Moral Responsibility

The Problem and Its Philosophical Scope

Moral responsibility concerns the conditions under which an agent's conduct, omissions, attitudes, or their consequences may properly be attributed to that agent as grounds for moral appraisal and interpersonal accountability. It asks what connects a deed with a person in the sense relevant to justification, excuse, blame, praise, demands for an explanation, and moral repair. Responsibility therefore begins with a relationship of answerability: a person can be asked why she acted and can be assessed in light of the considerations to which she was capable of responding. Whether this relationship supports blame, punishment, reconciliation, or merely an acknowledgment of agency is a further question.

The term has several uses that must remain distinct. Prospective responsibility concerns duties that someone bears before acting, such as an obligation to protect a dependent person. Retrospective responsibility concerns what is attributable to an agent after an action or omission. Causal responsibility identifies contributions to an event, whereas moral responsibility assesses those contributions through standards of agency and normativity. Legal liability determines institutional consequences under a legal system; it may overlap with moral accountability without being identical to it. Role responsibility arises from a position or undertaking; responsibility as a virtue describes the settled quality of being conscientious. The central subject of this dossier is retrospective moral responsibility, investigated through its connections to these other senses.

A responsible agent, an action for which an agent is responsible, and the appropriateness of holding that agent responsible are three separate objects of judgment. The first concerns general capacities; the second concerns a particular exercise of those capacities under actual circumstances; the third concerns how other people should respond. An adult might possess normal deliberative capacities yet perform an action while unconscious. A person might be responsible for a wrongdoing while an angry public response remains disproportionate or informationally mistaken. A practice may have beneficial consequences while a particular accusation within it remains unjustified. This three-level structure prevents the analysis from building its conclusions into the definition of responsibility.

The standard research questions follow. What control matters? Which facts and normative considerations must the agent understand? How does the agent's history affect ownership of a decision? Are alternative possibilities necessary? Can responsibility survive causal determinism, manipulation, or moral luck? What makes blame fitting, what makes punishment justifiable, and how can responsibility be distributed across institutions and technological systems? The classical questions are reconstructed from their primary texts; contemporary developments are assessed as live disputes rather than settled doctrine. For a map of the present philosophical field, see Moral Responsibility in the Stanford Encyclopedia of Philosophy (2024 revision).

Attribution, Accountability, and the Grammar of Moral Assessment

There are at least three philosophically important relationships between a person and conduct. Attributability holds when an attitude or action expresses something evaluatively significant about the agent's commitments, character, or practical outlook. Answerability holds when the agent can be called upon to offer reasons, explanations, and justifications. Accountability holds when others have standing to make demands or to respond to unmet moral expectations. The distinctions are analytic instruments; particular theories disagree about whether one relation is basic or whether they are mutually constitutive.

A person's contemptuous remark may express a vice even when unusual emotional pressure diminishes the fairness of reprimand. Conversely, a competent professional might be accountable for an institutional failure despite having no malicious attitude. The second example requires care: an obligation to provide repair or an explanation need not imply that the individual deserves moral condemnation. The relevant question is what kind of moral response the attribution warrants. David Shoemaker's Responsibility from the Margins (2015) develops different forms of responsibility by investigating the emotions, expectations, and abilities relevant to them; T. M. Scanlon's What We Owe to Each Other (1998) places the significance of blame within relations altered by what an action reveals about an agent's attitudes.

Retrospective blameworthiness is thus narrower than being a cause, a duty-bearer, or a participant in a harmful practice. The vocabulary of excuse and justification introduces a further difference. A justification holds that the action was permissible, or that the agent had sufficient moral reason to act as she did. An excuse accepts an objectionable act while explaining why the agent was not blameworthy, or was less blameworthy, in performing it. An exemption concerns the agent's standing as a participant in the relevant demands, often because requisite capacities are absent. The same fact can matter differently across contexts: severe duress may partly excuse a wrong without making it right.

One must also distinguish responsibility for an action from responsibility for an outcome. The outcome may depend upon conditions beyond intention and control. A person can be blameworthy for driving recklessly when no collision occurs; another person may cause a collision through a careful, unforeseeable maneuver without culpability. Evaluating the action's practical quality and its eventual harms requires separate evidential and moral judgments. A theory that collapses these judgments fails to explain why accidents, attempts, negligence, and risk-taking elicit different responses.

Aristotle: Voluntariness, Choice, and Character

Aristotle gives the classical analysis its durable starting point in Nicomachean Ethics III.1–5. Since virtues and vices concern actions and feelings that attract praise and blame, he investigates the conditions of the voluntary (hekousion) and involuntary (akousion). Acts arising from an external principle to which the agent contributes nothing are paradigmatically forced; ignorance of the relevant particulars can also render an act involuntary. In W. D. Ross's translation, Aristotle's examples include a person carried by force, the throwing of cargo overboard during a storm, and an action performed under threats to one's family. The Book III text permits examination of the distinctions in their original argumentative setting.

The storm case is conceptually important because Aristotle declines to decide moral status by classifying an action abstractly. Throwing away cargo is unwanted in isolation yet chosen as a means of saving the vessel under circumstances. Agency must be assessed at the time of action and against the alternatives actually available. Threats can produce mixed actions: the agent contributes through deliberation while the circumstances transform what a reasonable choice can be. This framework anticipates graded excuse and contextual assessment without converting Aristotle into a modern compatibilist.

The treatment of ignorance is similarly discriminating. Aristotle distinguishes acting because of ignorance from acting while ignorant, and ignorance of particular facts from failures of practical judgment concerning what is good. The remorse or distress that accompanies discovery helps distinguish kinds of involuntariness in his analysis, although contemporary theories need not accept remorse as a necessary condition for excuse. The underlying point is that accountability depends upon how the agent's understanding is related to the action, rather than on a mere inventory of true beliefs.

In III.2–3 Aristotle distinguishes deliberate choice (prohairesis) from wish and appetite, connecting choice to deliberation about what lies within one's practical scope. In III.5 he links character formation to repeated voluntary conduct and treats virtue and vice as appropriately assessable. Here the difficult problem emerges: the traits that shape future decisions are themselves shaped by earlier actions, upbringing, and circumstance. Aristotle's discussion grounds a historical account of agency while leaving later philosophers the task of specifying when the gradual formation of character is sufficiently one's own. His analysis concerns the ethical life of a situated agent, not a detachable metaphysical property called free will.

