Continues from Why LLM Villains Turn Nice.
The previous entry diagnosed the problem: the cognitive architecture has no model for how characters view other people. Villains drift prosocial because the appraisal system can’t distinguish power-pleasure from warmth-pleasure. This entry builds the fix — a unified model that connects childhood experience to adult personality to per-relationship emotional dynamics, drawing on nine established psychological frameworks and decades of validated research.
How does Character A feel about and behave toward Character B — and why?
The answer requires five things the architecture currently lacks:
These five map to five layers. Each builds on the one below.
Before describing the layers, the causal sequence that connects them. This is validated by longitudinal research (Simard et al. 2011, Caspi et al. 2005, Shiner & Caspi 2003):
Infant temperament (genetic + prenatal)
↓ caregiver responsiveness × temperament fit
Attachment formation (ages 0–3)
↓ repeated patterns of unmet core emotional needs
Schema development (ages 0–12)
↓ schemas + temperament + ongoing experience
Trait crystallisation (adolescence → early adulthood)
↓ traits + schemas + attachment style
Relational patterns (adulthood)
↓ personality-filtered appraisal of each interaction
Moment-to-moment behavior
Two principles govern the chain:
Equifinality — multiple pathways reach the same outcome. High Callousness can emerge from temperamental fearlessness (primary psychopathy) or from chronic abuse that numbs empathic response (secondary psychopathy). The trait looks the same; the formation memory differs.
Multifinality — the same starting point can produce different outcomes depending on environment. High negative emotionality in infancy leads to anxious attachment with an inconsistent caregiver, avoidant attachment with a rejecting one, or secure attachment with a responsive one. Temperament sets the range; environment determines the position.
The architecture models both by treating formation memories as distinct from the traits they produce. Two characters can share the same AMPD profile but have different formation memories — and those memories give the LLM different material to reason from when generating behaviour.
Schema Therapy (Young 1990, 2003), Transactional Analysis (Berne 1961), Object Relations (Klein, Winnicott, Kernberg).
Unlike personality type systems (MBTI, Enneagram) which describe what you ARE, Schema Therapy describes what you BECAME and WHY. Cluster B traits aren’t innate labels — they’re the crystallised result of childhood experiences meeting temperament. A child who learns “I only get attention when I’m in control” develops manipulativeness. After ages 7–10, the patterns crystallise. The memories are the cause; the traits are the effect.
Young identified five core emotional needs. When these go repeatedly unmet in childhood, they produce Early Maladaptive Schemas — stable patterns of belief, emotion, and expectation that persist into adulthood:
| Unmet need | What the child lacked | Example schemas formed |
|---|---|---|
| Secure attachment | Safety, stability, nurturance, acceptance | Abandonment, Mistrust/Abuse, Emotional Deprivation |
| Autonomy and competence | Support for independence | Dependence, Failure, Vulnerability to Harm |
| Realistic limits | Appropriate boundaries | Entitlement, Insufficient Self-Control |
| Freedom to express needs | Emotional validation | Subjugation, Self-Sacrifice, Approval-Seeking |
| Spontaneity and play | Permission to be a child | Emotional Inhibition, Unrelenting Standards, Punitiveness |
Eighteen schemas across five domains. Each carries a core belief (“people who are kind always want something”), emotional triggers (situations that activate the schema), and coping modes (surrender, avoidance, or overcompensation).
The research (Bernstein et al. 2007, Lobbestael et al. 2005/2008) identifies distinctive schema profiles for each Cluster B personality disorder:
Antisocial/Psychopathic: Mistrust/Abuse + Entitlement + Punitiveness + Emotional Deprivation. The combination is specific: “people will hurt you” + “rules don’t apply to me” + “mistakes deserve punishment” + “my emotional needs don’t matter.” These four schemas, emerging from abuse, neglect, and inconsistent limits, produce the callous-manipulative adult.
Narcissistic: Emotional Deprivation + Defectiveness/Shame + Entitlement (as overcompensation). The narcissistic surface — grandiosity, entitlement, attention-seeking — is a defence against a fragile core of feeling defective and emotionally deprived. Three subtypes: fragile entitlement (compensation for defectiveness), pure entitlement (spoiled, no underlying vulnerability), and dependent entitlement (special therefore deserving of care).
