changes the environment and future cues;
• action produces feedback;
• repeated action becomes habit;
• bodily and social states alter which goals feel urgent;
• reflection can reshape identity and long-term preference.

No single stage is “the moral faculty.” Moral behavior is a system-level achievement.

---

14. Which philosopher does the science favor?

Hume is strongly supported on these points

• Bare factual or propositional representation is normally insufficient for action.
• Motivation depends on the significance or value assigned to an outcome.
• Knowing, judging, and intending can dissociate from behavior.
• Calm, long-term motives can oppose acute impulses without constituting pure passionless reason.
• Social feeling and sympathy contribute to moral action.

The Stoics are strongly supported on these points

• Appraisals help constitute emotion and action readiness.
• Repeated judgments and attentional habits reshape future responses.
• Control-related processes can causally change behavior.
• Practical competence is more than verbal assent; it is stable, trained organization of perception, valuation, and response.
• Emotion and cognition are not separate substances fighting for control.

Hume is not scientifically established on these stronger claims

• That all ends ultimately come from passions that reason cannot transform.
• That a belief can never possess intrinsic motivational content.
• That reason’s role is adequately captured as mere service to antecedent desire.
• That sentimental origin settles the truth conditions of moral claims.

The Stoics are not scientifically established on these stronger claims

• That genuine moral knowledge guarantees right action.
• That all apparent akrasia is merely oscillating judgment.
• That passions are exhaustively reducible to false assent.
• That complete rational consistency is psychologically attainable.

Best overall verdict

Modern science is Humean about the insufficiency of bare belief, Stoic about the power of appraisal and trained practical cognition, and anti-dualist about the architecture of both. It most closely resembles a hybrid in which reasoning works by participating in valuation and control—not by issuing motivationally inert commands, but also not merely by obeying fixed passions.

---

15. What this means for moral education

If the Humean side is right, moral education cannot stop at argument:

• cultivate empathy and concern;
• make consequences vivid;
• connect principles to identity and relationships;
• design incentives and environments that support action;
• practice until conduct becomes easier and more automatic.

If the Stoic side is right, emotion need not be accepted as fixed:

• examine the appraisal embedded in the response;
• rehearse alternative interpretations;
• distinguish genuine goods from salient temptations;
• practice attention and assent before high-pressure situations;
• develop stable action policies and role-based commitments.

The synthesis is:

Teach the truth, cultivate the motive, rehearse the action, and shape the situation.

Knowing what is good matters. Caring about it matters. Possessing the control and habits to enact it matters. None is reliably sufficient alone.

---

16. Final assessment of the original claim

“Hume says reason cannot motivate our actions.”

Mostly correct with an essential qualifier: reason alone cannot originate motivation. Reason can redirect action by changing beliefs about objects, causes, consequences, and means.

“Therefore there is no connection between reason and morality.”

Incorrect. For Hume, reason is indispensable for informed, stable, impartial moral judgment. It does not supply the approving or disapproving sentiment by itself.

“This counters the Stoic view that one first knows the good and then acts on that knowledge.”

It counters a thin version of that view. It is powerful against the claim that possessing a correct proposition is sufficient for action. It is less decisive against the actual Stoic theory, where practical knowledge is an integrated and stable condition whose assent already generates impulse.

“What does science say?”

Science rejects the simple faculty duel. Motivation requires value-sensitive and action-energizing processes; control and reasoning can alter choices; explicit judgment often fails to become conduct; and affect, cognition, learning, identity, and context are dynamically integrated.

---

Confidence table

---

Handwritten Notebook Pages

Source

Open Can reason motivate moral action? notebook pages — start at notebook p. 147 (file not included in this release).

Sources

Primary philosophy

• David Hume, A Treatise of Human Nature, II.iii.3, “Of the influencing motives of the will”.
• David Hume, A Treatise of Human Nature, III.i.1, “Moral distinctions not derived from reason”.
• David Hume
• Hume Is-Ought Problem
• 60.12.1 Stoicism
• Stoicism and Emotion - Margaret R Graver

Philosophical scholarship

• Rachel Cohon, “Hume’s Moral Philosophy,” Stanford Encyclopedia of Philosophy.
• “Moral Motivation,” Stanford Encyclopedia of Philosophy.
• “Reasons for Action: Internal vs. External,” Stanford Encyclopedia of Philosophy.
• “Stoicism,” Stanford Encyclopedia of Philosophy.
• “Stoic Ethics,” Internet Encyclopedia of Philosophy.

