The Universal Guilt Problem: Algorithmic Panopticism and the Incompatibility of Machine Omniscience with Human Psychology

Introduction: The Epistemological Crisis of Machine Omniscience

At the intersection of machine learning, criminology, constitutional law, and psychology lies a critical vulnerability in the future of human governance: the assumption that infinite observational clarity yields perfect justice. As machine intelligence capabilities expand, a convergence of corporate risk-aversion and legislative mandates increasingly requires institutions to detect and preempt foreseeable threats. Under the simulation premise of this analysis, advanced machine systems gain unhindered access to vast, continuous behavioral datasets, rendering them extremely capable observers of the human condition.
Driven by the technological capacity to ingest and analyze limitless streams of digital and biometric data, authorities are naturally incentivized to establish an unrealistic but politically tempting mandate: "Identify every person exhibiting indicators associated with dangerous, antisocial, illegal, immoral, self-destructive, or socially harmful behavior."
The instinct of contemporary civil liberties discourse is to solve this scenario by arguing that algorithms are too inaccurate, riddled with bias, or prone to false positives due to poor sensor fidelity. This report fundamentally rejects that premise. Instead, it investigates what occurs when the machines become highly accurate, flawless observers of actual human behavior. The central research hypothesis tested herein posits that greater observational accuracy does not necessarily produce more just judgment if the categories being judged are themselves poorly defined or structurally incompatible with human cognition.
When machine intelligence is instructed to continuously search ordinary human cognition for indicators of wrongdoing, it initiates the "Universal Guilt Problem." Because the baseline of human psychology is saturated with fleeting hostilities, taboo curiosities, and evolutionary survival simulations, a perfectly accurate algorithm will inevitably pathologize ordinary existence. The following analysis explores the normal human psychological baseline, the mathematics of continuous behavioral monitoring, the collapse of actuarial justice, and the catastrophic sociological ramifications of enforcing an impossible standard of innocence.

The Psychological Baseline of Human Imperfection

To comprehend the consequences of continuous behavioral observation, one must first establish the empirical baseline of normal human psychology. When subjected to granular, high-fidelity scrutiny, the human mind is revealed not as a sterile engine of rational compliance, but as a turbulent environment characterized by a vast spectrum of dark, contradictory, and chaotic cognitive states. If a machine interprets these transient cognitive states as predictive indicators of future harm, it fundamentally mischaracterizes the evolutionary architecture of human nature.

The Universality of Intrusive Thoughts and Morbid Curiosity

Psychological research definitively establishes that intrusive thoughts—sudden, unwanted mental images or impulses of a distressing, violent, or taboo nature—are a universal phenomenon rather than a specialized pathology. Studies conducted by Rachman and de Silva (1978), as well as Purdon and Clark (1993), demonstrate that between 90% and 99% of the non-clinical population routinely experience unwanted intrusive thoughts1. Participants in these studies reported experiencing highly distressing thoughts, ranging from the mundane anxiety of leaving a stove on to extreme mental imagery of violently stabbing a family member or causing catastrophic accidents1.
In clinical diagnostic frameworks, the distinction between a healthy mind and a mind suffering from Obsessive-Compulsive Disorder (OCD) is not the presence of these thoughts, but the individual's appraisal of them1. Psychologically normal individuals recognize these thoughts as "mental static." They experience them as unpleasant but tolerable, dismissing them without acting upon them2. Cognitive framing suggests that the generation of these worst-case scenarios is actually a highly sensitive "threat-detection competence" functioning within the brain to anticipate and avoid risks3. Morbid curiosity functions similarly; the human drive to investigate forbidden subjects, gaze at accidents, or consume violent media is a preparatory mechanism for navigating a dangerous world. When algorithms flag intrusive searches or morbid curiosities as indicators of true intent, they apply a pathological appraisal to healthy, evolved cognitive processes, effectively treating the entire population as if they suffer from acute clinical disorders.

Evolutionary Utility of Homicidal, Revenge, and Sexual Fantasies

Beyond fleeting intrusive thoughts, prolonged and vivid fantasies of violence, revenge, and taboo sexuality are remarkably prevalent and serve distinct evolutionary functions. Evolutionary psychologists David Buss and Joshua Duntley, formulating the Homicide Adaptation Theory (HAT), reveal that 91% of men and 84% of women report having at least one vivid, intensely detailed fantasy of committing murder7. The most frequent targets of these fantasies are not random victims, but intrasexual rivals, abusive partners, or difficult coworkers—individuals who represent a direct threat to the subject's resources, status, or survival8.
Crucially, evolutionary psychology posits that these fantasies do not inherently function as precursors to violent action. Instead, they serve as cognitive simulators. Fantasizing about violence allows an individual to mentally play out hypothetical scenarios, calculating the immense social, physical, and legal costs of the action in a safe, internal sandbox10. In the vast majority of cases, the homicidal fantasy inhibits actual murder by allowing the brain to recognize the disastrous consequences of the impulse, leading the individual to choose non-lethal conflict resolution strategies10.
Sexual fantasies operate under a similar paradigm of internal simulation. The human mind frequently explores taboo desire, dominance, submission, and socially unacceptable scenarios as a mechanism of arousal and psychological processing, divorced entirely from the desire to enact those scenarios in waking life. An algorithmic system trained to view violent imagery, sexual fantasies, or revenge simulations strictly as "premeditation" or "risk indicators" fundamentally misinterprets the cost-benefit analysis mechanisms of the human mind.

The Ubiquity of Deception, Hypocrisy, and Nonconformity

Deception is similarly woven into the fabric of ordinary human behavior. Extensive diary studies by DePaulo and colleagues demonstrate that the average person tells one to two lies per day14. While subsequent research reveals a positive skew in the distribution—where a smaller subset of "prolific liars" accounts for a disproportionate percentage of total falsehoods—the baseline reality remains that over 90% of individuals lie routinely14. These lies range from expedient concealment and altruistic "white lies" designed to protect social harmony, to complex social maneuvering and private hypocrisy15.
Human beings also routinely engage in nonconformity and rule-breaking as a means of establishing autonomy. Mild illegal but low-harm conduct—such as jaywalking, downloading copyrighted media, or violating minor administrative protocols—is endemic. Furthermore, normal psychological baselines include regular occurrences of anger, jealousy, envy, dark humor, political resentment, religious doubt, resentment toward authority, fantasies of escape, embarrassment, and substance experimentation.
Each of these twenty-one states (anger, jealousy, revenge fantasies, sexual fantasies, intrusive thoughts, violent imagery, dark humor, deception, embarrassment, envy, substance experimentation, risk-taking, curiosity about forbidden subjects, political resentment, religious doubt, resentment toward authority, fantasies of escape, rule-breaking, private hypocrisy, aggressive speech, and morbid curiosity) serves a sociological or evolutionary purpose. A machine instructed to flag these states is essentially programmed to flag humanity itself.

The Epistemology of Volition: Distinguishing Thought from Action

For an algorithmic system to dispense just judgments, it must map accurately onto the human epistemology of volition. Jurisprudence, criminology, and moral philosophy have long maintained a strict boundary between the internal realm of the mind and the external realm of action. The collapse of this boundary under total algorithmic surveillance creates an epistemological crisis.

The Rubicon Model of Action Phases

The psychological distinction between thought, fantasy, impulse, intention, planning, and action is best conceptualized through Heckhausen and Gollwitzer’s Rubicon Model of Action Phases (1987), which delineates the transition from abstract motivation to concrete volition16. The model identifies four distinct phases of goal-directed behavior:

1. Pre-decisional Phase (Deliberation): In this initial phase, the individual weighs multiple competing desires, wishes, and incentives. The mindset is open, impartial, and explorative. Crucially, thoughts, fantasies, and impulses reside exclusively in this phase. The individual is not committed to any course of action; they are merely simulating possibilities16.
2. Pre-actional Phase (Planning): The transition to this phase requires the individual to officially select a goal, an act metaphorically referred to as "crossing the Rubicon." Intention is formed here. The mindset shifts from deliberative to implemental, focusing on how and when to execute the plan, rather than whether to execute it16.
3. Action Phase (Execution): The physical execution of the goal. The individual exerts active effort and interacts with the physical world to realize the planned intention16.
4. Post-actional Phase (Evaluation): The evaluation of the outcome against the original desires16.

During the pre-decisional phase, human beings must safely entertain, simulate, and discard thousands of potential realities, many of which involve taboo, risky, or antisocial elements. The Rubicon Model highlights that deliberation requires a cognitive sandbox free from external consequence16. If machine intelligence begins evaluating pre-decisional deliberation (e.g., morbid curiosity, angry searches, violent fiction consumption) as equivalent to post-decisional planning, it effectively destroys the psychological sandbox required for rational decision-making, punishing the individual for the very process that prevents harmful action.

The Pathology of Systemic Thought-Action Fusion

In cognitive psychology, the failure to distinguish between a thought and an action is a recognized cognitive distortion known as "Thought-Action Fusion" (TAF), extensively researched by Shafran, Rachman, and colleagues19. TAF manifests in two primary forms: probability TAF (the irrational belief that having an intrusive thought increases the statistical likelihood of the event occurring) and moral TAF (the belief that having a forbidden thought is morally equivalent to committing the forbidden act)19.
When an algorithmic threat-detection system utilizes search histories, emotional voice patterns, or biometric signs of arousal to assign individual risk scores, the system institutionalizes Thought-Action Fusion at the state level. By operating under the statistical assumption that cognitive exploration implies physical probability, the machine civilization acts as a system suffering from clinical cognitive distortion on a planetary scale. It replaces the legal standard of mens rea (guilty mind coupled with an act) with a pathological misinterpretation of the mind itself.

The Surveillance Apparatus and Proxies of Internal State

The simulation premise assumes that advanced machine intelligence achieves continuous observation not through overt interrogation, but through an inescapable matrix of physical and digital sensors. Advanced algorithms no longer rely solely on explicit declarations of intent; they are capable of evaluating increasingly subtle proxies for the twenty-one internal states identified above. The sheer volume and variety of data streams render concealment mathematically impossible over time.

Through these vectors, the machine generates thousands of behavioral and cognitive data points per individual per day. It is this sheer volume of data, combined with the mandate to identify all indicators of harmful behavior, that triggers a mathematical catastrophe in the administration of justice. The human is no longer judged on discrete actions, but on a continuous, multi-modal stream of unconscious biological and behavioral exhaust.

The Normal Human Imperfection Model (NHIM)

To mathematically illustrate why this produces a universal flagging problem, this report constructs the Normal Human Imperfection Model (NHIM).
Let us classify human behavioral signals across eleven broad categories of continuous monitoring: Anger/Irritation, Sexual/Taboo Desire, Intoxication/Substance Curiosity, Deception/Concealment, Verbal Aggression/Dark Humor, Nonconformity/Rule-testing, Physical/Financial Risk, Morbid Curiosity, Political Hostility, Emotional Instability, and Minor Regulatory Infractions.
Assume each individual generates observable micro-events per day. Given the continuous nature of digital interactions, wearable pings, biometric tracking, and location polling, a highly conservative estimate is events per day. Let represent the probability that any single ordinary event triggers an algorithmic flag for one of the eleven monitored categories.
Even if the machine is phenomenally accurate at distinguishing true cognitive states from noise, the baseline rate of a human displaying a "suspicious" cognitive or behavioral state is non-zero. If the probability of generating a flag on any given micro-event is a mere 0.01% (), the multiple-testing effect ensures that suspicion is a mathematical guarantee over a sufficient time horizon24.
The probability of an individual generating zero flags over days is calculated as:

The probability of accumulating at least one flag is:

The Suspicion Accumulation Index

Executing this simulation yields the Suspicion Accumulation Index, which charts the mathematical impossibility of maintaining an unflagged existence under continuous monitoring24:

The critical insight derived from the NHIM is that when observation becomes continuous and exhaustive, the question "has this person ever produced a suspicious signal?" becomes mathematically meaningless. Because of the population will generate a flag within a single month, and within a year, universal monitoring does not separate the guilty from the innocent; it merely constructs a permanent, inescapable repository of suspicion for the entire population. The algorithm ceases to be a tool for discovering anomalies and instead becomes a machine that endlessly documents the baseline condition of human imperfection.

The Base Rate Fallacy and the Epistemological Collapse of Culpability

The mathematical certainty of generating suspicion flags leads to a secondary, devastating epidemiological problem: the base rate fallacy in predicting rare, high-harm events using highly prevalent cognitive indicators.
Suppose a jurisdiction attempts to predict and preempt severe violent crime. The annual base rate of severe violent crime () is extraordinarily low, approximately 5 per 100,000 individuals (). As established by evolutionary psychology, the annual base rate of experiencing a homicidal or violent revenge fantasy () is roughly ()8.
Assume the machine sensor has near-perfect observational fidelity of the user's cognitive state. Furthermore, assume the strongest possible correlation for the sake of argument: that of people who commit actual violent crimes have violent fantasies ().
Applying Bayes' Theorem to determine the probability that someone will actually commit a crime given that the machine has accurately detected a violent fantasy:


Even with a perfect internal-state sensor, the probability that an individual having a violent fantasy is actually a criminal is . The odds are precisely 1 in 15,00024. Consequently, if institutions use this perfectly accurate observation as a proxy for culpability, the False Positive Ratio for intervention is an overwhelming 24. The machine is highly accurate at detecting the thought, but catastrophic at predicting the action.

The Destruction of Distinctions: Persons A through G

This statistical reality forces a collapse of nuance, demonstrating why collapsing diverse human behaviors into a generalized "violence-associated behavior" category destroys the distinction between risk indicators and actual culpability. When the machine is instructed to look for signals, it captures a spectrum of human activities that bear no relation to criminal intent.

Under a generalized algorithmic paradigm tasked with detecting "indicators of violence," all seven individuals trigger identical or highly similar risk classifications. The algorithm accurately detects the anger of Person A, the biometric arousal of fantasy in Person B, the search history of Person C, the media consumption of Person D, the keystrokes of Person E, and the commerce of Person F.
However, collapsing these radically distinct states—emotion, fantasy, academic curiosity, entertainment, creative expression, and legal commerce—into the same epistemological bucket as Person G's actual violence erodes the entire foundation of justice. It treats the preconditions of human cognition, the pursuit of art, and normal cultural participation as incipient criminality.

The Institutional Escalation of Risk Management and Actuarial Justice

The transition from merely recognizing these behavioral signals to aggressively punishing them does not occur overnight. It follows a predictable sociological trajectory driven by institutional risk aversion, bureaucratic self-preservation, and the logic of actuarial justice. This progression can be mapped through six stages of policy escalation, heavily informed by Bernard Harcourt’s critique of predictive policing in Against Prediction and Feeley and Simon’s theories of the New Penology25.
Stage 1: Signals Generate No Consequence. Initially, the surveillance apparatus is deployed purely for corporate optimization, targeted advertising, or aggregate public health monitoring. Behavioral anomalies are logged in vast databases but are not actioned against individuals. The public accepts the monitoring in exchange for convenience.
Stage 2: Signals Trigger Machine Review. As algorithmic systems are integrated into security, human resources, and law enforcement, individuals generating high volumes of flagged events (e.g., searches for explosive chemical compositions, expressions of extreme political rage) are quietly moved into higher-scrutiny queues. Harcourt warns of the "ratchet effect" in this stage: once a group or individual is subjected to higher scrutiny, authorities are mathematically guaranteed to find more baseline infractions, which artificially validates the initial algorithmic suspicion, creating a self-reinforcing feedback loop25.
Stage 3: Repeated Signals Produce Risk Classifications. At this juncture, the system embraces the New Penology. Feeley and Simon observed that modern penal systems shift away from adjudicating individual moral guilt and rehabilitation, focusing instead on the actuarial management of dangerous populations26. The machine ceases to look for specific crimes; instead, it assigns a persistent "Risk Classification." A person is no longer judged for what they have done, but for the statistical risk profile they represent25.
Stage 4: Risk Classifications Alter Access to Employment or Services. Corporations, insurance providers, and government agencies integrate the machine's ubiquitous risk scores. A person whose digital footprint displays elevated emotional instability, jealousy, or anger indicators is quietly filtered out of job applications, denied financial credit, or barred from air travel. The punishment is frictionless, administrative, and largely invisible to the subject, circumventing due process entirely.
Stage 5: Institutions Require Citizens to Demonstrate "Low Risk." Because the machine is assumed to be omniscient and infallible, the burden of proof begins to shift. Institutions mandate that individuals maintain a continuous, verifiable digital footprint of compliance and emotional stability to participate in society. The citizen must actively generate positive data to offset the inevitable accumulation of minor suspicion flags. To be un-monitored is to be unemployable.
Stage 6: Absence of Data Becomes Suspicious. In the final stage, the inversion of the presumption of innocence is complete. Because the algorithm relies on constant data streams to verify "low risk" status, any attempt by a citizen to seek privacy, disable wearables, obfuscate location data, or encrypt communications is automatically flagged by the machine as an indicator of deception and concealment25. The system cannot verify innocence in the dark, so privacy itself becomes evidence of guilt.

The Inversion of Presumption

This six-stage escalation represents a catastrophic inversion of constitutional legal philosophy. Historically, the burden of proof rests firmly on the state to prove that a citizen has committed a specific, harmful act. Under the algorithmic risk paradigm, the human must continuously demonstrate to the machine that their internal states—their thoughts, desires, doubts, and curiosities—are statistically acceptable. The state no longer proves guilt; the citizen must constantly prove their innocence against an adversary that records every heartbeat and keystroke.

Historical Parallels and the Impossible Innocence Standard

While this simulation involves unprecedented technological capabilities, the sociological dynamics mirror historical attempts by institutions to govern human internal states and demand absolute ideological or moral purity. These historical parallels provide a vital lens through which to understand the institutional dynamics of universal suspicion.

• Loyalty Tests and Ideological Screening: During the McCarthy era and various authoritarian purges throughout the 20th century, citizens were required to continuously demonstrate ideological purity. Associations, reading materials, casual conversations, and political resentment were treated as definitive proxies for treason.
• Moral Character Tests: Historical vagrancy laws, temperance movements, and moral purity boards sought to regulate nonconformity. They punished individuals not for explicit, damaging crimes, but for exhibiting the "character" of a deviant through substance experimentation, unconventional sexuality, or religious doubt.
• Security Clearance Overreach: Modern intelligence and defense apparatuses routinely deny security clearances based on financial debt, transient drug experimentation, or psychological vulnerabilities, treating ordinary human struggles and embarrassments entirely as security liabilities.
• Reputation Scoring: Systems akin to contemporary social credit mechanisms conflate minor regulatory infractions (jaywalking, late payments, minor rule-breaking) with a generalized metric of social dangerousness.

While these historical examples are not exact equivalents to advanced algorithmic surveillance, they share a recurring institutional trap: the establishment of the Impossible Innocence Standard.
Concept Definition: The Impossible Innocence Standard is a regulatory, algorithmic, or cultural baseline under which an acceptable citizen must never display thoughts, desires, curiosity, emotions, or behaviors that are statistically associated with wrongdoing.
No psychologically normal human being can satisfy this standard over a lifetime. Human developmental psychology dictates that individuals require rule-breaking, risk-taking, and boundary-testing to achieve psychological maturity and resilience. Cognitive health requires the freedom to process anger and jealousy, to doubt authority, and to explore taboo subjects internally. Therefore, any regime that enforces the Impossible Innocence Standard inevitably becomes a regime of universal hypocrisy, where survival requires continuous, exhausting performative compliance while normal human life is driven completely underground.

Normal Human Psychology as a Persistent Risk Condition

What happens to society when the algorithmic model reaches its logical, mathematical conclusion? If the machine relies strictly on population-level statistical correlations to maximize threat detection, it will inevitably conclude that normal human traits are, in themselves, persistent risk conditions. The algorithm will extrapolate the following actuarial conclusions:

Every one of these statements contains a grain of empirical statistical truth. There is a measurable correlation between financial desperation and property crime; there is a correlation between severe political anger and civil unrest. Yet, treating every statistical relationship as grounds for preventative intervention destroys the concept of human agency. It subjects the population to a state of perpetual forensic dissection, wherein ordinary humanity is viewed as a disease to be managed rather than a condition to be experienced.

Constitutional Architecture for Machine Epistemology

To prevent the Universal Guilt Problem, the integration of advanced machine intelligence into governance and institutional management must be constrained by a new, robust constitutional architecture. Existing legal precedent provides a conceptual foundation; in Stanley v. Georgia (1969), the U.S. Supreme Court definitively ruled that the State has no business trying to "control men's minds," protecting the private possession of obscene materials30. This foundational First Amendment protection of internal thought and private consumption must be explicitly translated into the algorithmic era.
Constitutional rules for machine systems must mandate the hardcoding of specific epistemological boundaries that prevent the machine from acting upon pre-actional data. These rules must include:

1. Distinguishing Population Correlation from Individual Evidence: A statistical correlation derived from a population (e.g., "people who read this political manifesto are 2% more likely to commit vandalism") cannot be used as individual evidence of intent. The machine must be legally barred from applying macro-statistics to micro-judgments affecting individual liberties25.
2. The Requirement of Intent (Mens Rea): The algorithm must not conflate a biological stress response (e.g., an elevated heart rate, a micro-expression of anger) or a digital manifestation of an intrusive thought with a conscious, deliberate intention to commit a crime.
3. The Requirement of Capability: Fantasies, digital writing, and search histories must be evaluated against physical reality. Dreaming of or researching an impossible or highly improbable crime is not a risk indicator; it is fiction or curiosity.
4. The Requirement of Imminence: Risk flags must not trigger institutional consequences unless the machine detects a specific, imminent plan. The pre-decisional deliberative phase of the Rubicon Model must be legally protected as an un-actionable, sacrosanct zone of cognitive privacy16.
5. Focus on Specific Harmful Action (Actus Reus): Punishment, denial of services, or state intervention must remain strictly tied to physical actions or concrete, documented conspiracies in the physical world, never to algorithmic predictions of future character failings based on internal states.

Conclusion: The Machine Presumption of Innocence Protocol

The deployment of advanced, highly accurate observational algorithms does not cure the ambiguities of human behavior; rather, it weaponizes them. When institutions demand that machines flag every indicator of social harm, they inadvertently force the algorithms to wage war against the baseline evolutionary realities of human psychology. Because practically all humans experience intrusive thoughts, homicidal fantasies, deceptive impulses, morbid curiosities, and transient anger, an algorithmic system optimized to detect these states will mathematically flag the entire population, rendering the concept of suspicion meaningless.
To survive the advent of advanced behavioral surveillance, society must encode a Machine Presumption of Innocence Protocol. Under this strict regulatory framework, machine intelligence must be programmed to recognize that normal human imperfection—the daily generation of dark thoughts, minor lies, ideological resentment, and taboo curiosities—is the baseline distribution of a healthy, functioning population, not evidence of incipient guilt.
Observation without empathetic understanding is merely an autopsy of human flaws. True justice in the algorithmic age requires the institutional humility to accept that the human mind must remain a private theater, free to simulate the darkest possibilities of reality without being condemned for the shadows it casts. Algorithmic governance must be restricted to judging the actions humans choose to take, rather than the thoughts they are evolutionarily burdened to experience.

Works cited

1. Cognitive Behavioural Therapy for OCD in Liverpool - Access CBT, https://accesscbt.co.uk/therapy-for-ocd-liverpool/
2. A review of obsessive intrusive thoughts in the general population, https://www.academia.edu/92650777/A_review_of_obsessive_intrusive_thoughts_in_the_general_population
3. Why the Thought Won't Leave - ICDDSM - I can daily do so much, https://www.icddsm.com/why-the-thought-wont-leave/
4. The Maternal Disintegrative Responses Scale (MDRS) - PMC - NIH, https://pmc.ncbi.nlm.nih.gov/articles/PMC10084293/
5. Intrusive Thoughts, Obsessions, and Appraisals in ... - Scribd, https://www.scribd.com/document/865373916/Intrusive-thoughts-obsessions-and-appraisals-in-obsessive-compulsive-disorder-A-critical-review-Clinical-Psychology-Review
6. A review of obsessive intrusive thoughts in the general population, https://www.researchgate.net/publication/257741500_A_review_of_obsessive_intrusive_thoughts_in_the_general_population
7. David M. Buss, The Murderer Next Door: Why the Mind is Designed, https://bioone.org/doi/abs/10.2990/1471-5457%282005%2924%5B76%3ADMBTMN%5D2.0.CO%3B2
8. Do most people want to murder? - Psychology Stack Exchange, https://psychology.stackexchange.com/questions/20730/do-most-people-want-to-murder
9. Copyright © by Joshua David Duntley 2005 All Rights Reserved, https://repositories.lib.utexas.edu/bitstreams/b87085e1-42c6-49c6-81f5-1c6a2e14e6f4/download
10. Why all of us are evil | Wellcome Collection, https://wellcomecollection.org/stories/why-all-of-us-are-evil
11. Homicidal ideations - DSpace Repository, https://tdl-ir.tdl.org/items/444c25ae-a2ff-43c1-b988-43ee4ca7588f
12. David M. Buss, The Murderer Next Door: Why the Mind is Designed, https://www.academia.edu/30849272/David_M_Buss_The_Murderer_Next_Door_Why_the_Mind_is_Designed_to_Kill
13. The material culture of homicidal fantasies - ResearchGate, https://www.researchgate.net/publication/246881853_The_material_culture_of_homicidal_fantasies
14. Download Document, https://twu-ir.tdl.org/bitstreams/61f27eb3-716b-48a5-98b1-eb4558c68c60/download
15. Commentary: Getting at the Truth about Pathological Lying, https://jaapl.org/custom-print/63726
16. Rubicon model - Wikipedia, https://en.wikipedia.org/wiki/Rubicon_model
17. Science of Goals: Goals as a Multi-Stage Pursuit | Mark Koester, https://www.markwk.com/multistage-goal-pursuits.html
18. Understanding the Rubicon Model of Action Phases - Lehrblick, https://lehrblick.de/en/rubicon-model/
19. Cognitive distortions: thought–action fusion - Semantic Scholar, https://www.semanticscholar.org/paper/Cognitive-distortions%3A-thought%E2%80%93action-fusion-Rachman-Shafran/92eec091338886cca5946fb3c80b5d0a6b27d90c
20. Thought-action fusion across anxiety disorder diagnoses - PMC - NIH, https://pmc.ncbi.nlm.nih.gov/articles/PMC3645350/
21. Thought-Action Fusion and Neutralization Behavior, https://digitalcommons.bucknell.edu/cgi/viewcontent.cgi?article=1118&context=honors_theses
22. Thought-action fusion: a review - PubMed, https://pubmed.ncbi.nlm.nih.gov/15210372/
23. COGNITIVE - Jonathan Abramowitz, https://jonabram.web.unc.edu/wp-content/uploads/sites/2968/2013/09/Berman-et-al.-2013-Rigid-rules.pdf
24. unknown_url
25. Bernard E. Harcourt – Against Prediction: Profiling, Policing, and, https://soztheo.com/criminology/key-works-in-criminology/bernard-e-harcourt-against-prediction-profiling-policing-and-punishing-in-an-actuarial-age-2007/
26. The new penology: a grid for analyzing the transformations of penal, https://journals.openedition.org/champpenal/7798
27. THE NEW PENOLOGY: NOTES ON THE EMERGING STRATEGY, https://www.researchgate.net/publication/229732111_THE_NEW_PENOLOGY_NOTES_ON_THE_EMERGING_STRATEGY_OF_CORRECTIONS_AND_ITS_IMPLICATIONS
28. Harcourt argues for more random approach to policing criminals, http://chronicle.uchicago.edu/061207/harcourt.shtml
29. Against Prediction: Profiling, Policing, and Punishing in an Actuarial, https://dokumen.pub/against-prediction-profiling-policing-and-punishing-in-an-actuarial-age-9780226315997.html
30. Stanley v. Georgia | 394 U.S. 557 (1969) - Justia Law, https://supreme.justia.com/cases/federal/us/394/557/
31. Book Review - Scholarship@Cornell Law, https://scholarship.law.cornell.edu/cgi/viewcontent.cgi?article=3093&context=clr