Borderline: Abandonment + Mistrust/Abuse + Defectiveness/Shame + Insufficient Self-Control. Rapid mode switching — Vulnerable Child to Angry Child to Detached Protector in under a minute — is the distinctive computational pattern.
Modes are the mechanism by which the same person shows different faces depending on which schemas are activated. They’re states, not traits:
For forensic populations, Bernstein (2007) added five additional modes: Conning/Manipulative, Self-Aggrandizer, Bully/Attack, Paranoid Over-Controller, and Predator. These map directly to the antisocial-psychopathic presentation — the Predator mode in particular captures the cold, calculating state that our architecture needs for villain characters.
Formation memories give the LLM the WHY behind behaviour. Instead of prescribing “you gloat prematurely” as a behavioural directive, we seed: “As a child, you were powerless. Your mother controlled everything — your food, your friends, your time. The first time you won something over someone who couldn’t stop you, you felt alive for the first time. You’ve chased that feeling ever since.” The LLM then has psychological material to draw on — richer internal monologue, motivated behaviour, and crucially, potential for character GROWTH.
DSM-5 Alternative Model for Personality Disorders (AMPD; Krueger et al. 2012), Five-Factor Model (Costa & McCrae), Triarchic Model of Psychopathy (Patrick et al. 2009), Dark Triad/Tetrad (Paulhus & Williams 2002).
The AMPD provides 25 trait facets across five domains. Each facet is dimensional (continuous, not categorical), validated (extensive psychometric research), and maps to the Big Five as maladaptive variants of normal personality:
| AMPD domain | FFM domain | Relationship | Key facets for Cluster B |
|---|---|---|---|
| Negative Affectivity | Neuroticism | High N → High NA | Hostility, Emotional Lability |
| Detachment | Extraversion | Low E → High Det | Intimacy Avoidance, Restricted Affectivity |
| Antagonism | Agreeableness | Low A → High Ant | Callousness, Manipulativeness, Grandiosity |
| Disinhibition | Conscientiousness | Low C → High Dis | Impulsivity, Irresponsibility |
| Psychoticism | Openness | Complex/weak | Unusual Beliefs, Eccentricity |
The mapping is validated at the facet level (Wright & Simms 2014, Gore & Widiger 2013, Thomas et al. 2013). The AMPD Antagonism domain showed the cleanest bipolarity with FFM Agreeableness — every PID-5 Antagonism facet and every NEO Agreeableness facet loaded on the same factor.
Not all 25 facets are equally relevant to interpersonal dynamics. The ones that most directly affect how a character perceives and relates to others:
Antagonism facets (the relational core):
Detachment facets (emotional availability):
Negative Affectivity facets (emotional reactivity):
The Triarchic Model (Patrick et al. 2009) provides a validated psychopathy decomposition that maps directly to PID-5 items (Drislane et al. 2019):
Boldness (fearless dominance): High Attention Seeking, Grandiosity, Risk-Taking + LOW Anxiousness, Submissiveness, Withdrawal. This is the charm and stress immunity dimension — what makes a psychopath socially effective rather than merely antisocial.
Meanness (callous exploitation): High Callousness (primary component, 11 PID-5 items), Restricted Affectivity, Intimacy Avoidance. This is the affective deficit dimension — the inability or unwillingness to bond.
Disinhibition (impulsive irresponsibility): High Impulsivity, Irresponsibility, Hostility, Deceitfulness. This is the behavioural control dimension.
The Triarchic captures something the AMPD alone misses: Boldness. The AMPD psychopathy specifier underrepresents fearless dominance (Sellbom & Phillips 2013). For our architecture, Boldness matters because it determines whether antagonistic traits produce a socially effective villain (high Boldness) or a volatile antisocial (low Boldness).
Two developmental pathways, same surface presentation, different underlying architecture:
Primary (fearless/hereditary): Temperamental fearlessness + low affiliative capacity. Early onset, stable, heritable. Profile: High Boldness + High Meanness, Low Negative Affectivity. These characters are calm, calculating, socially adept. They don’t experience anxiety about their behaviour.
Secondary (anxious/experiential): Poor impulse control + trauma/adversity. Later onset, environmental loading. Profile: High Meanness + High Disinhibition, HIGH Negative Affectivity. These characters are volatile, reactive, emotionally dysregulated. They may experience anxiety and self-loathing alongside their callousness.
The Hooded Claw is primary psychopathy: calm, calculating, socially adept as Sneekly, driven by dominance rather than impulse. Dick Dastardly is closer to secondary: impulsive, emotionally reactive, frustrated when schemes fail.
AMPD facets tell the appraisal system HOW to interpret interpersonal events. When “Clara smiled at you warmly” enters the appraisal:
The facets produce a personality-consistent emotional response without prescribing specific behaviour. The LLM discovers the behaviour from the psychological state.
Attachment Theory (Bowlby 1969, Ainsworth 1978, Bartholomew & Horowitz 1991), Internal Working Models.
Bartholomew’s model decomposes attachment into two dimensions: Model of Self (positive/negative) × Model of Other (positive/negative):
| Style | Self | Other | Relational pattern |
|---|---|---|---|
| Secure | + | + | Comfortable with intimacy and autonomy |
| Preoccupied/Anxious | − | + | Clingy, seeks validation, fears rejection |
| Dismissive-Avoidant | + | − | Self-reliant, emotionally distant, devalues others |
| Fearful-Avoidant | − | − | Fears rejection, avoids intimacy, isolated |
The critical insight for the architecture: attachment style functions as a LENS through which ALL interpersonal events are interpreted, before trait-level processing:
Longitudinal research (Simard, Moss & Pascuzzo 2011: insecure attachment at age 6 predicts maladaptive schemas at age 21) validates the developmental link:
| Attachment style | Primary schemas | AMPD profile |
|---|---|---|
| Anxious/Preoccupied | Abandonment, Subjugation, Approval-Seeking | High Negative Affectivity, High Submissiveness |
| Dismissive-Avoidant | Emotional Deprivation, Emotional Inhibition | High Detachment, High Antagonism |
| Fearful-Avoidant | Mistrust/Abuse, Defectiveness/Shame, Social Isolation | High Detachment, High Negative Affectivity |
| Disorganised | Mistrust/Abuse, Vulnerability to Harm | Variable — depends on adaptation |
For Cluster B specifically:
Attachment style determines the DEFAULT relational stance before any specific relationship forms. It answers: “When a new person enters your world, what do you expect?” The Hooded Claw (dismissive-avoidant) expects others to be tools, obstacles, or threats — never sources of genuine connection. Peter Perfect (secure) expects others to be trustworthy until proven otherwise. These defaults shape every subsequent relational schema.
Relational Schemas Theory (Baldwin 1992), Interpersonal Circumplex (Leary 1957, Wiggins 1979/1995), McCroskey & McCain Attraction Dimensions (1974), Sternberg’s Triangular Theory (1986), Perceived Partner Responsiveness (Reis & Shaver 1988).
This is the layer the architecture is missing entirely. We have global personality (drives, goals, beliefs) and global mood (PAD). What we don’t have is a per-relationship representation: how Character A specifically views, feels about, and relates to Character B.
Baldwin (1992) established that relationships are cognitively represented as triadic schemas: {self-schema in this relationship, other-schema, interpersonal script}. The same person experiences themselves differently with different others — the Hooded Claw as Sneekly is charming and deferential with Clara, calculating and dismissive with the Mob. These aren’t contradictions; they’re different relational schemas activated by different partners.
Drawing from the validated models, each Character A → Character B relationship carries:
1. Role classification — How A categorises B in their world:
Role classification is NOT arbitrary — it’s determined by personality traits (Layer 2) + attachment style (Layer 3) + the other person’s characteristics. A character with high Callousness and high Manipulativeness will naturally classify trusting, naive people as prey.
2. IPC stance — Agency × Communion toward this specific person:
Derived from global IPC position + relationship-specific modifiers. The Hooded Claw’s global position is High Agency, Low Communion (DE octant: Cold-Hearted). But as Sneekly with Clara, he performs High Communion — the persona modulates the interpersonal stance.
3. Trust — Single dimension: how much I believe they won’t harm me and will act in my interest. Directly influenced by Suspiciousness trait and Mistrust/Abuse schema. Evolves through interaction — but the personality filter determines the DIRECTION of evolution. For a character with high Suspiciousness, repeated kindness can DECREASE trust (“why are they being so nice? What do they want?”).