Scientific evidence and reviews

• Brassard et al. (2024), “Neurobiological foundations and clinical relevance of effort-based decision-making”, Brain Imaging and Behavior.
• Lockwood et al. (2024), “Human ventromedial prefrontal cortex is necessary for prosocial motivation”, Nature Human Behaviour.
• Lin & Feng (2024), “Lateralization of self-control over the dorsolateral prefrontal cortex in decision-making”, Cognitive, Affective, & Behavioral Neuroscience.
• Ellemers et al. (2019), “The Psychology of Morality: A Review and Analysis of Empirical Studies Published From 1940 Through 2017”, Personality and Social Psychology Review.
• Feil, Fritsch & Rhodes (2023), “The intention–behaviour gap in physical activity”, British Journal of Sports Medicine.
• Garrigan, Adlam & Langdon (2016), “Moral judgment as information processing: an integrative review”, Frontiers in Psychology.
• Gawronski et al. (2017), “Consequences, norms, and generalized inaction in moral dilemmas: The CNI model”, Journal of Personality and Social Psychology.
• Eres et al. (2025), “Intuitive deontology? A systematic review and multivariate, multilevel meta-analysis”.
• Wemm & Wulfert (2025), “Decision-making and performance in the Iowa Gambling Task: recent ERP findings and clinical implications”.
• Simonovic et al. (2019), “Sweating the small stuff: A meta-analysis of skin conductance on the Iowa Gambling Task”.
• Watabe-Uchida, Eshel & Uchida (2017), “Neural Circuitry of Reward Prediction Error”, Annual Review of Neuroscience.

Footnotes

1. Hume, Treatise II.iii.3.2–10. The primary text explicitly assigns reason the discovery of causal relations and effective means while locating the impulse in aversion, propensity, and calm or violent passion. ↩ ↩²

2. “Moral Motivation,” Stanford Encyclopedia of Philosophy, on Humean and anti-Humean theories and the direction-of-fit argument. ↩

3. Hume, Treatise III.i.1.5–8. The argument moves from morality’s practical influence and reason’s alleged inertness to the claim that moral rules are not conclusions of reason alone. ↩

4. See Cohon, “Hume’s Moral Philosophy,” sections on moral sentiment, sympathy, and the common point of view. ↩

5. “Stoicism,” Stanford Encyclopedia of Philosophy, §3.6: assent to a practical impression yields rational impulse; the view rejects simultaneous conflicting impulses and reinterprets akrasia. ↩

6. Ellemers et al. (2019), content and bibliometric analysis of 1,278 empirical studies. The review emphasizes the field’s heavy reliance on judgments and self-reports and its limited integration of actual behavior, emotion, identity, and group context. ↩ ↩²

7. Feil, Fritsch & Rhodes (2023), 25 samples and 29,600 participants. Physical activity is used only as general evidence about intention translation, not as a direct estimate of the moral judgment–action gap. ↩

8. “Reasons for Action: Internal vs. External,” Stanford Encyclopedia of Philosophy, especially the distinction between the Humean theory of motivation and stronger claims about normative reasons. ↩

9. Brassard et al. (2024), review of 39 effort-based decision studies. Regional summaries are functional tendencies within interacting networks, not one-region/one-function identities. ↩

10. Lockwood et al. (2024): 25 vmPFC-lesion patients, 15 other-lesion controls, and 40 healthy controls completed incentivized choices involving effort for self or another person. ↩

11. Lin & Feng (2024): systematic review and random-effects meta-analysis of 33 studies and 130 effect sizes. Effects varied by hemisphere, stimulation type, and decision domain. ↩

12. Garrigan et al. (2016); Gawronski et al. (2017); Eres et al. (2025). Together they support separating information inputs, consequence sensitivity, norm sensitivity, action bias, intuitive processing, and deliberation rather than inferring a single process from a deontological or utilitarian response. ↩

13. Wemm & Wulfert (2025); Simonovic et al. (2019); and Maia & McClelland (2004), “A reexamination of the evidence for the somatic marker hypothesis”. ↩

14. Watabe-Uchida, Eshel & Uchida (2017), review of the circuits producing reward-prediction-error signals. This does not reduce every dopamine signal or motivational function to one computation. ↩