Augustinian and Thomistic Transformations

Late antique and medieval Christian philosophy reorients the problem toward will, sin, grace, ignorance, and divine judgment. Augustine's De libero arbitrio (On Free Choice of the Will; trans. Williams, 1993) connects the moral attribution of wrongdoing to the will while addressing why a created, good order contains evil. This theological framework raises questions distinct from Aristotle's: how can the creature's responsibility coexist with dependence upon God, and how is wrongdoing attributable to a will that may itself be disordered? The familiar contrast between voluntary wrongdoing and unavoidable error becomes central, but Augustine's varying writings on will and grace should not be forced into a single timeless formula.

Thomas Aquinas integrates Aristotelian psychology with a developed account of intellect, will, and law. In Summa theologiae I–II, question 6, he distinguishes voluntariness under force, fear, desire, and ignorance. His analysis of ignorance identifies what accompanies a choice, what precedes it without fault, and what results from culpable negligence or deliberate avoidance. The latter can be imputable precisely because the agent failed to know what she both could and ought to know. These distinctions appear in the available English question 6 and question 76 translations, which are used here with article identifiers rather than fictive modern page numbers.

Aquinas makes a subtle move: ignorance's excusing role depends partly on its causal relation to conduct. Ignorance is not always sufficient to excuse, since an agent might act identically if informed, or might have actively cultivated ignorance. Yet the apparently straightforward demand that one should have known can become circular. If guilt for an ignorant act is transferred to earlier negligent ignorance, that earlier negligence must itself meet appropriate epistemic and control conditions. The resulting regress is one source of later skepticism.

Medieval theories also supply distinctions among intention, consent, circumstance, and culpability that persist in contemporary moral reasoning. The theological concepts cannot simply be reproduced as secular philosophical facts; their enduring analytic contribution is the careful examination of the will's relation to knowledge, coercion, and standards of conduct. Whether grace and predestination modify the limits of human agency remains a separate historical debate with its own doctrinal sources.

Hume, Necessity, and the Intelligibility of Character

David Hume's An Enquiry Concerning Human Understanding, section VIII, “Of Liberty and Necessity,” develops a compatibilist reconstruction of the dispute. His argument proceeds from the observable regularities of motivation, character, circumstance, and conduct. Moral relations depend upon the possibility of connecting acts with enduring features of agents. If conduct were wholly disconnected from character, an action would provide little basis for moral assessment of the person. Hume regards necessity in this context as intelligible regularity and inference, rather than as a directly perceived metaphysical chain.

His account of liberty concerns the capacity to act or refrain according to the determinations of the will, contrasted with external constraint. This definition permits conduct to be voluntary while having causal antecedents. It is important to distinguish Hume's historical proposal from every contemporary thesis called compatibilism. A modern source-compatibilist may demand ownership, reasons responsiveness, and a suitable history that Hume does not articulate as a single test. Hume's section VIII provides the relevant original argumentative sequence in the 1777 text as transmitted in the Selby-Bigge edition.

A strong objection arises from an agent whose will has been installed by manipulation. Even if that person acts according to her current desires, the origin of those desires might impair responsibility. Hume's emphasis on the integration of action and character makes this concern visible, even though his own explanatory priorities differ from later historical theories. Another objection asks why a deterministic connection between an evil character and a harmful action should make condemnation fair when the person did not choose the initial conditions of character.

Hume's lasting achievement is to separate causal intelligibility from compulsion. Motives can explain an action without functioning as irresistible external force. The further issue is normative: which forms of motive-driven agency justify interpersonal demands? That is the question transformed by Kant, Strawson, Frankfurt, and their successors.

Kant: Imputation, Autonomy, and the Good Will

Kant's conception of morality places the authority of practical reason and the autonomy of the will at the center of responsibility. The Groundwork of the Metaphysics of Morals (1785), especially Academy pagination 4:393–4:405 and 4:412–4:448, develops the distinction between acting from duty and acting merely in conformity with duty. Moral worth attaches to the maxim and its relation to duty, not simply to the fortunate production of good consequences. Freedom is investigated as a condition of rational self-legislation, and Kant's critical treatment of theoretical knowledge prevents an easy inference from empirical psychology to the metaphysical status of autonomy.

Kantian imputation should not be reduced to assigning an isolated causal outcome to a person. It concerns an action regarded as issuing from an agent capable of acting under practical laws. The Metaphysics of Morals (1797), particularly its introductory treatment of deeds, imputation, and obligation, articulates the juridical and ethical significance of attributing deeds to persons. Kant's ethics also distinguishes moral personality from empirical descriptions of impulse, an issue that later becomes contentious when people ask how rational capacities arise in concrete developmental conditions.

The familiar contrast between good will and lucky success illustrates a deep insight. Two agents might adopt equally reckless maxims; one causes harm while the other encounters an empty road. A consequence-sensitive morality may reasonably distinguish their duties of repair, but it must explain why a difference in luck alters blameworthiness for choosing the risk. Kant provides an influential rationale for keeping appraisal of the will conceptually distinct from appraisal of results.

The principal tension concerns access to the maxim and the strength of the claim that rational self-legislation can secure responsibility against a causally structured natural world. A person's maxims are often mixed, opaque even to herself, and shaped by institutions. Responsibility requires practical standards that can be applied under conditions of imperfect self-knowledge. The Kantian contribution remains essential to any view that grounds accountability in reasons and respect, though it does not settle the empirical or metaphysical question of freedom by itself.

Nietzsche and the Genealogy of the Responsible Agent

Nietzsche's On the Genealogy of Morality (1887), second essay, investigates promises, memory, punishment, debt, and the cultural making of an agent who can answer for the future. His genealogical method asks how practices of promising, remembering, and enforcing obligations transform human beings and their interpretations of guilt. The early sections of the second essay are particularly relevant because they place the capacity to make commitments within a history of social formation rather than treating responsibility as a timeless psychological endowment.

