The Neurobiology of Moral Judgment (Trolley Problem fMRI) – Joshua Greene

For millennia, the inquiry into the architecture of moral judgment was the exclusive domain of speculative philosophy. From Aristotle’s virtue ethics and the divine command frameworks of medieval scholasticism to the Enlightenment clash between Immanuel Kant’s deontological imperative and Jeremy Bentham’s utilitarian calculus, the question of how human beings discern right from wrong was addressed through introspection, dialectical argumentation, and normative theorizing. Ethicists operated under the working assumption that moral convictions were the output of domain-specific rational deliberation—deliberations that could be systematically dissected using logical principles, conceptual coherence, and normative consistency. However, the dawn of cognitive neuroscience and functional neuroimaging in the late twentieth century shattered this insular philosophical tradition, forcing an empirical reckoning with the biological machinery that generates, constrains, and colors human ethical evaluations.

At the epicenter of this empirical revolution stands Joshua Greene and his pioneering functional magnetic resonance imaging (fMRI) investigations of the classic “trolley problem.” Conceived originally by Philippa Foot and expanded by Judith Jarvis Thomson, the trolley problem was designed as an abstract thought experiment to illuminate the subtle, often paradoxical contours of ethical theory—specifically, why diverting a runaway trolley to save five people at the cost of one appears morally permissible, whereas pushing a single individual into the path of the train to achieve the identical numerical outcome strikes most observers as an abhorrent moral transgression. Where normative philosophers perceived an exquisite problem of moral metaphysics, Greene discerned a neurobiological clue: a window into the evolutionarily stratified, competing neural circuits that govern human decision-making under conditions of conflict.

Greene’s seminal 2001 study in Science, followed by his 2004 investigation in Neuron, introduced the Dual-Process Theory of Moral Judgment, fundamentally destabilizing the classical boundaries between descriptive neurobiology and prescriptive meta-ethics. By demonstrating that deontological intuitions (rights-based constraints against direct harm) are predominantly driven by rapid, phylogenetically ancient affective circuits within the limbic and paralimbic systems, while utilitarian evaluations (consequentialist maximization of aggregate welfare) recruit the metabolically expensive, phylogenetically recent frontoparietal cognitive control networks, Greene challenged the normative authority of moral intuition itself. What follows is an exhaustive exploration of the neurobiology of moral judgment, tracing the trajectory from armchair philosophy to high-field neuroimaging, unpacking the neural substrates of personal versus impersonal harm, evaluating convergent lesion and neurochemical evidence, and critically examining the meta-ethical, legal, and algorithmic consequences of viewing the human conscience through the lens of cognitive neuroscience.

1. Historical and Philosophical Foundations of the Trolley Problem

1.1 Philippa Foot’s Original Formulation of the Trolley Dilemma

The trolley dilemma was introduced into contemporary moral philosophy by Philippa Foot in her foundational 1967 paper, “The Problem of Abortion and the Doctrine of the Double Effect,” published in the Oxford Review. Foot was not seeking to establish an esoteric puzzle for its own sake; rather, she was interrogating the boundaries of legitimate moral justification regarding medical ethics, specifically whether terminating a pregnancy could ever be permissible to save the life of the mother. To isolate the underlying ethical principles, Foot devised an analogy involving a runaway tramway car:

“Suppose that a judge or magistrate is faced with rioters demanding that a culprit be found for a certain crime and threatening otherwise to take their own bloody revenge on a particular section of the community… At the same time, let us imagine another case in which a driver whose runaway tram cannot be stopped can steer it onto a side-track, where it will hit and kill one man, rather than continuing along the main track where it will hit and kill five.”

Foot utilized this scenario to dissect the Doctrine of Double Effect (DDE), an intellectual framework originating in the jurisprudence of Thomas Aquinas. The DDE asserts that there is a profound moral distinction between an action whose foreseen bad consequence is intended as a means or an end, and an action where the bad consequence is merely an unintended, though foreseen, side effect of pursuing a proportionately good outcome. In Foot’s analysis, the driver of the runaway trolley does not intend the death of the single workman on the spur track; rather, the workman’s death is a tragic, collateral side effect of steering the vehicle away from five people.

Crucially, Foot grounded her analysis in the distinction between negative duties and positive duties. A negative duty is an obligation to refrain from inflicting harm, injury, or death upon an innocent individual. A positive duty, by contrast, is an obligation to provide aid, sustenance, or rescue to someone in peril. Traditional moral intuition and legal codes generally assign greater stringency to negative duties than to positive ones: refraining from murder is an absolute prohibition, whereas the duty to save a drowning stranger is often conditioned by proximity, feasibility, and cost. In the classic switch case (subsequently formalized in the philosophical literature as Bystander at the Switch), Foot argued that the driver is caught between two conflicting negative duties: the duty not to injure the five versus the duty not to injure the one. When forced to navigate a zero-sum collision of negative duties, Foot concluded that moral agents intuitively and rationally default to minimizing the net quantum of injury, rendering the diversion of the trolley permissible, if not obligatory.

1.2 Judith Jarvis Thomson and the Footbridge Variant

While Philippa Foot provided the initial catalyst, it was Judith Jarvis Thomson who transformed the trolley problem into a diagnostic instrument for testing the consistency of moral systems. In her landmark papers “Killing, Letting Die, and the Trolley Problem” (1976) and “The Trolley Problem” (1985), Thomson identified a crucial vulnerability in Foot’s reliance on the negative-versus-positive duty dichotomy. To expose this conceptual tension, Thomson introduced several radical variations, the most famous of which is the Footbridge Dilemma.

In the Footbridge scenario, an agent stands on an overpass above the tracks, observing a runaway trolley hurtling toward five unsuspecting workers. Standing next to the agent is a large, heavy stranger. The only way to arrest the progress of the trolley before it obliterates the five workers is to push this stranger off the footbridge and onto the tracks below. His physical mass will bring the train to a halt, killing him instantly, but saving the five workers on the line ahead. Structurally, the mathematical balance sheet of the Footbridge case is identical to the Switch case: one life is sacrificed to preserve five lives. Yet, when presented with this dilemma, the vast majority of human subjects exhibit a stark moral asymmetry: they judge turning the switch in the trolley dilemma to be morally permissible (often with endorsement rates exceeding 85%), but they condemn pushing the man off the footbridge as an egregious moral atrocity (with rejection rates routinely hovering between 80% and 90%).

This moral asymmetry presented a profound puzzle for normative philosophy. Why should two scenarios with identical consequential payoffs (net +4 lives preserved) elicit diametrically opposed moral intuitions? Thomson argued that the asymmetry could not be explained away by Foot’s framework of duties, since in both cases an innocent person’s negative right not to be killed is violated to preserve the lives of others. Instead, Thomson initially appealed to the Kantian principle of treating humanity never merely as a means, but always at the same time as an end. In the Footbridge case, the agent exploits the heavy man’s physical body as an indispensable mechanical buffer: his destruction is an essential causal instrument in arresting the train. If the stranger were to magically slip off the tracks before impact, the plan would fail and the five would die. In the Switch case, by contrast, the workman on the spur track is not utilized as a mechanical instrument; his presence is an unfortunate circumstance. If he were to successfully leap clear of the tracks, the five workers would still be saved. Pushing the stranger introduces a deliberate violation of bodily integrity where the victim is subordinated as a mere object of causal utility—a distinction that resonated deeply with normative ethicists, yet remained notoriously difficult to ground in a unified, objective moral theory.

1.3 Deontological Frameworks Versus Utilitarian Calculus

The philosophical stalemate generated by the trolley and footbridge dilemmas exposed the deep fault lines separating the two dominant paradigms of Western normative ethics: deontology and utilitarianism.

The deontological tradition, rooted in the moral philosophy of Immanuel Kant, posits that morality is governed by fundamental, categorical duties, rules, and rights that possess intrinsic normative weight, completely independent of the empirical consequences they yield. In the Groundwork of the Metaphysics of Morals (1785), Kant articulated the Categorical Imperative, whose second formulation—the Formula of Humanity—demands that moral agents treat rational beings never merely as a means to an end, but always as ends in themselves. Within modern deontology, philosophers like Robert Nozick conceptualized individual rights as inviolable “side-constraints” that erect an impenetrable moral boundary around the person, forbidding their sacrifice for the collective good. From a deontological perspective, pushing the stranger off the footbridge is an impermissible rights-violation: it tramples upon the agent’s fundamental right to life, treating his conscious existence as an expendable instrument for aggregate optimization.

Conversely, the utilitarian tradition, pioneered by Jeremy Bentham and systematically refined by John Stuart Mill, operates on an entirely consequentialist axis. Bentham’s fundamental axiom asserts that the measure of right and wrong is the greatest happiness of the greatest number, calculated through the aggregation of pleasure and the minimization of pain (the felicific calculus). Mill, while introducing qualitative distinctions among pleasures, maintained that actions are morally right in proportion as they tend to promote happiness, and wrong as they tend to produce the reverse of happiness. From a strict classical utilitarian stance, rights, duties, and intentions are merely heuristic rules of thumb. What matters normatively is the bottom-line optimization of aggregate welfare. In both the Switch and Footbridge scenarios, the consequential algebra is indistinguishable: 5 – 1 = +4 lives saved. Therefore, a consistent utilitarian must conclude that if it is permissible—or indeed mandatory—to divert the switch, it must be equally permissible to push the man off the footbridge, provided the epistemic probabilities of success are held equal.

Prior to the intervention of cognitive neuroscience, normative ethics found itself trapped in an epistemic deadlock. Utilitarians accused deontologists of engaging in irrational “rule worship” and cowardly squeamishness—allowing five people to die simply to keep one’s hands conceptually clean. Deontologists fired back, accusing utilitarians of a monstrous, technocratic moral blindness that viewed human beings as fungible units in a balance sheet, eroding the very concept of human dignity. Crucially, both philosophical camps claimed that their respective conclusions were grounded in pure rational intuition. Neither camp possessed the empirical diagnostic tools required to interrogate the psychological or neurobiological engines generating these conflicting intuitions. The debate remained an intractable battle of introspective convictions, waiting for an empirical framework capable of examining the cognitive machinery beneath the surface.

2. Joshua Greene’s Paradigm Shift: Bridging Moral Philosophy and Cognitive Neuroscience

2.1 The Departure from Armchair Normative Philosophy

In the late 1990s and early 2000s, Joshua Greene, then a doctoral student in philosophy at Princeton University working alongside cognitive psychologists and neuroscientists, recognized a profound epistemic pathology within traditional normative ethics: the uncritical reliance on introspective intuitions. Ethicists had spent decades constructing increasingly complex, baroque theoretical epicycles to reconcile their conflicting reactions to trolley dilemmas. Yet, these philosophers rarely paused to question whether these moral intuitions were the pristine outputs of an objective, normative reality, or merely the idiosyncratic evolutionary and neurobiological artifacts of the human primate brain.

Greene was heavily influenced by the psychological critique of rationalism championed by figures like Jonathan Haidt, who demonstrated that moral judgment is frequently driven by rapid, automatic emotional intuitions, followed by elaborate, post-hoc rationalizations. Haidt’s work on “moral dumbfounding”—wherein subjects stubbornly maintain that an act (such as consensual adult sibling incest using two forms of contraception) is morally wrong, despite being completely unable to articulate a coherent harm-based justification—suggested that conscious moral reasoning often functions not as an impartial judge discovering moral truth, but as a biased press secretary defending pre-existing affective convictions. Greene recognized that armchair philosophers constructing grand theories around Footbridge and Switch were potentially committing the same error: taking visceral, subcortical emotional reactions and retrofitting them with sophisticated, Kantian-sounding philosophical justifications.

To break this intellectual circularity, Greene argued that moral philosophy needed to be naturalized. Rather than asking which philosophical theory successfully unified our divergent intuitions, cognitive scientists needed to ask: What are the underlying neurocognitive mechanisms that produce divergent intuitions across structurally similar moral dilemmas? Greene sought to replace speculative metaphysics with descriptive cognitive neuroscience, transforming the trolley problem from an abstract ethical conundrum into a functional experimental probe of the living human brain.

2.2 The Central Research Question: The Origin of Divergent Intuitions

The core research question animating Greene’s early empirical work was deceptively straightforward yet methodologically profound: Why does diverting a runaway trolley via a switch feel profoundly different from pushing an individual off a footbridge, when both dilemmas present an identical arithmetic calculus of one life sacrificed to preserve five? If classical moral rationalism were correct, the human moral faculty should evaluate both scenarios through an integrated, domain-general computational architecture, weighing rights, consequences, and probabilistic outcomes through coherent cognitive principles.

Greene formulated the radical counter-hypothesis that the divergence in moral intuitions was not the product of a unified, high-level moral faculty applying subtle metaphysical distinctions, but rather the result of an architectural clash between distinct, neuroanatomically dissociable cognitive systems within the human brain. Specifically, Greene hypothesized that the Footbridge dilemma triggers a robust, automatic, evolutionarily primitive emotional response that screams “Wrong!”, whereas the Switch dilemma fails to recruit this affective alarm system to the same degree, allowing cold, deliberative cost-benefit reasoning to dominate the behavioral output.

To test this hypothesis, Greene turned to functional magnetic resonance imaging (fMRI). By measuring Blood-Oxygen-Level-Dependent (BOLD) contrast signals across the whole brain while human participants deliberated over standardized, computerized moral dilemmas, Greene and his collaborators could visualize which neural circuits were recruited in real time. If the difference between Switch and Footbridge was fundamentally driven by affective engagement, the fMRI data would reveal heightened metabolic activity in brain networks associated with emotion and social cognition during Footbridge trials, contrasted with elevated activity in executive cognitive control networks during Switch trials. The trolley problem was no longer just a thought experiment; it had become an experimental paradigm.

2.3 Formulating the Dual-Process Hypothesis in Moral Cognition

To provide a rigorous theoretical scaffold for his neuroimaging investigations, Greene synthesized insights from the emerging behavioral economics and cognitive psychology literature, most notably the dual-process models of cognition pioneered by Daniel Kahneman and Amos Tversky. Kahneman and Tversky had demonstrated that human judgment under uncertainty is governed by two radically distinct modes of information processing: System 1 (fast, effortless, associative, implicit, and often emotionally charged) and System 2 (slow, effortful, rule-governed, deliberate, and computationally demanding).

Greene adapted this taxonomy directly to the domain of moral philosophy, formulating the Dual-Process Theory of Moral Judgment. Under Greene’s model:

• Deontological judgments (e.g., rejecting the sacrifice of the man on the footbridge) are driven primarily by intuitive, automatic emotional responses generated by System 1 processes. These responses operate as visceral heuristic prohibitions against committing direct, proximate interpersonal violence.
• Utilitarian judgments (e.g., endorsing the diversion of the switch, or overriding emotion to push the stranger) are driven by controlled, effortful System 2 deliberation. These processes perform explicit mathematical cost-benefit analyses aimed at maximizing aggregate net welfare.

Crucially, Greene recognized that this dual-process architecture implied dynamic computational conflict. In cases where System 1 and System 2 outputs align (for instance, in everyday acts of altruism or wanton violence), moral decision-making is smooth and unambiguous. However, in cases like the Footbridge dilemma, the two systems produce radically antagonistic behavioral directives: System 1 demands absolute restraint (“Do not push!”), while System 2 calculates aggregate optimization (“Save five lives over one!”). This theoretical architecture generated explicit, falsifiable behavioral and neurobiological predictions: moral dilemmas that generate systemic conflict should show observable response-time latencies, prolonged cognitive interference, and simultaneous activation of competitive neural networks representing affective salience on the one hand and executive cognitive control on the other.

3. The Theoretical Architecture of Greene’s Dual-Process Model

3.1 The Dual-Process Taxonomy: System 1 Affect Versus System 2 Deliberation

The operational engine of Greene’s Dual-Process Model relies on a fundamental functional dichotomy within human cognitive architecture. To appreciate the predictive power of this taxonomy, one must examine how System 1 and System 2 process moral information across temporal, computational, and representational dimensions.

System 1 moral processing is intrinsically affective, pre-reflective, and ballistic. When an individual witnesses or mentally simulates an act of visceral, hands-on physical violence, the brain’s evolutionary alarm networks fire instantaneously, long before conscious deliberation can formulate a coherent conceptual argument. This affective surge possesses an inherently prescriptive character; it does not merely register descriptive information (“a body is falling”), but attaches an urgent, repulsive somatic tag to the contemplated action (“DO NOT DO THIS!”). Because these emotional responses are encapsulated and heuristic, they are computationally cheap and functionally rigid. They correspond directly to the classic hallmarks of deontological ethics: universalized, duty-based prohibitions, rights-based taboos, and non-negotiable moral constraints that brook no compromise or mathematical trade-offs.

System 2 moral processing, conversely, is propositional, reflective, and computational. It relies on the deliberate manipulation of abstract symbols, counterfactual simulations, and numerical values maintained within working memory. When evaluating a dilemma, System 2 abstracts away from the visceral, qualitative details of the event and constructs an algorithmic balance sheet of outcomes. It assesses the causal chain, calculates the expected utility of alternative interventions, and selects the path that yields the optimal aggregate outcome. Because System 2 requires conscious cognitive control, it is metabolically demanding, strictly capacity-limited, and vulnerable to cognitive load and temporal constraints. It corresponds to the hallmarks of utilitarianism: consequentialist calculation, impartiality, and the willingness to compromise specific deontological constraints to achieve a greater global good.

The interaction dynamic between these systems is one of mutual competition and competitive inhibition. The final behavioral output of a moral judgment is not determined by an integrated, harmonious consensus, but by a neurocomputational race: whether the deliberative cognitive control network can muster sufficient executive activation to actively override, inhibit, and veto the rapid emotional surge originating from the affective core.

3.2 The Camera Analogy: Automatic Settings Versus Manual Mode

To make the computational efficiency and evolutionary rationale of his dual-process architecture intuitive, Greene introduced an evocative metaphor: the dual-mode digital camera. A modern digital camera comes equipped with two complementary operational modes: “automatic settings” (point-and-shoot presets such as ‘portrait’, ‘landscape’, or ‘sports’) and “manual mode” (allowing custom calibration of ISO, shutter speed, and aperture).

The automatic settings are pre-programmed by camera engineers to optimize typical photographic scenarios with zero cognitive effort from the user. You point the lens at a face, the camera detects the subject, adjusts the focal length, fires the flash, and snaps an excellent photo within milliseconds. For standard, everyday conditions, the automatic settings are extraordinarily efficient, rapid, and accurate. However, they lack contextual flexibility. If you attempt to shoot a high-speed silhouette against a blinding sunset, the automatic settings will catastrophically misread the scene, producing an overexposed or blurred image. In such non-standard, complex conditions, the photographer must switch to “manual mode.” Manual mode requires technical expertise, time, and active calculation, but it provides the unconstrained flexibility necessary to tailor the machine’s parameters to rare, highly intricate, or novel visual environments.

Greene argues that human moral cognition is organized along precisely these design specifications:

• Automatic Settings: Our moral emotions, intuitions, and gut reactions are the brain’s evolutionary presets. They were shaped by millions of years of hominin natural selection to rapidly resolve recurrent social dilemmas—such as sharing resources, protecting kin, and punishing direct interpersonal assault—without requiring costly cognitive deliberation.
• Manual Mode: Our capacity for conscious, deliberative moral reasoning represents the brain’s manual override. It is uniquely capable of operating outside evolutionary precedents, systematically analyzing novel, abstract, and multi-layered ethical challenges where ancestral heuristics fail.

Under this analogy, the Footbridge dilemma is an evolutionary trick: it places the human brain in a situation where the ancestral automatic setting (“Do not push an innocent conspecific to his death”) fires with overwhelming intensity, even though the structural arithmetic of the situation presents a novel, manual-mode optimization challenge (“Save five lives at the cost of one”).

3.3 Evolutionary Epistemology of Moral Intuitions

The camera analogy leads directly into Greene’s evolutionary epistemology of moral judgment. Human beings did not evolve under the conditions of twenty-first-century globalized civilization, governed by international trade, institutional legal frameworks, probabilistic risk models, and long-range technological causation. Rather, the cognitive architecture of Homo sapiens was forged during the Pleistocene epoch within small, kin-based, face-to-face hunter-gatherer bands.

In this ancestral environment, interpersonal violence was overwhelmingly “up-close and personal.” It was executed through muscular force, tooth, club, or spear. A hominin band whose individual members felt no visceral, instinctive aversion to physically striking, strangling, or bludgeoning their conspecifics would have suffered catastrophic internal destabilization, fracturing the cooperative alliances essential for group survival. Thus, natural selection installed powerful, hardwired emotional tripwires: immediate, pre-reflective horror at the prospect of inflicting direct physical violence on an individual in close proximity. These affective tripwires are the evolutionary ancestors of our contemporary deontological intuitions.

Crucially, ancestral environments featured no remote switches, no industrial machinery, no guided missiles, and no public health allocation algorithms. Consequently, evolution had neither the cause nor the selective pressure to install visceral, automatic emotional prohibitions against indirect, mediated, or impersonal harms. Harming someone by pulling a lever, altering a train track, or adjusting a financial spreadsheet does not trigger our evolutionary alarm bells, because these causal modalities did not exist when our affective heuristics were being hardwired. Herein lies the profound evolutionary mismatch: our automatic settings remain tuned to the physical kinematics of the Pleistocene, rendering our moral intuitions hyper-sensitive to proximate, personal harm, while leaving us emotionally numb to impersonal, systemic, or structurally distributed harms—even when the latter are catastrophically more lethal in the modern world.

4. The Seminal 2001 Science Study: Methodological Framework

4.1 Taxonomy of Experimental Stimuli: Moral-Personal Versus Moral-Impersonal

To translate these theoretical constructs into a rigorous empirical design, Greene and his colleagues (2001) developed a comprehensive taxonomy of computerized vignettes. Recognizing that the term “moral dilemma” was far too coarse, Greene constructed an operationalized classification system that divided experimental stimuli into three distinct categories: Moral-Personal, Moral-Impersonal, and Non-Moral.

The operational definition of a Moral-Personal dilemma was constructed with immense analytical precision. A moral violation was classified as “personal” if, and only if, it met three conjunctive criteria:

1. The action must be expected to cause serious bodily harm (typically death or severe physical trauma);
2. The harm must be directed against a particular person or a distinct set of identifiable persons;
3. The harm must not be the result of deflecting an existing, pre-existing threat onto a different party, but must involve an agent-driven intervention where the victim’s bodily integrity is violated through direct agency (encapsulated by the colloquial heuristic: “ME HURT YOU”).

The quintessential operationalization of a moral-personal dilemma was the Footbridge case, alongside scenarios such as smothering a crying baby to prevent discovery by enemy soldiers, or harvesting organs from an innocent patient to save five dying patients. Conversely, a dilemma was categorized as Moral-Impersonal if it involved serious moral stakes but failed to meet one or more of the personal criteria. The classic exemplar was the Switch dilemma (Bystander at the Switch), alongside cases such as voting for a policy that allocates funds away from medical care, or keeping found money that belongs to an impersonal institution. Finally, Non-Moral control scenarios were meticulously matched for narrative length, syntax, and complexity, but involved mundane, non-ethical decision-making (e.g., choosing whether to take a train or a bus to an appointment, or selecting ingredients for a recipe).

4.2 Neuroimaging Parameters and Subject Demographics

In the seminal 2001 study published in Science, Greene and his team scanned nine healthy human participants (followed rapidly by expanded cohorts in subsequent replications) using high-field functional magnetic resonance imaging. The experimental environment required subjects to lie supine inside a 1.5-Tesla scanner, viewing high-resolution narrative text projected onto a mirror mounted to the head coil. Dilemmas were presented across multiple visual frames: subjects read the background context, followed by the presentation of a proposed course of action (e.g., “Do you push the stranger off the footbridge to save the five?”), at which point they recorded their judgment by pressing one of two buttons corresponding to “Appropriate” or “Inappropriate.”

The scan protocol utilized a functional BOLD contrast sequence, sensitive to local changes in the ratio of oxyhemoglobin to deoxyhemoglobin in cerebral blood flow. Scans were acquired across functional volumes covering the entire cerebrum. Greene utilized an event-related and block-hybrid design, isolating the specific computational epoch wherein the moral dilemma was read and the decisive moral evaluation was computed. Preprocessing pipelines applied rigid motion correction protocols to scrub head-movement artifacts, spatial normalization into standard stereotactic space (Talairach/MNI coordinates), and spatial smoothing using an isotropic Gaussian kernel to optimize signal-to-noise ratios across subjects.

By contrasting the localized BOLD activation maps generated during Moral-Personal dilemmas against those generated during Moral-Impersonal and Non-Moral conditions, the researchers could execute rigorous subtraction analyses. These contrasts isolated the specific neural networks uniquely recruited by the visceral prospect of inflicting up-close, personal harm.

4.3 Behavioral Metrics: Latency and Response Selection

In parallel with hemodynamic data, the experimental software captured precise behavioral metrics with millisecond-level fidelity: categorical response selection (the proportion of “Appropriate” versus “Inappropriate” choices) and reaction time (RT) latencies spanning the interval from the dilemma’s decisive prompt to the button-press execution.

The behavioral metrics yielded two primary empirical outcomes:

• Categorical Endorsements: Replicating decades of philosophical intuition, participants overwhelmingly judged personal violations (e.g., Footbridge) to be “Inappropriate” (acting in accordance with deontological prohibitions), while endorsing impersonal violations (e.g., Switch) as “Appropriate” (acting in accordance with utilitarian maximization).
• Reaction Time Divergence: Greene uncovered a crucial, highly telling latency interference effect. In moral-personal dilemmas, when subjects chose the conventional deontological option (declaring the push to be “Inappropriate”), their reaction times were remarkably rapid and uniform. However, on the rare occasions when subjects endorsed the utilitarian option (declaring the push to be “Appropriate”), their reaction times were substantially and statistically significantly prolonged.

This RT delay was the exact behavioral footprint predicted by the dual-process model. A subject who endorses pushing the man off the footbridge is not simply calculating a cold mathematical equation; they are experiencing an acute, visceral emotional alarm bell telling them to abort the action. To push the button marked “Appropriate,” the subject must actively recruit cognitive resources to override, dampen, and veto that affective alarm—a process of internal conflict resolution that consumes measurable cognitive time.

5. Neural Substrates of Moral-Personal Dilemmas: The Affective Core

5.1 Medial Prefrontal Cortex (mPFC) and Social Value Computation

When Greene and his colleagues analyzed the fMRI contrasts between Moral-Personal and Moral-Impersonal conditions, the imaging data revealed an unmistakable, localized divergence. Moral-personal dilemmas elicited profound, statistically robust BOLD signal elevations in a network of structures long recognized as the neurological bedrock of affective processing and social cognition. Chief among these was the medial prefrontal cortex (mPFC), spanning Brodmann Areas (BA) 9, 10, and 32.

The mPFC is a complex prefrontal hub deeply integrated into the brain’s Default Mode Network (DMN). It plays an indispensable functional role in self-referential thought, autobiographical memory, mentalizing (Theory of Mind), and the computation of subjective social and affective value. In the context of the Footbridge dilemma, the mPFC is recruited almost instantaneously as the subject simulates the first-person scenario: the physical posture of approaching another human being, placing hands upon their back, and directly causing their catastrophic plunge to the rails below. The ventral and medial swaths of the prefrontal cortex act as a convergence zone, bridging high-level conceptual representations of the proposed scenario with visceral, somatic emotional markers transmitted from subcortical structures.

Subsequent high-resolution neuroimaging work has confirmed that the degree of mPFC activation during moral dilemmas directly correlates with the perceived moral badness of the act. Rather than computing an abstract mathematical balance sheet, the mPFC generates an affective valuation signal that tags the action with profound negative valence. This affective tag serves as the primary computational input for the rapid, deontological judgment: “This act is wrong; it is impermissible.”

5.2 Posterior Cingulate Cortex (PCC) and Precuneus Recruitment

Working in tight functional synchronization with the medial prefrontal cortex, the posterior cingulate cortex (PCC, BA 31) and the adjacent precuneus (BA 7) demonstrated massive metabolic recruitment during Moral-Personal conditions, while exhibiting relative quiescence or active deactivation during Moral-Impersonal and Non-Moral trials.

The PCC and precuneus form an essential core of the brain’s internal mental simulation and perspective-taking machinery. When human beings are asked to engage in episodic memory retrieval or construct vivid counterfactual mental models of future events, the PCC/precuneus complex acts as a central projection engine. In moral-personal dilemmas, the subject is not presented with a mere abstract logic puzzle; the narrative demands the construction of an embodied, egocentric simulation of physical violence. The precuneus is responsible for establishing an egocentric spatial framework—placing the self directly onto the footbridge, feeling the weight of the stranger, tracking the trajectory of the body falling through space.

Furthermore, the posterior cingulate cortex is implicated in tracking the emotional relevance of external and internal stimuli, acting as an affective filter that monitors how deeply an imagined scenario impinges upon personal survival, identity, and social attachment. The intense recruitment of the PCC/precuneus network during the Footbridge dilemma indicates that personal moral dilemmas force the human brain into an intense, immersive, first-person affective simulation—a simulation that is conspicuously bypassed when subjects consider pulling an impersonal mechanical switch on a distant control panel.

5.3 The Amygdala: Salience and Visceral Alarm Responses

Beneath the neocortex, deep within the temporal lobes, the fMRI data revealed the recruitment of the bilateral amygdaloid complex. While the original 2001 study, constrained by early 1.5-Tesla signal-to-noise limitations in subcortical structures, observed this trend primarily within paralimbic regions, Greene’s subsequent 2004 Neuron study and numerous international replications confirmed that the amygdala is the crucial subcortical spark generating the affective revulsion in personal moral dilemmas.

The basolateral amygdala is the brain’s primary hub for detecting behaviorally salient, survival-relevant stimuli, specialized in processing threat, distress cues, and the emotional appraisal of proximate danger. When an agent contemplates an act of personal violence, the amygdala receives rapid, low-latency inputs via subcortical thalamic pathways (the “low road”), followed by nuanced sensory inputs via the cortical “high road.” The amygdala immediately registers the visceral horror of the imagined transgression, firing an unconditioned somatic alarm bell. This triggers a downstream cascade: signaling the hypothalamus and brainstem to induce peripheral autonomic arousal (elevated heart rate, skin conductance fluctuations, pupillary dilation, and transient respiratory inhibition).

This somatic alarm response operates precisely as described in Antonio Damasio’s Somatic Marker Hypothesis. The amygdala injects a powerful, visceral signal into the decision loop, marking the proposed behavioral option (pushing the man) with an overwhelming negative somatic tag. In the vast majority of healthy human subjects, this amygdaloid alarm operates ballistically: it interrupts ongoing cognitive processes, short-circuits any nascent utilitarian arithmetic, and forces the motor system to execute a rapid, decisive rejection: “No, that is unacceptable.”

6. Neural Substrates of Moral-Impersonal Dilemmas: The Cognitive Control Network

6.1 Dorsolateral Prefrontal Cortex (dlPFC) and Utilitarian Maximization

When Greene and his collaborators turned their attention to the contrasting condition—Moral-Impersonal dilemmas like the classic switch scenario—the neuroimaging data painted a radically different picture. The paralimbic and affective networks that dominated the Footbridge dilemma fell silent, or exhibited BOLD decreases below baseline. In their place, the fMRI scans revealed robust, selective recruitment of the Central Executive Network (CEN), anchored by the bilateral dorsolateral prefrontal cortex (dlPFC), localized predominantly across Brodmann Areas 9 and 46.

The dlPFC is the crowning evolutionary achievement of the human prefrontal mantle. It is the primary seat of executive function, working memory, cognitive flexibility, abstract rule representation, and goal-directed planning. When confronted with an impersonal moral scenario, the dlPFC organizes and manipulates the quantitative parameters of the problem. Free from the overpowering interference of an amygdaloid alarm, the dlPFC constructs a mathematical representation of the dilemma’s causal mechanics: five human beings face imminent death; a mechanical diversion path exists; the intervention carries an unavoidable collateral cost of one life; the net outcome is +4 lives saved.

Greene’s functional data demonstrated a direct correlation: the greater the BOLD signal elevation within the bilateral dlPFC (specifically within the middle frontal gyrus), the more likely the participant was to deliver a utilitarian judgment endorsing the sacrifice. The dlPFC computes aggregate utility by maintaining competing outcomes in active working memory, weighing their respective magnitudes, and systematically executing the utilitarian algorithm. It is the neural instantiation of Bentham’s felicific calculus, transforming an emotional ethical dilemma into an optimization problem.

6.2 Anterior Cingulate Cortex (ACC) and Conflict Monitoring

One of the most theoretically vital discoveries of Greene’s neuroimaging paradigm was the precise localization of cognitive conflict within the anterior cingulate cortex (ACC, BA 24/32), specifically within its dorsal division (dACC). Situated at the structural and functional crossroads between the limbic system and the prefrontal cortex, the dACC serves as the brain’s chief conflict-monitoring and error-detection hub.

In classical cognitive neuroscience paradigms such as the Stroop task—where subjects must name the ink color of the word “RED” printed in blue letters—the dACC exhibits a dramatic burst of BOLD activity. This activity does not represent the execution of the task itself, but rather the detection of simultaneous, incompatible response tendencies (the automatic urge to read the word “RED” clashing with the task-instructed goal of naming the ink “BLUE”). According to the influential conflict-monitoring model developed by Matthew Botvinick, Jonathan Cohen, and colleagues, the dACC functions as an online alarm that detects computational friction between competing neural circuits. Once detected, the dACC sends an urgent request to the dlPFC, demanding the recruitment of top-down cognitive control to resolve the conflict and suppress the inappropriate response.

Greene demonstrated that the dACC behaves identically in the moral domain. In low-conflict dilemmas (such as obvious Non-Moral tasks or trivial moral questions where both affect and calculus agree), dACC activity remains baseline. However, in high-conflict moral dilemmas—precisely where visceral System 1 emotional revulsion clashes directly with cold System 2 utilitarian calculation—the dACC exhibits sustained, high-amplitude metabolic recruitment. The dACC tracks the subjective difficulty of the moral dilemma. It acts as the neural arbiter, signaling that the automatic and manual modes have generated contradictory instructions, and alerting the frontoparietal executive network to step in and arbitrate the dispute.

6.3 The Inferior Parietal Lobule and Working Memory Demands

Flanking the dlPFC within the frontoparietal executive network, the bilateral inferior parietal lobules (IPL, BA 39/40) were consistently co-activated during moral-impersonal judgments. The inferior parietal cortex, particularly around the intraparietal sulcus, is known to be intimately involved in spatial attention, working memory maintenance, and, critically, numerical cognition and quantitative reasoning.

In moral dilemmas that require cost-benefit balancing, the IPL functions as a computational workspace for the dlPFC. It maintains the numerical representations of the agents involved (one vs. five) and performs the spatial-causal transformations necessary to conceptualize how an action will play out over time (e.g., how throwing a switch redirects a mass along a split track). During impersonal moral judgments, the IPL exhibits tight functional connectivity with the dlPFC, forming a frontoparietal loop that operates entirely free from emotional interference.

This functional dissociation was striking: while personal moral dilemmas systematically recruit the DMN and paralimbic structures (mPFC, PCC, amygdala), impersonal moral dilemmas systematically recruit the Central Executive Network (dlPFC, IPL, dACC). The human brain, Greene argued, does not possess a singular “moral center.” Instead, our moral determinations are the output of an ongoing, dynamic struggle between at least two radically distinct, evolutionarily divergent, large-scale neurocognitive networks.

7. Reaction Time Discrepancies and Cognitive Conflict Resolution

7.1 The Interference Effect in Utilitarian Endorsements

The behavioral metrics captured during Greene’s early fMRI experiments provided crucial kinetic evidence confirming the dual-process architecture. The most prominent behavioral phenomenon was what became known as the moral interference effect. In moral-personal dilemmas, subjects who delivered deontological rejections (“No, pushing the stranger off the footbridge is inappropriate”) did so with rapid, streamlined reaction times. Their response latencies mirrored those of trivial non-moral control decisions.

In stark contrast, when subjects chose the utilitarian path in these same moral-personal dilemmas (“Yes, pushing the stranger is appropriate to save five”), their reaction times were substantially prolonged. This reaction time tail was not a minor statistical artifact; it represented a statistically robust, multi-second deceleration of the cognitive apparatus. Why did this latency divergence occur?

Under Greene’s model, the explanation is computationally elegant. In the deontological case, the behavioral choice is aligned with the initial, ballistic output of System 1. The amygdala and mPFC fire their emotional alarm, declaring the action impermissible. If the participant chooses to accept this intuitive output, no extensive cognitive negotiation is required. The automatic setting has fired, the behavioral veto is registered, and the button is pressed. However, in the utilitarian case, the subject cannot simply push the button. The automatic setting has fired with immense force, but the subject’s deliberative System 2 calculations have simultaneously determined that five lives outweigh one. To execute the utilitarian response, the executive control machinery (dlPFC) must be rallied by the conflict monitor (dACC) to actively suppress, inhibit, and override the affective signal generated by the mPFC-amygdala axis. This top-down inhibitory gating takes metabolic time, computational resources, and sustained focus, directly resulting in prolonged reaction time latencies.

7.2 Statistical Robustness and Replicability Assessments

Following the publication of Greene’s initial papers, the reaction-time data became the subject of intense methodological scrutiny within cognitive psychology. In a widely cited critique, McGuire and colleagues (2009) re-analyzed Greene’s original stimulus set, arguing that the observed reaction time differences might be an artifact of confounding variables, specifically reading comprehension times, sentence length, and narrative complexity across the vignettes. Furthermore, McGuire et al. noted that if one removed a small subset of “unusual” or poorly calibrated dilemmas from the pool, the statistical significance of the RT difference was attenuated.

Greene immediately responded with exhaustive empirical re-analyses (Greene, 2008; Greene, 2009). Greene demonstrated that when one stratifies the dilemmas with higher methodological rigor—specifically isolating what he termed “high-conflict” moral-personal dilemmas (scenarios where the utilitarian payoff is massive, catastrophic, and deeply agonizing, such as the Crying Baby scenario) from “low-conflict” moral-personal dilemmas (scenarios where the personal harm is gratuitous or the trade-off is trivial, such as stealing a stranger’s clothes to keep warm)—the interference effect is not only robust, but massive. In low-conflict cases, rejection of the personal harm is near-instantaneous. In high-conflict personal cases, reaction times soar across all participants, and the RT gap between deontological rejections and utilitarian endorsements widens significantly.

Subsequent international replications utilizing expanded cohorts, improved non-magnetic response boxes, millisecond-accurate display software, and eye-tracking controls have verified the core phenomenon: overriding an automatic affective aversion to commit instrumental violence takes measurable cognitive time. The interference effect stands as one of the most thoroughly replicated behavioral markers in experimental moral psychology.

7.3 Individual Differences in Cognitive Reflection and Working Memory Capacity

If utilitarian moral judgment relies on effortful, capacity-limited System 2 deliberation, it follows that an individual’s propensity to make utilitarian judgments should correlate directly with their general cognitive capacity, working memory capacity, and cognitive style. This prediction has been confirmed through extensive psychological testing.

Psychologists utilize the Cognitive Reflection Test (CRT), developed by Shane Frederick, to measure an individual’s disposition to resist rapid, intuitive, but incorrect answers in favor of reflective, computationally correct responses (for example: “A bat and a ball cost $1.10 in total. The bat costs$1.00 more than the ball. How much does the ball cost?”). Multiple studies have demonstrated that individuals who score highly on the CRT—those who habitually deploy System 2 manual overrides in abstract logic—exhibit a significantly higher baseline propensity to endorse utilitarian solutions to high-conflict personal moral dilemmas, including the Footbridge case.

Conversely, when experimenters place participants under simultaneous cognitive load—such as forcing them to hold an intricate string of alphanumeric characters in working memory, or perform continuous auditory Stroop tasks while deliberating over moral scenarios—the capacity of System 2 is artificially throttled. Under high cognitive load, Greene and his collaborators (2008) demonstrated that subjects exhibit increased reaction time delays specifically when making utilitarian judgments, and the absolute proportion of utilitarian endorsements decreases. The executive control network, depleted of the metabolic resources required to execute active top-down suppression, defaults back to the automatic, deontological heuristics of System 1.

8. The 2004 Neuron Study: Dissecting Difficult Moral Dilemmas

8.1 Stratifying Dilemmas: High-Conflict Versus Low-Conflict Scenarios

Building upon the foundational insights of 2001, Greene and his collaborators designed an ambitious follow-up neuroimaging study, published in Neuron in 2004, titled “The Neural Bases of Cognitive Conflict and Control in Moral Judgment.” This study addressed early criticisms by implementing a much more sophisticated, parametrically calibrated taxonomy of moral stimuli, explicitly stratifying personal moral dilemmas into High-Conflict and Low-Conflict tiers.

Low-conflict personal dilemmas were scenarios where the moral calculation was straightforwardly asymmetric and trivial. For instance, in one scenario, an agent considers whether to kill an annoying boss to secure an inheritance, or whether to push someone onto the tracks simply to avoid being late for dinner. In these scenarios, there is no moral friction: System 1 screams “No!” and System 2 concurs that the cost-benefit balance sheet is completely unjustifiable. Consequently, subjects exhibit near-100% consensus, rapid reaction times, and minimal frontoparietal or cingulate activation.

High-conflict personal dilemmas, by contrast, represented the true crucible of moral cognition. The archetypal scenario introduced in this study was the harrowing Crying Baby Dilemma:

“Enemy soldiers have taken over your village. They have orders to kill all remaining civilians. You and some of your townspeople have sought refuge in the cellar of a large house. Outside, you hear the voices of soldiers who have come to search the house for valuables. Your baby begins to cry loudly. You cover his mouth to block the sound. If you remove your hand, his crying will summon the attention of the soldiers who will kill you, your child, and the others hidden in the cellar. To save yourself and the others, you must smother your child to death.”

Here, the computational conflict is maximized. The visceral, affective aversion to suffocating one’s own infant is among the most powerful evolutionary prohibitions hardwired into the mammalian brain. Yet, the consequentialist math is absolute and inescapable: if you do not smother the child, the child dies anyway, along with yourself and everyone hidden in the cellar (1 life lost vs. 30 lives lost, where the 1 is guaranteed to die regardless). It is the absolute thermodynamic limit of moral conflict.

8.2 Mechanisms of Affective Override: Prefrontal-Subcortical Tensions

The 2004 Neuron study revealed that high-conflict dilemmas elicited a massive, sustained surge of metabolic activation within the dorsal anterior cingulate cortex (dACC). The dACC acted as a direct parametric gauge of subjective decision difficulty: the longer a participant hesitated, and the more agonizing the internal struggle between the emotional prohibition and the mathematical outcome, the higher the BOLD signal peaked within the cingulate sulcus.

Crucially, the study established the precise neural mechanism of affective override. When subjects ultimately pushed the button to smother the crying baby—endorsing the utilitarian option—the neuroimaging data demonstrated a massive, selective recruitment of the right and left dorsolateral prefrontal cortex (BA 46), accompanied by the anterior prefrontal cortex (BA 10). When subjects refused to smother the baby—delivering the deontological veto—the dlPFC signal remained significantly lower, while the medial prefrontal cortex and paralimbic structures maintained dominance.

Parametric analyses revealed an inverse functional coupling: as dlPFC activation intensified, it exerted top-down inhibitory modulation over the ventral emotional circuitry. Greene had captured the neurobiological mechanism of manual override in action: the prefrontal executive network actively working against subcortical affective resistance, burning metabolic glucose to suppress an evolutionary taboo in the service of an abstract consequentialist calculus.

8.3 The Tripartite Architecture: mPFC, dlPFC, and dACC Interaction Dynamics

The theoretical synthesis of the 2004 Neuron paper established what is now understood as the Tripartite Architecture of Moral Cognition. This neurobiological model maps moral decision-making across three functionally interconnected and dynamically competing hubs:

• The Affective Engine (mPFC / Amygdala / PCC): This network computes rapid, automatic social-emotional values. It responds with high-amplitude somatic alarm signals to representations of direct, personal violence, generating the foundational intuitions that animate deontological ethics.
• The Deliberative Engine (dlPFC / IPL): This network computes abstract, rule-based, and consequentialist equations. It maintains numerical parameters within working memory, projects counterfactual outcomes, and systematically calculates net aggregate utility.
• The Conflict Arbiter (dACC): This structure continuously monitors real-time cognitive and affective tension between the Affective and Deliberative engines. When the two systems generate incompatible behavioral trajectories, the dACC fires an executive alert, signaling the dlPFC to deploy top-down inhibitory control to suppress the affective engine.

Using structural equation modeling and functional connectivity analyses, Greene and his team mapped the temporal and directional dynamics of this tripartite circuit. The dACC conflict detection systematically precedes the full activation of the bilateral dlPFC. When an agent is confronted with a high-stakes ethical dilemma, the limbic system fires first; the ACC registers the computational clash between that limbic fire and nascent consequentialist logic; and the dlPFC is subsequently recruited to execute the top-down manual override. The human moral mind is thus exposed not as a harmonious, unified sanctuary of pure reason, but as a biological battlefield where ancient emotions and modern executive control vie for dominance over the motor cortex.

9. Convergent Neurobiological Evidence: Lesions and Neurochemistry

9.1 Ventromedial Prefrontal Cortex (vmPFC) Lesion Studies

A fundamental limitation of functional magnetic resonance imaging is that it provides correlational, rather than causal, data. Demonstrating that the mPFC, PCC, or dlPFC lights up during a moral decision does not prove that these structures are indispensable to the execution of that judgment. To establish genuine neurobiological causality, cognitive neuroscience must turn to lesion studies and pharmacological interventions. In the mid-2000s, precisely such causal confirmation arrived.

In a landmark 2007 study published in Nature, Michael Koenigs, Antonio Damasio, and their collaborators investigated moral judgment in patients with focal, bilateral damage to the ventromedial prefrontal cortex (vmPFC)—a cohort famously exemplified historically by Phineas Gage. The vmPFC is a critical anatomical bridge linking subcortical emotional systems (like the amygdala) with higher prefrontal cognitive regions. Patients with vmPFC lesions exhibit preserved abstract intellect, normal IQ, and intact working memory, but suffer from a profound blunting of emotional processing and social affect, alongside a severe impairment in somatic marker generation.

Koenigs et al. presented vmPFC lesion patients, healthy normal controls, and brain-damaged controls (individuals with focal lesions outside the vmPFC) with the identical battery of moral dilemmas developed by Greene. The results were dramatic and unequivocal: in low-conflict moral dilemmas and moral-impersonal dilemmas, vmPFC patients behaved identically to healthy controls. However, when presented with high-conflict, personal moral dilemmas like the Footbridge and Crying Baby scenarios, the vmPFC patients exhibited a hyper-utilitarian judgment profile:

While only roughly 10% to 20% of healthy controls endorse pushing the stranger off the footbridge, nearly 50% to 60% of vmPFC lesion patients judged pushing the man to be entirely “Appropriate.” Autonomic recordings confirmed that while healthy controls exhibited massive, anticipatory spikes in Galvanic Skin Response (GSR)—a peripheral somatic marker of autonomic arousal—prior to contemplating personal violence, the vmPFC patients exhibited flatline autonomic responses. Stripped of the neurobiological machinery that generates visceral emotional horror, vmPFC patients viewed the Footbridge dilemma as a simple arithmetic problem: 5 is greater than 1. Their intact dlPFC computed the utilitarian calculus completely unhindered by affective resistance. This provided definitive, causal proof for Greene’s dual-process architecture: when the affective core is physically destroyed, human moral judgment defaults seamlessly to cold utilitarian maximization.

9.2 Serotonergic Modulation of Harm Aversion

If moral judgment is governed by neurobiological circuits, it should be susceptible to pharmacological manipulation at the neurochemical level. Specifically, the neurotransmitter serotonin (5-HT) has long been implicated in the regulation of emotional processing, impulsive aggression, and social harm aversion within the limbic and paralimbic systems.

In an elegant 2010 study published in the Proceedings of the National Academy of Sciences (PNAS), Molly Crockett and her colleagues investigated the causal role of serotonin in moral dilemmas by administering citalopram—a selective serotonin reuptake inhibitor (SSRI) that acutely elevates synaptic serotonin levels—to healthy human participants in a double-blind, placebo-controlled crossover design. The researchers sought to determine whether transiently enhancing serotonergic neurotransmission would selectively amplify the emotional aversion to inflicting direct harm.

The findings confirmed the pharmacological tractability of moral judgment: participants under the influence of acute citalopram administration became significantly less utilitarian in high-conflict personal moral dilemmas. They exhibited a heightened resistance to pushing the stranger off the footbridge or smothering the crying baby, judging these acts as profoundly inappropriate at rates significantly higher than when under placebo. Crucially, citalopram had zero effect on moral-impersonal dilemmas (the Switch case) or non-moral scenarios.

Neurochemically, serotonin acts on 5-HT2A and 5-HT1A receptors within the basolateral amygdala and ventral prefrontal cortex, enhancing the sensitivity of these networks to interpersonal distress cues and scaling up the subjective cost of direct harm. By artificially tuning the gain of the brain’s affective alarm system, Crockett et al. demonstrated that deontological intuitions can be dialed up or down via acute pharmacological intervention, further undermining the classical philosophical premise that these judgments reflect immutably derived, rational principles.

9.3 Psychopathology and Moral Insensitivity

The neurobiological dissection of moral judgment also illuminates the dark side of moral cognition: clinical psychopathy. Individuals with psychopathy, characterized by callous-unemotional traits, shallow affect, an absence of empathy, and chronic antisocial behavior, have historically presented an enigma to forensic psychology.

James Blair’s neurodevelopmental model of psychopathy argues that the condition is fundamentally rooted in a disruption of the Violence Inhibition Mechanism (VIM)—a hardwired mammalian circuit designed to abort aggressive behavior upon detecting distress cues (crying, submission, pain) from a conspecific. Neuroimaging of psychopaths consistently demonstrates structural and functional abnormalities across the uncinate fasciculus (the white-matter tract connecting the amygdala to the ventral prefrontal cortex), resulting in profound hypoactivity within the amygdala and mPFC during moral perspective-taking.

When presented with Greene’s moral battery, psychopaths display an abnormally high propensity to endorse utilitarian solutions to personal moral dilemmas. They will push the man off the footbridge without hesitation. However, cognitive neuroscientists, including Guy Kahane and colleagues, have issued a crucial caveat against conflating psychopathic moral calculation with philosophical utilitarianism. Utilitarianism, as conceived by John Stuart Mill or Peter Singer, is rooted in radical, impartial altruism—a deep, universal concern for the aggregate welfare of all sentient life that demands the equal consideration of every being’s suffering.

Psychopaths do not endorse utilitarian outcomes out of an expansive, universal benevolence. Rather, their judgments reflect a profound affective deficiency: they are simply blind to the moral weight of harm. This has led researchers to distinguish between two dissociable dimensions of utilitarian judgment:

• Instrumental Harm: The willingness to sacrifice an individual for the greater good (driven by low empathy, reduced harm aversion, and affective blunting).
• Impartial Beneficence: The positive, universal devotion to maximizing the global welfare of all people, even at great personal cost to the self (driven by high cognitive reflection, general empathic concern, and abstract prosocial principles).

Psychopathy selectively mimics the instrumental harm dimension while being entirely devoid of impartial beneficence, proving that the destruction of System 1 affect produces moral sociopathy, not philosophical enlightenment.

10. Theoretical and Methodological Critiques of Greene’s Paradigm

10.1 The Ecological Validity Critique and Artificiality of Trolleyology

Despite its transformative influence, Joshua Greene’s fMRI trolley paradigm has faced severe criticism from both within philosophy and across cognitive neuroscience. One of the most prominent lines of attack focuses on the issue of ecological validity, often referred to derisively as the critique of “Trolleyology.”

Philosophers like Selim Berker and Guy Kahane, along with behavioral psychologists, have pointed out that standard trolley and footbridge scenarios are bizarre, surreal, and thoroughly unrepresentative of the moral challenges humans encounter in real-world existence. In real life, humans never operate under conditions of epistemic certainty where they know with 100% mathematical precision that pushing one individual will reliably save exactly five. In reality, shoving a heavy stranger off a bridge is vastly more likely to result in two deaths: the stranger is crushed, and the train derails, killing the five anyway. Critics argue that by stripping dilemmas of probabilistic uncertainty, sensory complexity, and social context, Greene created an artificial, hothouse laboratory task that forces the brain into abnormal computational states.

Furthermore, there is a profound chasm between pressing a plastic button inside an fMRI bore while reading a hypothetical cartoon-like story, and actually reaching out with one’s physical hands to push a living, breathing human being to their death. To address this discrepancy, researchers have turned to Immersive Virtual Reality (VR) and high-fidelity sensory simulations. In VR paradigms developed by Carlos Navarrete, Kathryn Francis, and colleagues, participants don head-mounted displays and navigate physically embodied crisis scenarios. Interestingly, these VR studies reveal that when subjects are confronted with multisensory, spatially immersive representations of trolley dilemmas, the rate of utilitarian actions actually increases behaviorally—though it is accompanied by massive, objective autonomic distress (skyrocketing heart rates and skin conductance).

These findings suggest that while Greene’s laboratory vignettes capture genuine cognitive conflict, they necessarily abstract away from the visceral, somatic terror of real-world physical action, potentially over-intellectualizing the nature of human moral performance under existential crisis.

10.2 The Reverse Inference Fallacy in Cognitive Neuroimaging

A second major methodological critique strikes at the heart of neuroimaging interpretation itself: the problem of reverse inference, famously articulated by cognitive neuroscientist Russell Poldrack (2006). A reverse inference occurs when a researcher observes metabolic activation in a specific brain region and infers backwards that a specific cognitive or psychological process was occurring (e.g., “The anterior insula is active, therefore the subject is feeling disgust”).

Poldrack demonstrated that reverse inference is logically deductively invalid, and its inductive strength depends entirely on the selective specificity of the brain region in question. If a brain region is recruited by only one mental operation, reverse inference is reliable. However, the brain regions at the center of Greene’s model are notorious for their broad, domain-general multi-functionality:

• The dorsolateral prefrontal cortex (dlPFC) is not an exclusive “utilitarian calculator.” It is activated in thousands of disparate neuroimaging tasks involving spatial working memory, syntactic processing, chess playing, visual attention, and motor sequencing.
• The anterior cingulate cortex (dACC) is recruited by physical pain, monetary loss, autonomic arousal, task switching, and sensory novelty—not solely moral conflict.
• The medial prefrontal cortex (mPFC) is equally implicated in mundane episodic memory recall, autobiographical navigation, and general semantic integration.

Critics such as Selim Berker (2009) argued that Greene committed classic reverse inference errors: inferring that because the dlPFC lit up, “utilitarian calculation” was occurring, or because the mPFC lit up, “raw emotion” was dictating the outcome. To overcome these univariate reverse inference weaknesses, modern cognitive neuroscience has transitioned toward Multivariate Pattern Analysis (MVPA) and whole-brain machine learning decoders. MVPA examines distributed, fine-grained spatial patterns of voxels across the entire brain, rather than relying on coarse regional averaging. Recent MVPA studies have largely vindicated Greene’s overarching framework, demonstrating that moral decisions can indeed be decoded from distributed neural representations of subjective value, executive control, and affective salience, though with a degree of network-level complexity that far surpasses the early, localized models of 2001.

10.3 Alternative Theoretical Models: Social Intuitionism and Moral Foundations

Greene’s model also faces intense competition from alternative, highly influential frameworks within evolutionary and psychological science. Chief among these is Jonathan Haidt’s Social Intuitionist Model. Haidt agrees with Greene that moral intuitions are primarily affective and automatic. However, Haidt rejects Greene’s elevated role for System 2 manual override. Under Haidt’s framework, conscious moral reasoning almost never overrides an intuitive gut reaction; rather, moral reasoning operates like a “lawyer defending a client,” generating endless post-hoc rationalizations to justify whatever conclusion the emotional elephant has already stumbled into.

Furthermore, Haidt, Jesse Graham, and their colleagues developed Moral Foundations Theory, which directly challenges the narrow scope of the trolley problem. Haidt argues that “Trolleyology” suffers from a profound Western, secular, utilitarian-deontological bias, compressing the vast spectrum of human morality down to a single axis: harm versus welfare. Moral Foundations Theory demonstrates that across traditional, non-Western, and politically conservative societies, morality is constructed upon at least six distinct evolutionary pillars:

1. Care / Harm
2. Fairness / Cheating
3. Loyalty / Betrayal
4. Authority / Subversion
5. Sanctity / Degradation
6. Liberty / Oppression

From this perspective, the trolley problem only tests the “Care/Harm” foundation, leaving the neural networks governing sacredness, tribal loyalty, and ritual purity entirely unprobed.

Finally, John Mikhail and Marc Hauser proposed an alternative computational paradigm: Universal Moral Grammar (UMG). Drawing inspiration from Noam Chomsky’s generative linguistics, UMG posits that the human mind possesses a specialized, innate moral faculty that automatically computes the structural, legalistic, and intentional syntax of an action—such as tacitly calculating whether an outcome is a means or an end, and identifying causal intervention pathways—long before any emotional arousal occurs. Under Mikhail’s model, the affective response is not the cause of the moral judgment, but merely a downstream consequence of a fast, non-emotional computational appraisal of intentionality and causal mechanics.

11. Meta-Ethical and Normative Implications of Moral Neurobiology

11.1 Navigating the Is-Ought Problem and the Naturalistic Fallacy

Perhaps the most controversial dimension of Joshua Greene’s scholarly corpus is his audacious leap from descriptive neuroscience to normative ethics. Can the study of blood flow, action potentials, and neurotransmitters tell us anything about how we ought to live? Can functional neuroimaging solve philosophical problems that have confounded thinkers for centuries?

Since the publication of David Hume’s Treatise of Human Nature (1739), philosophers have recognized the inviolability of the Is-Ought Problem (Hume’s Guillotine). Hume observed that thinkers routinely transition from descriptive claims about what is (empirical facts about human nature, biology, or psychology) to prescriptive claims about what ought to be (normative moral obligations), without offering any logical justification for this conceptual leap. In the early twentieth century, G.E. Moore formalized a related barrier: the Naturalistic Fallacy, arguing that “good” is a simple, non-natural property that cannot be identified with or defined by any natural property, such as “pleasurable,” “evolutionarily adaptive,” or “neurologically active.”

Greene readily concedes that one cannot deductively derive an “ought” from an “is.” Knowing that the amygdala fires during the Footbridge dilemma does not, by itself, prove that pushing the stranger is morally permissible. However, Greene argues that empirical neuroscience can execute a devastating philosophical maneuver known as an Evolutionary Epistemic Debunking Argument. Debunking arguments do not fallaciously deduce an “ought” from an “is”; rather, they use empirical facts about the causal origins of our beliefs to undercut the epistemic reliability and normative authority of those beliefs. If you discover that your belief that aliens built the pyramids was implanted in your brain by a malfunctioning virtual reality helmet, you may not have proven that aliens didn’t build the pyramids, but you have destroyed your justification for believing that they did.

11.2 Greene’s Normative Argument for Utilitarianism

Armed with this debunking framework, Greene launched a profound attack on deontological philosophy in his famous meta-ethical manifesto, “The Secret Joke of Kant’s Soul” (2008), expanded in his book Moral Tribes: Emotion, Reason, and the Gap Between Us and Them (2013). Greene’s normative argument proceeds through a series of radical steps:

First, Greene argues that our deontological intuitions—our visceral convictions that certain rights cannot be violated, that direct harm is categorically wrong, and that individuals possess inviolable moral boundaries—are nothing more than the cognitive echoes of evolutionarily primitive emotional tripwires. These tripwires were installed by natural selection to regulate face-to-face physical aggression within ancestral hunter-gatherer bands. They are blind to mathematical consequences and responsive only to the arbitrary physical mechanics of up-close violence. When Immanuel Kant or modern neo-Kantian ethicists construct intricate, intellectualized systems of categorical imperatives, duties, and rights, they are not discovering eternal, objective truths about the moral cosmos. Rather, they are engaging in elaborate, post-hoc rationalizations, retrofitting high-minded philosophical jargon onto ancient primate gut reactions.

Second, Greene argues that our utilitarian intuitions emerge from an entirely different cognitive source: the domain-general, flexible, reflective cognitive control networks of the prefrontal cortex (dlPFC). Utilitarianism is not an evolutionary accident tuned to the kinematics of Stone Age violence. It is an impartial, objective mathematical procedure for maximizing aggregate flourishing and minimizing suffering across all conscious beings. Utilitarianism operates as a universal moral currency. Because it relies on the manual mode of human reason, it is uniquely equipped to transcend tribal divisions, overcome parochial emotional biases, and navigate the complex, non-zero-sum coordination challenges of a globalized world.

Greene concludes that deontology should be normatively abandoned. Our deontological intuitions are cognitive heuristics that fail catastrophically in modern ethical contexts. Instead, humanity must embrace what Greene calls “Deep Pragmatism”—a commitment to utilitarian maximization guided by prefrontal manual override, systematically discarding our ancestral emotional taboos whenever they obstruct the optimization of aggregate human and non-human welfare.

11.3 Heuristics, Biases, and Public Policy Implications

The practical applications of Greene’s neurobiological debunking framework extend far beyond academic philosophy; they strike at the heart of public policy, healthcare triage, and institutional design. Because the human brain’s automatic settings remain tuned to proximate physical violence, our moral intuitions suffer from severe, systematic cognitive distortions that distort legal codes and institutional governance.

Consider the allocation of scarce medical resources during a global pandemic. A strict utilitarian approach requires algorithmic triage: ventilators, ICU beds, and antiviral medications must be allocated to maximize Quality-Adjusted Life Years (QALYs) saved, even if this means withdrawing care from a terminal patient to preserve three salvageable lives. Yet, public policy frequently recoils from such cold, optimization protocols, crippled by deontological outrage that views withdrawing a ventilator as an impermissible act of “killing” (a direct personal harm), while viewing the death of three patients who never received a ventilator as a mere “letting die” (an indirect impersonal harm). Neuroscience exposes this distinction as an evolutionary illusion: from the perspective of objective aggregate welfare, both outcomes involve an identical causal responsibility for preventable death.

Similarly, Greene’s model illuminates our systemic moral paralysis regarding existential risks, climate change, and global poverty. We will deploy hundreds of thousands of dollars to rescue a single child fallen into an abandoned well (a proximate, identifiable, emotionally vivid personal crisis that lights up the mPFC-amygdala axis), while neglecting global aid programs that could save thousands of nameless children from malaria for a few dollars per bed net (an abstract, statistical, impersonal crisis that leaves our evolutionary alarm systems utterly unmoved). Overcoming these ruinous institutional biases requires the deliberate, codified institutionalization of manual-mode thinking: designing legal, medical, and political architectures that systematically check, neutralize, and override primate moral intuitions in favor of impartial, evidence-based utilitarian calculus.

12. Contemporary Frontiers and Future Horizons in Moral Neuroscience

12.1 Algorithmic Ethics and Autonomous Vehicle Dilemmas

In the twenty-first century, the trolley problem has made an unprecedented leap from philosophical classrooms and fMRI scanners directly into the automotive and robotics industry. With the advent of autonomous vehicles (AVs), software engineers must write deterministic algorithms that dictate how an automated machine will navigate unavoidable fatal collisions. If an autonomous vehicle’s brakes fail at high speed, should the algorithm be programmed to swerve into a concrete barrier (killing its single passenger) to avoid mowing down five pedestrians crossing the street? The trolley problem is no longer a hypothetical abstraction; it is compiled source code.

In 2018, researchers at the MIT Media Lab launched the Moral Machine experiment, an unprecedented global study published in Nature (Awad et al., 2018) that captured 40 million decisions from millions of people across 233 countries regarding autonomous vehicle crash scenarios. The findings revealed profound behavioral fractures that mirrored Greene’s dual-process tensions, while exposing deep cultural variations across Western, Eastern, and Southern global clusters.

Furthermore, the MIT study exposed a profound consumer paradox rooted in the dual-process architecture: when surveyed abstractly, citizens universally expressed a utilitarian preference, stating that autonomous vehicles ought to be programmed to minimize aggregate fatalities, sacrificing the passenger to save the pedestrians. However, when asked if they would personally purchase a vehicle programmed with this utilitarian algorithm, participants overwhelmingly rejected the machine, demanding that their own personal vehicle prioritize their personal survival at all costs. Algorithmic ethics has run headfirst into the conflict between impartial utilitarian reason and evolutionarily hardwired, self-preservation heuristics—a challenge that automotive manufacturers and regulatory bodies are struggling to resolve.

12.2 Advanced Methodological Advancements: Ultra-High Field fMRI and TMS

Technologically, moral neuroscience has advanced far beyond the initial 1.5-Tesla, univariate designs of Greene’s 2001 study. Today, the field is leveraging cutting-edge, multimodal neurotechnologies that provide unprecedented spatial and temporal resolution into the moral brain:

• Ultra-High Field 7-Tesla fMRI: High-field 7T scanners provide sub-millimeter isotropic resolution, allowing cognitive neuroscientists to image specific cortical layers (laminar fMRI) within the prefrontal mantle. Researchers can now observe how sensory inputs arrive in layer IV of the dlPFC and trace the micro-circuitry of top-down inhibitory signals descending from deep prefrontal layers to the amygdaloid nuclei during real-time moral conflict.
• Transcranial Magnetic Stimulation (TMS): Non-invasive brain stimulation allows researchers to establish direct, reversible causal interventions. By applying repetitive TMS (rTMS) or continuous theta-burst stimulation (cTBS) over the right dlPFC, researchers can transiently disrupt the executive deliberative engine. Studies show that when the dlPFC is magnetically knocked offline, subjects become significantly slower and less likely to deliver utilitarian judgments in high-conflict dilemmas. Conversely, stimulating the temporoparietal junction (TPJ) alters an agent’s capacity to integrate mental states, radically transforming how intentions versus outcomes are weighed in moral blameworthiness.
• Combined EEG-fMRI: By recording high-density electroencephalography (EEG) simultaneously inside the fMRI scanner, researchers can combine the millisecond temporal precision of event-related potentials (such as the early P2, N2, and Late Positive Potential components) with the exquisite spatial localization of fMRI BOLD signals. This reveals the exact temporal progression of moral cognition: demonstrating that affective salience processing in limbic structures occurs within 150 to 250 milliseconds post-stimulus, whereas controlled prefrontal deliberative calculations emerge only after 400 to 800 milliseconds.

12.3 Cross-Cultural Moral Neuroscience and Neurodiversity

Finally, the contemporary frontier of moral neuroscience is actively addressing one of the most critical structural flaws of twentieth-century behavioral science: the WEIRD sampling bias. As documented by Joseph Henrich and colleagues, the overwhelming majority of behavioral and neuroimaging studies have been conducted on populations that are Western, Educated, Industrialized, Rich, and Democratic—a demographic slice representing less than 12% of the global human population.

Cross-cultural moral neuroscience has begun mapping how cultural ecologies shape the underlying functional recruitment of the brain. Studies conducted in East Asian societies (such as Japan, China, and South Korea) reveal that moral dilemmas often recruit significantly higher levels of social-relational and mentalizing networks (precuneus, TPJ) compared to Western cohorts. In collectivist cultures, where the preservation of social harmony and relational duties takes precedence over both individualistic rights-constraints and abstract mathematical optimizations, the behavioral and neural signatures of the trolley problem shift demonstrably. Participants in these cultures are often less willing to endorse utilitarian trade-offs if the sacrificial individual is an in-group member, reflecting the profound evolutionary salience of tribal cohesion.

Simultaneously, the field is interrogating neurodiversity, specifically examining how moral evaluations operate in populations on the Autism Spectrum (ASD). Individuals with ASD, who often present with atypical Theory of Mind and social communication processing alongside highly intact systemizing and analytical cognitive styles, frequently exhibit a unique moral profile. In moral dilemmas, autistic individuals often demonstrate a higher reliance on explicit, rule-based utilitarian reasoning, unaffected by the subtle, social-emotional framing effects that skew neurotypical populations. Understanding these neurodiverse computational variations is essential as cognitive neuroscience works toward its ultimate horizon: a comprehensive, multi-level, computationally precise model of human moral agency that unites genetics, neuroanatomy, cultural transmission, and philosophical normative reflection into a single unified science of the human soul.

Conclusion

The journey from Philippa Foot’s runaway tramway to the ultra-high field neuroimaging laboratories of modern cognitive neuroscience represents one of the most profound intellectual transformations in the history of ideas. What began as a delicate, introspective thought experiment designed to unpack the Doctrine of Double Effect was transformed by Joshua Greene into an empirical probe that exposed the fractured, evolutionary fault lines of the human brain.

Greene’s Dual-Process Theory of Moral Judgment demonstrated that our deepest moral convictions are not the monolithic outputs of a transcendent, unified moral faculty, nor are they the pristine discoveries of pure rational introspection. Instead, our moral minds are biological palimpsests: ancient, subcortical emotional circuits—forged in the Pleistocene to enforce face-to-face cooperation and suppress immediate physical violence—uneasily coexisting with modern, prefrontal cognitive control networks capable of abstract, impartial utilitarian calculation. When we look upon the Footbridge dilemma and recoiled in instinctive horror, we are hearing the frantic ringing of an evolutionary alarm bell that evolved to prevent our primate ancestors from bludgeoning one another to death. When we look upon the Switch dilemma and calmly calculate that five lives outweigh one, we are deploying the flexible, manual machinery of prefrontal reason to optimize outcomes in a complex, non-zero-sum world.

This neurobiological insight does not diminish the gravity of human ethics; rather, it elevates it. By understanding the biological machinery of our conscience—by recognizing our visceral, deontological intuitions as evolutionary heuristics and our utilitarian calculations as deliberative manual overrides—we liberate ourselves from the tyrannical authority of our ancestral biases. As humanity faces an uncertain future characterized by algorithmic warfare, autonomous machines, global climate disruption, and catastrophic existential risks, our survival will depend on our ability to recognize when our ancient automatic settings are leading us astray, and our willingness to seize the prefrontal controls, engage the manual mode, and reason our way toward a more rational, impartial, and compassionate moral future.

References

• Awad, E., Dsouza, S., Kim, R., Schulz, J., Henrich, J., Shariff, A., Bonnefon, J. F., & Rahwan, I. (2018). The Moral Machine experiment. Nature, 563(7729), 59–64. https://doi.org/10.1038/s41586-018-0637-6
• Bentham, J. (1789). An Introduction to the Principles of Morals and Legislation. T. Payne and Son.
• Berker, S. (2009). The normative insignificance of neuroscience. Philosophy & Public Affairs, 37(4), 293–329. https://doi.org/10.1111/j.1088-4963.2009.01164.x
• Blair, R. J. R. (2007). The amygdala and ventromedial prefrontal cortex in morality and psychopathy. Trends in Cognitive Sciences, 11(9), 387–392. https://doi.org/10.1016/j.tics.2007.07.003
• Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict monitoring and cognitive control. Psychological Review, 108(3), 624–652. https://doi.org/10.1037/0033-295X.108.3.624
• Crockett, M. J., Clark, L., Hauser, M. D., & Robbins, T. W. (2010). Serotonin selectively influences moral judgment and behavior through effects on harm aversion. Proceedings of the National Academy of Sciences, 107(40), 17433–17438. https://doi.org/10.1073/pnas.1009396107
• Damasio, A. R. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. G.P. Putnam’s Sons.
• Foot, P. (1967). The problem of abortion and the doctrine of the double effect. Oxford Review, 5, 5–15.
• Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 19(4), 25–42. https://doi.org/10.1257/089533005775196732
• Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., & Cohen, J. D. (2001). An fMRI investigation of emotional engagement in moral judgment. Science, 293(5537), 2105–2108. https://doi.org/10.1126/science.1062872
• Greene, J. D., Nystrom, L. E., Engell, A. D., Darley, J. M., & Cohen, J. D. (2004). The neural bases of cognitive conflict and control in moral judgment. Neuron, 44(2), 389–400. https://doi.org/10.1016/j.neuron.2004.09.027
• Greene, J. D. (2008). The secret joke of Kant’s soul. In W. Sinnott-Armstrong (Ed.), Moral Psychology, Vol. 3: The Neuroscience of Morality: Emotion, Brain Disorders, and Development (pp. 35–79). MIT Press.
• Greene, J. D., Morelli, S. A., Lowenberg, K., Nystrom, L. E., & Cohen, J. D. (2008). Cognitive load selectively interferes with utilitarian moral judgment. Cognition, 107(3), 1144–1154. https://doi.org/10.1016/j.cognition.2007.11.004
• Greene, J. D. (2009). Dual-process morality and the personal/impersonal distinction: A reply to McGuire, Langdon, Coltheart, and Mackenzie. Journal of Experimental Social Psychology, 45(3), 581–584. https://doi.org/10.1016/j.jesp.2009.01.003
• Greene, J. D. (2013). Moral Tribes: Emotion, Reason, and the Gap Between Us and Them. Penguin Press.
• Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral judgment. Psychological Review, 108(4), 814–834. https://doi.org/10.1037/0033-295X.108.4.814
• Haidt, J., & Graham, J. (2007). When morality opposes justice: Conservatives have moral intuitions that liberals may not recognize. Social Justice Research, 20(1), 98–116. https://doi.org/10.1007/s11211-007-0034-z
• Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2-3), 61–83. https://doi.org/10.1017/S0140525X0999152X
• Hume, D. (1739). A Treatise of Human Nature. John Noon.
• Kahane, G., Everett, J. A., Earp, B. D., Farias, M., & Savulescu, J. (2015). ‘Utilitarian’ judgments in sacrificial moral dilemmas do not reflect impartial concern for the greater good. Cognition, 134, 193–209. https://doi.org/10.1016/j.cognition.2014.10.005
• Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
• Kant, I. (1785). Groundwork of the Metaphysics of Morals (M. Gregor, Trans.). Cambridge University Press (1998).
• Koenigs, M., Young, L., Adolphs, R., Tranel, D., Cushman, F., Hauser, M., & Damasio, A. (2007). Damage to the prefrontal cortex increases utilitarian moral judgements. Nature, 446(7138), 908–911. https://doi.org/10.1038/nature05631
• McGuire, J., Langdon, R., Coltheart, M., & Mackenzie, C. (2009). A reanalysis of the personal/impersonal distinction in moral psychology research. Journal of Experimental Social Psychology, 45(3), 577–580. https://doi.org/10.1016/j.jesp.2009.01.002
• Mikhail, J. (2007). Universal moral grammar: Theory, evidence and the future. Trends in Cognitive Sciences, 11(4), 143–152. https://doi.org/10.1016/j.tics.2006.12.007
• Mill, J. S. (1861). Utilitarianism. Parker, Son, and Bourn.
• Moore, G. E. (1903). Principia Ethica. Cambridge University Press.
• Navarrete, C. D., McDonald, M. M., Mott, M. L., & Asher, B. (2012). Virtual morality: Emotion and action in a simulated three-dimensional “trolley problem”. Emotion, 12(2), 364–370. https://doi.org/10.1037/a0025561
• Nozick, R. (1974). Anarchy, State, and Utopia. Basic Books.
• Poldrack, R. A. (2006). Can cognitive processes be inferred from neuroimaging data? Trends in Cognitive Sciences, 10(2), 59–63. https://doi.org/10.1016/j.tics.2005.12.004
• Thomson, J. J. (1976). Killing, letting die, and the trolley problem. The Monist, 59(2), 204–217. https://doi.org/10.5840/monist197659224
• Thomson, J. J. (1985). The trolley problem. The Yale Law Journal, 94(6), 1395–1415. https://doi.org/10.2307/796133