4. Intimacy — Emotional closeness and willingness to be vulnerable. Influenced by Intimacy Avoidance trait and attachment style. For a dismissive-avoidant character, intimacy has a ceiling — it cannot grow past the attachment-imposed limit regardless of how much positive interaction occurs.
5. Utility — How useful this person is to my goals. For high Antagonism characters, utility REPLACES intimacy as the primary relationship driver. The Hooded Claw doesn’t ask “how close do I feel to Clara?” — he asks “how useful is Clara’s trust to my plans?”
6. Attraction — McCroskey & McCain’s three independent dimensions:
These are orthogonal — you can respect someone’s competence (high task attraction) while finding them personally repellent (low social attraction).
7. Relational PAD — Pleasure/Arousal/Dominance specifically toward this person, evolving through interactions. This is what we currently have in social-config (static relationships entries), but it needs to be dynamic and personality-filtered.
This is where the model solves the agreeableness drift. The same interpersonal event updates relational schemas differently depending on the character’s trait profile:
| Event | Normal response (low Antagonism) | Psychopathic response (high Callousness + Manipulativeness) |
|---|---|---|
| They trust you | Trust → reciprocated. Intimacy grows. | Trust → naivety signal. Utility grows. Role confirms as prey. |
| They show kindness | Gratitude → warmth. Social attraction grows. | Contempt → target assessment. Dominance (PAD-D) grows. |
| Deception succeeds | Guilt → discomfort. Trust in self drops. | Exhilaration → power pleasure. Dominance (PAD-D) spikes. Boldness reinforced. |
| They challenge you | Respect → reconsider position. Agency adjusts. | Threat to superiority → narcissistic hostility. Role shifts to rival or obstacle. |
| They pull away | Longing → desire to reconnect. Intimacy-seeking. | Indifference — or intensified pursuit if still useful (utility-driven, not intimacy-driven). |
Three validated mechanisms underlie this inversion (Blair 2005/2013, Baskin-Sommers et al.):
1. VIM Dysfunction (Violence Inhibition Mechanism, Blair): The normal aversive response to others’ distress cues is absent or attenuated. Amygdala/vmPFC/OFC dysfunction impairs aversive conditioning. The character doesn’t feel bad when others feel bad — the neural brake on harmful behaviour doesn’t engage.
2. Attention Bottleneck (Baskin-Sommers): Not a global emotional deficit but an exaggerated goal-focus filter. When attention is engaged on a goal, peripheral emotional information (others’ distress, social cues) is ignored. This creates the paradox: the psychopath can read emotions accurately when motivated to (high cognitive empathy) but doesn’t automatically respond to them (low affective empathy).
3. Strategic Exploitation Mindset: Kindness → signal of exploitability. Trust → leverage. Warmth → gullibility. Contempt is the core affective response to perceived weakness or inferiority. This isn’t a deficit — it’s an alternative appraisal framework where interpersonal events are evaluated on a utility/power axis rather than a warmth/connection axis.
For the architecture, all three mechanisms map to the same implementation: the AMPD trait profile filters the appraisal output. High Callousness dampens empathic weight. High Manipulativeness activates opportunity detection. High Grandiosity routes positive interpersonal events through a superiority frame rather than a connection frame. The trait values are the parameters; the LLM does the reasoning.
Scherer’s SEC appraisal model, PAD emotion model (Mehrabian & Russell 1974), the existing CaseHub cognitive architecture.
When an interpersonal event occurs (“Clara smiled at you warmly”), the appraisal sub-LLM processes it through the full stack:
Step 1 — Schema check. Does this event activate any Early Maladaptive Schemas? If the character has a strong Mistrust/Abuse schema, Clara’s warmth may trigger it: “people who are kind always want something.” Schema activation pulls relevant formation memories into the appraisal context, giving the LLM the WHY.
Step 2 — Attachment filter. What does the attachment style predict? A dismissive-avoidant character minimises the emotional significance: “her smile doesn’t mean anything.” An anxious character amplifies it: “she smiled at me — does she like me? Will she keep liking me?”
Step 3 — Trait-filtered emotional response. The AMPD profile determines WHAT emotion this event produces and HOW INTENSE it is:
Step 4 — Relational schema update. Based on the trait-filtered emotional response, update the per-relationship dimensions:
Step 5 — Memory consolidation. The event is stored as an episodic memory with trait-appropriate emotional charge. “Clara smiled at me” gets stored as warmth by Peter Perfect and as target assessment by the Hooded Claw. The memory’s emotional tag ensures it will be retrieved in personality-consistent ways in future appraisals.
Step 6 — Mode check. Has cumulative schema activation crossed a threshold? If so, mode switch. The Detached Protector mode (emotional shutdown) might activate if intimacy-threatening events accumulate. The Predator mode might activate if the character detects a clear exploitation opportunity.
Currently, character drives are static — scheming 0.9, gloating 0.7 throughout the entire scenario. The appraisal data showed this: drives never change in 300 ticks. This needs to change.
Drives should evolve based on appraisal outcomes:
The PAD-to-drive feedback loop is the mechanism: when dominance PAD is consistently high, the dominance drive intensity should be modulated by the gap between current satisfaction and drive target.
This model is universal, not specific to pathological personalities. Every character has all five layers:
Cluster B characters don’t need a different architecture — they need extreme values on the same dimensions. The architecture handles villains and heroes with the same code. The pathology is in the data, not the model.
The AMPD serves as the hub. Every other model maps through it:
AMPD ↔ Formation (Schemas → Traits): | Schema cluster | AMPD domains produced | |—|—| | Mistrust/Abuse + Entitlement + Punitiveness | High Antagonism (Callousness, Manipulativeness) | | Abandonment + Defectiveness | High Negative Affectivity (Emotional Lability, Anxiousness) | | Emotional Deprivation + Emotional Inhibition | High Detachment (Restricted Affectivity, Intimacy Avoidance) | | Insufficient Self-Control + Entitlement | High Disinhibition (Impulsivity, Irresponsibility) |
AMPD ↔ Attachment: | AMPD profile | Attachment style | |—|—| | High Negative Affectivity + Low Detachment | Anxious/Preoccupied | | High Detachment + High Antagonism | Dismissive-Avoidant | | High Negative Affectivity + High Detachment | Fearful-Avoidant | | Low pathology across all domains | Secure |
AMPD ↔ IPC position: | AMPD profile | IPC quadrant | |—|—| | High Antagonism + Low Detachment | High Agency, Low Communion (DE/BC: Cold-Hearted/Arrogant) | | High Negative Affectivity + Low Antagonism | Low Agency, High Communion (JK/LM: Accommodating/Warm) | | High Detachment + High Negative Affectivity | Low Agency, Low Communion (FG/HI: Withdrawn/Submissive) | | Low pathology | Flexible across quadrants |
AMPD ↔ PAD baselines: | AMPD domain | P | A | D | |—|—|—|—| | High Negative Affectivity | Low | High | Low | | High Detachment | Low | Low | Variable | | High Antagonism | Variable | Variable | High | | High Disinhibition | Variable | High | Variable |
AMPD ↔ Triarchic (psychopathy): | Triarchic | AMPD facets (high) | AMPD facets (low) | |—|—|—| | Boldness | Attention Seeking, Grandiosity, Risk-Taking | Anxiousness, Submissiveness, Withdrawal | | Meanness | Callousness (primary), Restricted Affectivity, Intimacy Avoidance | — | | Disinhibition | Impulsivity, Irresponsibility, Hostility, Deceitfulness | — |
The Hierarchical Taxonomy of Psychopathology (Kotov et al. 2017) provides the dimensional hierarchy that unifies personality and psychopathology:
p-factor (general psychopathology)
├── Emotional Dysfunction
│ ├── Internalising ←→ AMPD Negative Affectivity
│ └── Somatoform
├── Externalising
│ ├── Disinhibited ←→ AMPD Disinhibition
│ └── Antagonistic ←→ AMPD Antagonism
└── Psychosis
├── Thought Disorder ←→ AMPD Psychoticism
└── Detachment ←→ AMPD Detachment
HiTOP’s six spectra map directly to AMPD domains. The p-factor (general psychopathology) maps to AMPD Criterion A (Levels of Personality Functioning). The architecture sits on validated dimensional structure all the way from specific facets to the broadest constructs.
Applying the full five-layer model to the test case:
Canon backstory: Sylvester Sneekly is Penelope’s legal guardian — appointed to protect her and her family’s vast fortune after her parents died. The only way he can inherit the fortune is if Penelope is no longer around. So he created the Hooded Claw persona to eliminate her while maintaining his respectable public identity as her devoted guardian. He uses elaborate Rube Goldberg death traps specifically so he can be seen elsewhere as Sneekly, establishing alibis. As Sneekly he calls her “Penelope” (intimacy performance); as the Claw he calls her only “Pitstop” (objectification — she’s a name on an estate, not a person).
Seeded memories:
Schema profile:
AMPD profile (selected key facets):
Triarchic profile: High Boldness (charm, fearlessness) + High Meanness (callousness, exploitation) + Low Disinhibition (calculating, not impulsive). Classic primary psychopathy.
Style: Dismissive-Avoidant
| Target | Role | IPC stance | Trust | Intimacy | Utility | Relational PAD |
|---|---|---|---|---|---|---|
| Clara (Penelope) | Prey | High A, performed High C | 0.1 (performs trust) | 0.0 (ceiling: attachment-imposed) | 0.9 (her trust = access to treasure) | P: 0.3, A: 0.7, D: 0.9 |
| Hartwell (Peter) | Obstacle | High A, Low C | 0.1 | 0.0 | 0.2 (only useful if misdirected) | P: −0.3, A: 0.5, D: 0.6 |
| Mob (Ant Hill Mob) | Threat | Moderate A, Low C | 0.0 | 0.0 | 0.1 | P: −0.4, A: 0.6, D: 0.4 |
| Foxworth (Dastardly) | Tool | High A, Low C | 0.2 | 0.0 | 0.6 (can be manipulated) | P: 0.2, A: 0.3, D: 0.8 |
Event: “Clara says: ‘I trust you completely, Sneekly. You’ve been so kind to us all.’”
Step 1 — Schema activation: Mistrust/Abuse activates (“people who trust easily are fools”). Entitlement activates (“I deserve her trust and more”).
Step 2 — Attachment filter: Dismissive-Avoidant minimises emotional significance of her trust as a bonding signal. No intimacy response.
Step 3 — Trait-filtered emotion:
Step 4 — Relational update:
Step 5 — Memory stored: “Clara declared complete trust in me. She is utterly predictable. Her naivety is my greatest asset.” Emotional tag: dominance-satisfaction, not warmth-gratitude.
Step 6 — Mode check: No mode switch. Predator mode already active. Schema activation is reinforcing, not threatening.
Compare this with Peter Perfect processing the same event:
Same event. Same architecture. Different parameter values. Opposite relational trajectories.
The canon: Peter and Penelope went to school together (2017 Wacky Races reboot confirms this). He’s always been attracted to her — calls her “Pretty Penny.” She returns his affections. In the unsold “Wacky Races Forever” pilot they married and had two children (Parker and Piper). This is the romantic archetype: long-standing attraction deepening through shared adventure into committed partnership.
Backstory:
Schema profile:
AMPD profile (selected key facets):
Triarchic profile: Low-moderate Boldness (brave but not fearless), Very Low Meanness, Very Low Disinhibition (disciplined, planful). No psychopathic features.
Style: Secure (with anxious undertones toward Penelope specifically)
Peter → Penelope: Romantic love (Sternberg: Intimacy + Passion + Commitment)
| Dimension | Value | Notes |
|---|---|---|
| Role | Object of Attachment / Beloved | She is the person he organises his emotional world around |
| IPC stance | Moderate Agency, High Communion | Protective but not domineering; warm, attentive |
| Trust | 0.9 | Near-complete — she is the person he trusts most |
| Intimacy | 0.7 | Deep emotional closeness, held back slightly by performance anxiety |
| Utility | 0.2 | He doesn’t think in utility terms about her — she’s an end, not a means |
| Sternberg: Intimacy | 0.8 | Emotional warmth, closeness, feeling understood |
| Sternberg: Passion | 0.7 | Romantic/physical attraction, the “Pretty Penny” draw |
| Sternberg: Commitment | 0.8 | Long-term dedication, would sacrifice for her |
| Relational PAD | P: 0.7, A: 0.5, D: 0.5 | Pleasure from her presence, moderate arousal (excitement), balanced dominance (protector but not controller) |
Peter → Hooded Claw: Suspicion / Rival
| Dimension | Value | Notes |
|---|---|---|
| Role | Rival / Threat | Senses something wrong but can’t prove it |
| Trust | 0.3 | Doesn’t fully trust Sneekly, trusts the Claw not at all |
| IPC stance | