The studies cited herein are employed exclusively at the group and systems level to describe population-level mechanisms, risk patterns, and structural dynamics.
They are non-diagnostic, do not support individual attribution, and are not intended for clinical, predictive, or forensic identification of specific persons.
6.1 Scope and limits
Supports Chapter 6’s functional model (salience ? reinforcement ? inhibition attenuation ? normalization) using peer-reviewed neuroscience and forensic psychology.
Findings are group-level, task-dependent, and non-diagnostic. “Cruelty” is operationalized via adjacent constructs (aggression, antisocial traits, psychopathy traits, sadism traits, and responses to violent imagery). No individual attribution is implied.
No claim in this appendix establishes sufficiency: all mechanisms described are conditional, context-dependent, and non-predictive at the individual level.
6.2 Amygdala: salience and threat prioritization
The amygdala participates in rapid salience and threat detection and biases downstream learning systems toward stimuli tagged as significant. Across antisocial and psychopathy-related phenotypes, studies report atypical limbic responsivity and altered limbic–prefrontal connectivity during affective and moral tasks. These findings are consistent with disrupted affective learning and altered valuation of harm-related cues rather than a single “hyper/hypo” profile [124–126].
6.3 Ventral striatum: reward prediction and social reinforcement
The ventral striatum (VS) supports reinforcement learning and reward prediction error. Social rewards (status cues, feedback metrics, group approval) recruit VS circuitry, enabling non-material incentives to function as reinforcers. This mechanism generalizes to digital contexts where engagement signals and peer feedback act as reward proxies without ideological commitment [127–129].
6.4 vmPFC/OFC: value integration and inhibitory control
Ventromedial prefrontal cortex (vmPFC) and orbitofrontal cortex (OFC) integrate value, social norms, and consequence simulation. Lesion and stress-modulation literatures show that compromised vmPFC/OFC function can impair norm integration and inhibitory control while leaving general intelligence intact. Acute and chronic stress degrade prefrontal regulation, increasing reliance on habitual and reactive responding [130–133].
6.5 Exposure effects: habituation and desensitization
Repeated exposure to violent or aversive stimuli can produce habituation (reduced autonomic and neural response). Laboratory paradigms show reduced physiological reactivity following repeated exposure to violent media: effects vary by task, population, and context, and do not imply inevitability or universality. These findings support a friction-reduction account rather than a deterministic pathway [134–136].
6.6 Sadism vs. consensual BDSM: empirical separation
Sexual Sadism Disorder (SSD) requires recurrent sexual arousal from another’s suffering plus distress/impairment or non-consent: consensual BDSM does not meet criteria absent these conditions. Non-clinical constructs such as “everyday sadism” capture enjoyment of others’ suffering without paraphilic arousal. In cruelty networks where reinforcement is primarily status, clout, intimidation, or access, SSD is not the parsimonious primary model: sexualized content, when present, requires separate criteria and cannot be inferred from harm alone [137–140].
6.7 Hormonal modulation (non-deterministic)
Testosterone is associated with status-seeking and dominance behaviors in context-dependent ways. It is not a deterministic violence driver [141–142].
Cortisol/stress systems can impair prefrontal control under sustained arousal, biasing toward reactive responding [132, 143].
Serotonergic systems contribute to impulse control and aggression modulation: associations are complex and non-diagnostic [144–145].
6.8 Moral disengagement and compartmentalization
Mechanisms such as diffusion of responsibility, dehumanization, and euphemistic labeling enable participation in harm without global self-concept collapse. These mechanisms are empirically supported and help explain intact executive function alongside harmful participation [146–147].
6.9. Bounded inference rule
Where salience tagging (amygdala/insula), social reinforcement (VS), and reduced inhibitory integration (vmPFC/OFC) co-occur under stress, repeated exposure is associated with reduced friction to continued participation. This describes a mechanism of risk, not a diagnosis.
Constraint Set, Chapter 6 (When The Brain Becomes Arena)
This addendum translates the neurobehavioral mechanisms described in Chapter 6 into systemic-risk categories legible to regulatory, supervisory, and institutional audiences. It does not diagnose individuals. It specifies risk pathways and duty-bearing loci.
6.10 Risk Propagation Mechanism
Repeated exposure to extreme violent or humiliating material within engagement-optimized environments recalibrates salience and reinforcement systems while attenuating inhibitory control. When social reward (status, visibility, access) is contingently paired with cruelty-linked cues, participation becomes self-stabilizing at the population level. Risk propagates through neurobehavioral adaptation synchronized with platform distribution dynamics, not through isolated intent.
6.11 Foreseeable Amplification Pathway
Recommendation systems that optimize for retention and interaction reliably surface content that elicits strong affect. Where such systems repeatedly deliver extreme material and couple it with social reinforcement signals, escalation and desensitization are foreseeable outcomes. The pathway (exposure ? habituation ? reward coupling ? drift into participation) is predictable, recurrent, and observable across platforms.
6.12 Failure of Duty of Care
A duty-of-care failure occurs where platforms maintain architectures that (a) predictably reinforce harmful engagement loops, (b) fail to introduce effective friction after repeated exposure signals, and (c) rely on post hoc content removal while leaving reinforcement contingencies intact. Mitigation that addresses instances but not learned reward pathways is insufficient where risk mechanisms are known and measurable.
6.13 Structural Selection Effect
Platform architectures selectively advantage behaviors that convert domination, shock, or humiliation into engagement metrics. This constitutes a structural selection effect: systems disproportionately retain and amplify actors and content that exploit reinforcement loops, irrespective of moral or legal classification. Cruelty persists not due to rarity or deviance, but because it is competitively selected under prevailing optimization criteria.
Regulatory Implication
Where systemic design predictably trains salience and reward systems toward harmful participation, risk attribution extends beyond individual actors to infrastructural choices. Effective mitigation requires altering reinforcement conditions, exposure gradients, and escalation affordances, not solely removing content after harm has been instantiated.