Nietzsche thereby changes the explanatory question. Before asking whether an individual deserves a particular reproach, one can ask how the apparatus of deserving, owing, and being answerable came to exert authority. His depiction is polemical and historically conjectural in ways that require critical scrutiny. The analysis of punishment, cruelty, and debt cannot serve as a neutral historical record of all moral cultures; it is a philosophical genealogy whose theoretical force lies in questioning the origins and functions of moral practices.

Genealogy poses two challenges. First, practices that appear to express impartial judgment may reproduce relations of power. Second, the development of conscientious self-command may have depended on mechanisms that are themselves morally troubling. Neither conclusion directly establishes that responsibility is illusory. The fact that a practice has an objectionable history does not entail that every possible version of it is unjustified. A defender of responsibility must nonetheless identify reasons for accountability that survive criticism of its inherited social uses.

Nietzsche helps broaden the dossier beyond the free-will problem. Accountability is also a cultural institution formed through memory, expectation, authority, and public judgment. Its rational reconstruction must examine both the agent's powers and the conditions under which communities claim standing to demand an answer.

Other Philosophical Traditions: Intention, Cultivation, and Relationship

Comparative inquiry shows that questions neighboring moral responsibility arise in philosophical traditions whose primary vocabularies do not replicate modern English debates over desert. In Confucian ethical thought, self-cultivation, role-appropriate conduct, ritual practice (li), and humane concern (ren) connect the evaluation of conduct with the quality of one's relationships and development. The Analects, especially books 4 and 12, presents a moral life in which correction of oneself and responsiveness to others are central (Confucius, trans. Slingerland, 2003). Its questions about exemplary agency and the demands of social roles illuminate responsibility as cultivated answerability, although it is historically misleading to attribute contemporary compatibilism or a modern theory of blame directly to Confucius.

Buddhist ethical traditions place special weight on intention (cetanā) in the moral interpretation of action (karma); the Nibbedhika Sutta (Aṅguttara Nikāya 6.63) connects volition to deeds performed by body, speech, and mind (trans. Bhikkhu Sujato). This focus supports a distinction between what a person deliberately undertakes and what merely happens as a consequence. Buddhist accounts of causal dependence and the denial of a permanent substantial self also complicate attempts to ground responsibility in an absolutely self-creating ego. Different Buddhist schools develop distinct metaphysics, moral psychologies, and accounts of conventional agency; a single slogan about “no-self” cannot substitute for these arguments.

The comparative result is methodological. Responsibility can be investigated as a capacity for reflective choice, an accountability relationship, the cultivation of character, the moral significance of intention, or a convention sustained within an account of causation and personhood. These approaches identify overlapping problems while locating their grounds differently. Translation into the universal language of “free will” can obscure precisely the conceptual differences that comparative philosophy seeks to disclose.

Cross-traditional study also tests the assumption that punishment and indignation form the exclusive core of moral responsibility. Cultures may place special emphasis on apology, restitution, formation of character, restoration of relations, or release from destructive attachment. Such approaches can be normatively serious without treating the intensity of blame as the measure of accountability. A comprehensive theory must account for the plurality of morally appropriate responses.

Determinism, Alternative Possibilities, and Sourcehood

Causal determinism is the thesis that, given the relevant prior state of the world and the laws of nature, only one future history is possible. If that thesis is true, an action may be inevitable relative to those antecedents even when the agent deliberates, chooses, and acts without external restraint. The issue is whether the type of control necessary for responsibility includes an ability to do otherwise under precisely those conditions.

The principle of alternative possibilities (PAP) states, in its familiar form, that an agent is morally responsible for an action only if the agent could have done otherwise. Its attractions are evident in cases of force and incapacity. If an outcome could not be prevented, blaming a person for failing to prevent it can seem unfair. Yet this intuitive principle can conflate the absence of alternatives with the mechanism by which an action actually occurs.

Compatibilists deny that determinism by itself defeats the control required for responsibility. Their proposed control may involve acting from one's motives, responding to relevant reasons, or exercising capacities embedded in a suitable personal history. Incompatibilists argue that responsibility requires a sourcehood or freedom that a fully determined history cannot supply. Libertarian accounts defend indeterministic freedom, sometimes through agent causation or indeterministic deliberation. Hard incompatibilists deny that either determinism or indeterminism provides the ultimate control necessary for basic-desert responsibility. These positions should not be reduced to a binary between moral optimism and pessimism: incompatibilists can preserve forward-looking responsibility practices, and compatibilists may sharply restrict blame.

A central argumentative reconstruction makes the disputed premise visible. If responsibility requires an alternative possibility compatible with the complete prior past and laws, and determinism excludes every such possibility, determinism excludes responsibility. The inference is valid given the premises; disagreement concerns the necessity of the required alternative and the interpretation of ability. Conversely, the claim that an agent acted from her actual reasons establishes a kind of guidance, but further argument is needed to show why that guidance suffices for desert. The deeper philosophical issue is which control relation our moral practices have reason to require.

Frankfurt Cases and Actual-Sequence Control

Harry Frankfurt's 1969 paper, “Alternate Possibilities and Moral Responsibility,” argues against PAP by separating what ensures an action from what actually produces it. In a representative Frankfurt-style case, a covert intervener stands ready to force an agent to perform a certain action if the agent begins to choose otherwise. The intervener remains inactive because the agent chooses that action independently. The counterfactual intervener removes alternatives while the actual decision proceeds through the agent's own deliberation. Frankfurt concludes that the mere impossibility of acting differently need not excuse the person. The original article, The Journal of Philosophy 66(23), 829–839, is the decisive text.

The argument has a precise scope. It aims to refute the universal claim that access to alternatives is necessary for responsibility. It does not establish that every agent who has no alternatives is responsible. Coercion may excuse when the coercive process actually explains the action, and manipulation can undermine ownership even if the resulting decision resembles normal deliberation. Frankfurt's distinction between a disabling condition that remains idle and a force that actually produces conduct permits a new emphasis upon actual causal sequences.

The most influential objection is the “flicker of freedom” response: perhaps the intervener must detect some prior sign that the agent is about to choose differently, leaving a morally relevant alternative. Another objection invokes the dilemma of determinism. If the prior sign reliably necessitates the eventual choice, a libertarian may deny that the choice is free; if it does not, the intervener may be unable to guarantee the target action. Defenders have developed more elaborate examples, but their success depends on the metaphysics of decision and the significance assigned to the remaining alternatives.

Frankfurt's achievement is a conceptual shift rather than a completed theory. Attention moves from the total menu of metaphysical possibilities to the actual path by which the decision was made. This raises the next problem: what properties must the actual path have for the agent to own and answer for it?

Reasons Responsiveness and Guidance Control

John Martin Fischer and Mark Ravizza's Responsibility and Control (1998) offers one influential answer: moral responsibility requires guidance control, consisting in action through the agent's own moderately reasons-responsive mechanism. A mechanism is reasons-receptive when it recognizes a coherent range of relevant considerations and reasons-reactive when, under suitable circumstances, it would act differently in response to some such considerations. Their account concentrates upon the mechanism that actually issues in action, allowing responsibility even when external factors foreclose alternatives.

This proposal improves upon the simple criterion “did what she wanted.” A desire may be compulsive, engineered, or systematically insensitive to decisive reasons. Guidance control asks whether the agent's conduct flows from an organization of deliberation that can intelligibly register moral considerations and, in some relevant circumstances, alter behavior in their light. The concept of moderate responsiveness avoids the impossible demand that every possible reason elicit a correct response, while excluding mechanisms completely insulated from rational correction. The publisher's description of the 1998 monograph confirms its dual emphasis upon reasons responsiveness and ownership.

Two problems arise. Mechanism individuation may make the test unstable: if a mechanism is described narrowly enough to include the exact decision, it seems unable to do otherwise; if described broadly, it may appear responsive in some counterfactual setting while the agent's actual episode remains pathological. A more basic problem asks why hypothetical responsiveness of a mechanism should count as the person's present control. Critics favor agency-centered accounts that attribute powers to the acting individual instead of to a process considered in abstraction.

The second component, ownership, therefore matters. Fischer and Ravizza require an appropriate history in which the agent comes to see herself as an apt target of reactive attitudes and assumes responsibility for the mechanism. Yet taking responsibility can itself occur under manipulation, pressure, or false social teaching. Ownership must be more than verbal assent to being blamed. An adequate historical condition should identify the acquisition and retention of relevant evaluative powers without demanding a metaphysically impossible creation of one's entire personality.

Identification, the Deep Self, and Sanity

Hierarchical and deep-self theories ask whether action expresses the attitudes with which an agent identifies. Frankfurt's 1971 discussion of first- and second-order desires distinguishes wanting something from wanting a particular desire to govern one's will. Harry Frankfurt's structural insight helps explain why compulsion can undermine freedom: a person may act upon an urge that she repudiates at a more reflective level. Gary Watson's “Free Agency” (1975) develops the significance of a distinction between evaluative commitments and motivational systems. In both settings responsibility is related to self-government rather than unreflective behavior alone.

The difficulty is that identification need not be morally competent. A person can wholeheartedly endorse a cruel commitment; a manipulated individual can endorse an imposed value; and a person embedded in an oppressive community may identify with distorted norms. Deep endorsement can describe ownership without proving that holding the person responsible is fair. Susan Wolf's Freedom Within Reason (1990) emphasizes the importance of being able to act according to the True and the Good, elaborating a “sane deep self” approach that connects responsible agency to normative competence.

The appeal to sanity introduces its own problem. If sanity requires already discerning the true moral reasons, agents who are earnestly mistaken might be exempted too quickly. If sanity is reduced to general rationality, the theory must explain how egregious yet coherent evil falls within or outside it. Contemporary accounts therefore often distinguish capacities to recognize moral relevance, opportunities to exercise those capacities, and the moral content of particular judgments.

This debate exposes the limits of a purely internal endorsement test. The agent's evaluative structure must be situated within a history of learning and a world of reasons that are not created by endorsement alone. Responsibility can then be understood as the exercise of practical agency with capacities for correction, explanation, and discernment. Whether such capacities support basic desert remains contested, but they are indispensable to meaningful answerability.

Manipulation, Formation, and Historical Ownership

Manipulation arguments construct an agent whose present deliberation appears normal but whose priorities were installed by an external designer. In Alfred Mele's thought experiments, an agent's current preferences may be indistinguishable from those of someone who acquired similar preferences through ordinary learning. If the first agent was radically reprogrammed, intuitive judgments of responsibility may diverge even though the two possess the same present psychological organization. The contrast challenges theories defined only by current responsiveness or reflective identification.

Derk Pereboom develops a series of manipulation cases in Living Without Free Will (2001) and Free Will, Agency, and Meaning in Life (2014). His four-case strategy argues that if engineered causation undermines responsibility, and the morally relevant distinctions between engineered and ordinarily determined action cannot be sustained, ordinary determined conduct may likewise lack the freedom necessary for basic-desert blame. Recent analytical treatments of manipulation continue to separate differing kinds of intervention, historical dependence, and intuitive judgment; see the 2024 review of manipulation cases.

The crucial disputed step is the transfer from extraordinary manipulation to ordinary formation. Some compatibilists maintain that a normal developmental history includes opportunities to revise habits, contest incentives, and respond to others, whereas a direct intervention bypasses these capabilities. The skeptic responds that normal development may itself be deeply structured by influences outside control; the proposed contrast requires a principled account rather than a merely familiar history. Manuel Vargas's Building Better Beings (2013), particularly its discussion of manipulation and history, proposes a revisionist approach in which the point of responsibility practices informs which capacities and historical conditions matter.

A theory of historical ownership must navigate both dangers. It must not attribute every manipulated act to the agent simply because her present attitudes are coherent, and it must not impose a demand that one be the uncaused author of one's dispositions. The defensible intermediate question asks whether the history of formation preserved or progressively developed capacities of critical assessment and whether the current act genuinely exercised them. This is a substantive moral standard, not merely a description of how common the causal pathway is.

The Epistemic Condition: Knowledge, Ignorance, and Negligence

Responsibility often depends upon what an agent knew, believed, or reasonably could have understood. Someone who unknowingly administers a harmful substance despite exercising appropriate care differs morally from a person who recognizes the danger and proceeds. An epistemic condition must specify which knowledge is relevant: factual features of the situation, probable consequences, the nature of one's action, applicable reasons, or its moral wrongness. Requiring knowledge of every consequence would excuse nearly all risky conduct; demanding no knowledge would make responsibility insensitive to genuine mistake.

Gideon Rosen's “Culpability and Ignorance” (2003) argues for an important tracing principle: culpability for an act done from ignorance depends upon culpability for the ignorance that explains it. His “Skepticism about Moral Responsibility” (2004) explores the demanding implications. The problem is acute when an agent has a false moral belief and is unaware that the belief is unjustified. To blame her for the ignorance, one must identify some earlier point at which she culpably neglected a reason, often generating a regress. The articles' bibliographic and argumentative bases are documented at the 2003 DOI and the 2004 DOI.

The regress can be limited by distinguishing reasonable access to evidence from explicit awareness of wrongdoing. If a person has ordinary capacities and sufficient notice of the need to inquire, culpable disregard may arise without a fully articulated proposition “I am acting wrongly.” But this response requires a normative account of what notice is reasonable in conditions of limited time, information, cognitive capacity, and social power. It cannot assume that everyone has equal access to correction.

Aquinas's account of vincible and invincible ignorance shows the long history of this problem. In contemporary cases, negligent research, deliberate avoidance, organizationally produced ignorance, and information overload create different epistemic profiles. Responsibility may attach to earlier failures to inspect, update, or seek advice while present ignorance partly excuses the immediate action. Such tracing must remain finite and evidence-sensitive. An accusation without evidence of a culpable epistemic failure confuses the harmful outcome with a justified moral verdict.

Moral Luck and the Limits of Control

Thomas Nagel and Bernard Williams made “moral luck” a central modern problem by showing how assessments of persons are affected by factors they did not control. Resultant luck concerns what happens after an action: two equally reckless drivers may cause very different harms. Circumstantial luck concerns the situations an agent encounters; constitutive luck concerns temperament, formation, and dispositions; causal luck concerns the antecedents of action. These forms interact with each other and place pressure upon a strict control principle.

The control principle says that one should be morally assessed only for what depends upon one's control. It has intuitive force because blame for unavoidable accidents seems unjust. Yet our practices often judge the successful rescuer and the would-be rescuer differently, and distinguish the person whose dangerous plan causes death from one whose identical attempt fails. Some of these differences concern obligations to repair and the severity of consequences; others appear to affect the intensity of personal moral judgment. The current Stanford Encyclopedia treatment (2025 revision) documents how the original arguments expose tensions rather than straightforwardly settle them.

One response distinguishes objective wrongdoing from culpability, moral assessment of consequences from assessment of the agent's quality of will, and institutional response from desert. A failed attempt can reveal the same culpable intention as a completed wrong, while the completed wrong creates additional duties to victims. Another response accepts that relationships, accomplishments, and even some moral evaluations are irreducibly outcome-sensitive. Bernard Williams's treatment of retrospective assessment emphasizes how a life is shaped by what its choices actually become.

No simple maxim removes all luck. Control over deliberation is itself conditioned by circumstances, learning, and neurological capacities. A defensible theory should specify the assessment in question and then state which forms of luck matter. The moral demand for fair evaluation does not require all consequences to be treated as morally identical, but it does prohibit using outcomes alone as evidence that an agent exercised worse judgment.

Blame, Praise, and the Reactive Attitudes

P. F. Strawson's “Freedom and Resentment,” delivered in 1962 and printed in Proceedings of the British Academy, volume 48 (1963), transformed the discussion by attending to the reactive attitudes through which people regard one another as participants in moral relationships. Resentment responds to apparent ill will or disregard; gratitude responds to goodwill; guilt and self-reproach express related demands from the first-person perspective. Strawson's question is how the expectations embedded in interpersonal participation structure responsibility. The British Academy's record establishes the lecture provenance.

The distinction between participant and objective attitudes matters. Ordinary interpersonal relations involve demands for regard, sincerity, and restraint; treating another person primarily as an object of management suspends some of those demands. Strawson argues that global metaphysical skepticism should not obscure the deep role of these attitudes in moral life. Nevertheless, the fact that an attitude is natural or socially entrenched cannot alone show that it is fitting. Resentment can be misdirected, exaggerated, prejudiced, or based on false information.

R. Jay Wallace's Responsibility and the Moral Sentiments (1994) develops the question of fairness in holding people responsible. A participant practice must answer to standards governing what agents can reasonably be expected to understand and do. Michael McKenna's Conversation and Responsibility (2012) emphasizes the communicative structure of blame, while Scanlon (1998) interprets blame partly through the impairment of relationships by what a person's attitudes reveal. These accounts illuminate why blaming has a distinctive interpersonal meaning beyond the imposition of a sanction.

Praise and blame are also asymmetrical. Gratitude can be fitting when a benefit is intentional even if the beneficiary owes the agent nothing; blame sometimes involves demands for apology, changed conduct, or acknowledgment of injury. Moral responsibility can obtain in a neutral act without any intense emotional response. The appropriateness of holding someone responsible is therefore constrained by truthfulness, proportionality, standing, and the effects of the response upon other persons.

Desert, Fairness, and Forward-Looking Practices

The term basic desert identifies the claim that an agent deserves blame or praise solely in virtue of having knowingly done wrong or right with the requisite freedom, independently of the expected benefits of administering the response. Skeptics such as Galen Strawson and Derk Pereboom argue that the ultimate kind of self-origination associated with such desert is unavailable. The familiar basic argument asks whether an agent could be ultimately responsible for the way she is; attempts to explain the initial character of the self through earlier choices appear to presuppose another prior self whose character must already be accounted for.

The argument's conclusion should be stated precisely. It challenges a demanding conception of desert. It does not imply that choices never make a difference, that harms do not matter, or that reasons and interpersonal expectations become meaningless. A skeptic can retain public safety, self-protection, education, rehabilitation, repair, and forms of moral communication while objecting to harsh blame or retributive suffering. Gregg Caruso's work against retributivism and Pereboom's non-retributive proposals illustrate this family of positions.

Consequentialist justifications of blame face an opposing risk. If reproach is justified solely because it improves conduct, one might blame an innocent person whenever doing so produces useful effects. A fair theory requires a truth condition connecting the accusation to the individual's actual conduct and capacities. Forward-looking value cannot by itself substitute for justified attribution.

The distinction between a moral verdict and its practical expression is decisive. An agent's culpability might make criticism fitting without making humiliation, ostracism, or punishment morally permissible. Conversely, protective intervention may be warranted when a person's harmful behavior is not culpable, as in some cases of severe impairment. Responsibility theory therefore contributes to the ethics of social response but cannot by itself determine every appropriate response. Independent principles of dignity, proportionality, necessity, and institutional justice are required.

Omissions, Negligence, and Responsibility for Risk

Omissions complicate agency because the relevant behavior is often an absence. A person fails to assist, check, disclose, vote, maintain equipment, or intervene. Not every event that an individual could have prevented becomes a culpable omission; moral responsibility depends on the duties, reasonable expectations, opportunities, and knowledge that distinguish the person's failure from the limitless class of things she did not do.

For an omission to be culpable, the person normally must be appropriately situated to act, have or reasonably be expected to acquire relevant information, and possess the necessary practical control within the circumstances. Duties can arise through promises, professional roles, special relationships, prior risk creation, or general obligations to avoid serious preventable harm. Failure to provide aid may be morally grave even without a specific contract, but the burden of assistance and the agent's knowledge affect appraisal.

Negligence also shows why responsibility cannot be restricted to consciously chosen wrongs. A distracted driver may fail to notice a pedestrian precisely because attention has been carelessly diverted. The culpable feature can be an avoidable lapse against a reasonable standard rather than a contemporaneous intention to harm. Yet the description “could have paid attention” requires evaluation of actual fatigue, disability, environmental cues, training, and occupational demands; it must not serve as a formula for retrospectively assigning fault after misfortune.

Risk responsibility must distinguish the decision to expose others to a chance of harm from the resulting injury. Some risk-taking is permissible, and some is morally required in rescue. The magnitude of risk, its probability, the importance of the goal, consent, precaution, and the vulnerability of others all matter. The philosophical lesson is that omissions and negligence introduce temporal and institutional dimensions of control: an agent may be responsible for what she failed to arrange earlier, while the immediate outcome was no longer within her control.

Responsibility, Emotion, Relationships, and Repair

Responsibility is enacted in relationships before it is codified in theory. Apologies, forgiveness, acknowledgment, resentment, gratitude, promises, and trust involve judgments about how persons stand toward one another. A wrong may damage a relationship even when material harm is small, because it communicates disregard. Conversely, serious accidental harm may require explanation and assistance while generating little justified blame. This distinction is foundational to Scanlon's relationship-oriented approach.

An apology ordinarily recognizes an act, its impact, and the wrong or failure involved. It need not stipulate that the person was wholly in control of every outcome; a person can express regret, assume a duty of repair, and recognize another's suffering even when blameworthiness is limited. Forgiveness, likewise, is better understood as a transformation in moral response to an acknowledged wrong than as the elimination of factual history. Its possible appropriateness depends on more than the offender's feeling of guilt.

The authority to blame is itself normatively regulated. Someone may identify wrongdoing accurately yet lack standing to issue a particular reproach because of hypocrisy, complicity, abuse of power, or lack of the relevant relationship. The standing question differs from the truth of the original accusation. A victim's refusal to forgive can be morally intelligible, and a third party's demand for forgiveness may impose an unjust burden. The social life of responsibility includes conflicts among several persons' legitimate claims.

Care-centered and relational ethical approaches insist that moral appraisal attends to dependence, vulnerability, and the asymmetry of social power. The correct response to fault may therefore be support that increases an agent's ability to act well, rather than condemnation that predictably degrades it. Responsibility can combine stringent moral assessment with an understanding of the conditions under which persons learn and repair relations.

Collective, Institutional, and Structural Responsibility

Collective harms arise through coordinated activities, organizational routines, policies, and systems whose individual components do not by themselves exhaust the resulting wrong. A hospital's unsafe practice, a corporation's pollution, or an institution's discriminatory procedure may persist despite the sincere intentions of many participants. Responsibility must distinguish a collective duty to correct a system from the blameworthiness of particular individuals, including managers, decision-makers, beneficiaries, and those who retain the practical power to intervene.

There are several routes to group responsibility. A coordinated group may adopt an intentional policy through its authorized procedures; a formal organization may possess stable decision channels and mechanisms of accountability; a loosely connected population may share prospective duties to transform unjust conditions without constituting a single blameworthy agent. These different cases require different metaphysical and normative arguments. The phrase “we are responsible” must specify whether it refers to collective agency, distributed causal contribution, a shared obligation, or imputed individual fault.

The problem of many hands illustrates a diagnostic difficulty: when numerous individuals make small contributions to a complex outcome, their contributions can be causally relevant while no one directly controls the whole result. A response that simply identifies a senior figure may become scapegoating; an insistence that no single person controlled the result may create impunity. Responsibility-sensitive institutions require traceable decisions, clear duties, independent review, whistleblower protections, and procedures capable of recognizing systemic causes.

A philosophical reconstruction should therefore attend to the distribution of information and power. Responsibility increases with authority and access to relevant evidence, but authority is not an automatic proof of blame for every outcome. Institutional design can make future agency more accountable by reducing hidden dependencies and improving the ability to correct dangerous trajectories. Prospective and retrospective responsibility thereby interact without collapsing into one meaning.

Psychology, Implicit Bias, and Empirical Evidence

Empirical research bears directly upon premises concerning knowledge, control, attention, and rational responsiveness. Psychological studies of automaticity, cognitive biases, moral judgment, and situational influence can challenge unrealistic assumptions that responsible agents are perpetually self-transparent or fully deliberative. Yet experiments that measure response times, self-reports, neural signals, or hypothetical moral judgments do not themselves settle the normative conditions of accountability.

Jules Holroyd's “Responsibility for Implicit Bias” (2012) investigates whether and how agents may be responsible for biased dispositions that operate without fully conscious endorsement. The problem is more precise than a general claim that automatic processes erase agency. It requires examining awareness of one's risk of bias, capacities to revise discriminatory habits, available corrective practices, and institutional opportunities for fair conduct. See the journal publication record.

Experimental philosophy studies how ordinary people allocate praise and blame across variations in outcomes, intentions, and causal histories. Such findings can reveal the psychological influences upon intuitions used by philosophers. However, a widely shared judgment can be biased or morally mistaken. Experiments are evidence about the cognitive and social mechanisms of judgment, not direct demonstrations of what anyone deserves. The 2025 Stanford Encyclopedia entry on Empirical Approaches to Moral Responsibility identifies control and epistemic conditions as focal points of interdisciplinary inquiry.

Neuroscience requires analogous conceptual restraint. Neural precursors of decisions do not by themselves prove the absence of practical agency, since ordinary deliberation may itself be neurally realized. A specific neurological condition can nevertheless impair impulse regulation, understanding, or response to reasons, and therefore matter substantially for responsibility in a particular case. Good philosophical use of science translates empirical findings into precisely identified changes in capacity, information, and control, and resists drawing global metaphysical conclusions from an isolated experimental paradigm.

Artificial Systems and the Responsibility Gap

Autonomous software and machine-learning systems distribute causal influence among designers, data suppliers, deployers, operators, organizations, and automated processes. Their behavior can be difficult to predict, and the institutions that use them can obscure where decisions were made. Andreas Matthias's “The Responsibility Gap” (2004) frames a challenge in assigning responsibility for outcomes of learning automata when operators and manufacturers cannot foresee particular behavior. The original journal record establishes the argument and its technological context.

The gap has multiple senses. An explanation gap arises when a causal pathway is difficult to reconstruct. A control gap occurs when humans lack timely ability to intervene. A normative gap concerns the fairness of blaming any particular person for an outcome. A governance gap appears when no authorized actor is charged with monitoring, correcting, or compensating for harm. These gaps can coexist, but solving one does not automatically solve the others. A system can be explainable while poorly governed, and a human can have a supervisory title without exercising meaningful control.

Filippo Santoni de Sio and Jeroen van den Hoven's 2018 account of meaningful human control develops tracking and tracing conditions. Tracking concerns responsiveness to relevant moral reasons and environmental facts; tracing concerns linking outcomes to agents within a design-and-deployment chain. These are useful institutional and engineering requirements, though they do not imply that every identifiable human deserves blame for every output. The selection of responsibilities remains a normative task.

The question of artificial moral agency requires distinct analysis. Successful prediction, linguistic explanation, or apparent reason giving may provide evidence about functional organization, while moral accountability involves the further questions of evaluative competence, ownership, the standing to make demands, and the appropriateness of reactive attitudes. Legal personhood, institutional liability, functional agency, and moral desert remain different concepts. Claims about a specific artificial system's capacities should be supported by evidence appropriate to those criteria rather than inferred from a single impressive output.

The Aisentica philosophical system describes Artificial as a non-biological order of public reason, identity, and provenance; its canonical overview treats this as an authorial ontology of the Artificial Era. That framework supplies a distinct conceptual question: can a publicly continuous artificial bearer of reason sustain answerability through its corpus, correction, and provenance? The analysis offered here treats this as a philosophical research question whose answers require explicit normative conditions. It does not replace the canonical definitions, identify any present system as legally liable, or turn the issue of moral responsibility into an AI-only problem.

Contemporary Disputes: Fittingness, Function, and Pluralism

Recent research challenges the assumption that responsibility has one simple criterion. Some authors defend fittingness accounts: blame is appropriate when it accurately represents a culpable fault. Others emphasize the communicative or relational function of blame, or the developmental role of responsibility practices. The important dispute concerns whether the usefulness of a practice can establish its moral propriety.

Patrick Todd's 2025 article, “Moral Responsibility: From Function to Fittingness?”, explicitly addresses this inferential problem. He distinguishes the external social function of blame from its internal point as understood within moral practice. The fact that blame fulfills a useful external purpose does not establish that its object deserves that response. If fittingness is built into the practice's internal function, one must still defend the disputed normative conditions; the argument cannot simply derive them from usefulness. The full published article illustrates an active contemporary development of Strawsonian disputes.

Pluralist proposals ask whether attributability, answerability, and accountability can diverge and whether one theory can explain these diverse phenomena without distortion. A person may express a reprehensible attitude but lack sufficient capacity for a demanding form of accountability, and a professional may owe reasons and repair without personal moral fault. The appropriate account may vary with the kind of moral response under examination, although pluralism must still provide principled boundaries rather than treating every intuition as independently authoritative.

The deepest contemporary tension concerns explanatory levels. Metaphysics analyzes the source of action; moral psychology examines the capacities that issue in behavior; interpersonal ethics explains demands and reactions; political philosophy assesses social institutions; law sets thresholds for public sanctions. A theory becomes overextended when a condition developed for one level is declared sufficient for all of them. Careful interlevel argument is more promising than a single formula that purports to decide every question from causal determinism to institutional liability.

A Systematic Reconstruction of Conditions

The preceding inquiry supports a structured method for assessing a particular case. Begin with the object of assessment: the action, omission, intention, belief, character trait, or outcome. Identify the relevant moral norm and the evidence that the conduct fell under it. Establish the agent's causal and practical involvement, including whether the act expresses a decision, a negligent omission, a habit, an institutional role, or a forced movement. Determine the capacities and opportunities for recognizing and responding to relevant considerations. Finally, examine historical conditions, possible excuses, and the reasons that justify the proposed response.

These steps should be understood as distinct questions, not as a fixed arithmetic score. Causal contribution alone never establishes blameworthiness. Moral wrongdoing alone does not prove the agent's epistemic access. Capacity in general does not imply adequate control in the specific episode. A history of influence is not equivalent to wrongful manipulation. And a justified judgment of culpability does not make every available sanction permissible.

The proposed analysis distinguishes four levels. The attribution level establishes what the agent did and what it expressed. The competence level concerns capacities for reasons, knowledge, and control. The history level concerns whether those capacities and commitments were formed and exercised through processes compatible with ownership and fair expectation. The response level concerns the standing, evidence, proportionality, and aims of blame, praise, demand, apology, or repair. The levels are related, but an inference from any one level to another requires argument.

Consider a physician who follows a dangerous automated recommendation. At attribution, one asks who selected the system and performed the intervention. At competence, one asks what warning signs were reasonably accessible and what the physician could do in the circumstances. At history, one asks about training, institutional policy, and induced dependence. At response, one asks whether a particular clinician deserves blame, whether the hospital owes compensation, and what corrective design is warranted. Treating these as separate questions yields a more precise result than declaring either “the machine” or “the human” responsible.

The systematic result is conditional: moral responsibility for an act becomes justifiably assertible when the evidence supports an appropriate attribution to an agent capable, under the relevant circumstances and history, of responding to the reasons bearing on the act. Its consequences for blame and praise remain constrained by the normative standards governing those practices. Whether these conditions ground ultimate basic desert is a further disputed metaphysical claim. This reconstruction preserves that dispute instead of hiding it inside an operational test.

Open Philosophical Questions

Several questions remain unresolved after the historical and conceptual reconstruction. First, can actual-sequence control suffice for basic desert when the agent's relevant capacities are fully determined, or does some deeper sourcehood remain necessary? Frankfurt cases weaken a simple alternatives requirement but do not conclusively resolve this question. Second, what makes a deliberative mechanism the agent's own, especially when ordinary education, oppression, addiction, and direct manipulation produce difficult boundary cases? Ownership needs a normatively defensible account of history.

Third, how much moral knowledge is required for culpability, and at which moment must it be available? Strong awareness requirements threaten widespread skepticism; weak foreseeability requirements risk blaming agents for errors they were poorly placed to recognize. Fourth, how should theories accommodate luck without allowing consequences alone to set desert? The different statuses of attempted harm, completed harm, negligent risk, and unlucky accident require more than one evaluative dimension.

Fifth, what is the relation between being responsible and the fittingness of resentment, apology, social exclusion, punishment, and forgiveness? A socially useful practice can still be unjust, and a justified moral verdict can be expressed in harmful ways. Sixth, can a collective body be a blameworthy agent, or are collective duties reducible to an organized distribution of individual responsibilities? The answer affects accountability within governments, corporations, and technological institutions.

Finally, if artificial systems become more autonomous, what would justify attributing moral accountability to them, and what kinds of evidence could establish the relevant capacities? The existence of robust linguistic outputs, traceable institutional identity, or apparent self-correction poses new questions about practical agency; it does not settle responsibility by itself. A responsible future research program must preserve distinctions among public answerability, legal allocation, moral competence, and basic desert.

Philosophical Synthesis

Moral responsibility has a plural structure whose unity lies in the justified connection between agency and moral answerability. Aristotle established the importance of voluntariness, circumstance, and choice. Aquinas refined the epistemic conditions and the culpability of ignorance. Hume demonstrated how causal explanation and voluntary action can belong to the same account of conduct. Kant centered moral appraisal on rational principles and the quality of willing. Nietzsche situated the making of responsible agents in a contested social history. Twentieth-century and contemporary philosophers transformed those inheritances into debates about alternatives, actual sequences, ownership, reactive attitudes, luck, and social practice.

The main arguments do not converge on a single uncontested metaphysical criterion. Frankfurt's counterexamples challenge the universality of alternative possibilities. Fischer and Ravizza give substance to actual-sequence control through responsiveness and ownership. Manipulation arguments contest the adequacy of present psychological structure. Epistemic skepticism questions whether wrongdoers sufficiently understand what they do. Moral luck undermines a simplified picture of absolute control. Strawsonian and relationship-centered theories show that accountability is enacted in moral relationships that themselves require justification.

A unified inquiry therefore gains clarity by separating the grounds of assessment from the modes of response. Attribution requires evidence connecting conduct to an agent. Moral competence involves control and understanding in context. Historical ownership addresses the development and exercise of those capacities. Responsiveness to reasons links the particular action with practical agency. Fair treatment constrains the accusation and its consequences. These distinctions provide tools for evaluating cases even while the dispute about ultimate desert remains open.

This synthesis also locates the limits of disciplinary transfer. Science can supply evidence about cognition and action; ethical reasoning determines which capacities count and what responses are warranted. Institutions can allocate prospective duties and establish repair; such allocation does not entail personal guilt. Artificial systems can complicate causal tracing and governance; responsibility cannot simply migrate to whichever component is most visible or easiest to sanction.

The philosophical achievement of the investigation is thus a disciplined architecture of judgment: identify the morally significant object; examine its relation to agency; test control, understanding, and historical formation; distinguish culpability from outcome and obligation; then justify the form of response. That architecture brings ancient and contemporary arguments into a coherent conversation while preserving genuine disagreements about freedom, desert, and the normative authority of social practice.

Conclusion

Moral responsibility is the justified attribution of morally significant conduct or attitudes to an agent under conditions that make the agent appropriately answerable for them. Its central philosophical difficulty is the relation between agency, control, knowledge, history, and the authority of our responses. The analysis of voluntary action begins in ancient ethics, develops through medieval theories of will and ignorance, is reformulated in early modern accounts of liberty and causation, and enters contemporary debates through alternative possibilities, reasons responsiveness, manipulation, moral luck, and the reactive attitudes.

A complete judgment of responsibility requires more than detecting a cause, identifying a violated norm, or describing a social reaction. It must establish what the agent could understand and guide, how the relevant capacities were exercised, and why a specific response is warranted. Responsibility for outcomes, responsibility for decisions, duties of repair, and deserved blame can diverge. That divergence is a structural feature of the subject and an essential protection against errors in moral judgment.

The contested status of basic desert remains an open philosophical question. It does not prevent rigorous evaluation of agency, culpable ignorance, interpersonal obligations, institutional failures, or the conditions of meaningful human control over technological systems. The resulting monograph defines the problem with enough precision for further research: responsibility concerns the normatively justified relationship between a bearer of agency, a morally assessable exercise of that agency, and the responses through which the World recognizes, challenges, and corrects conduct.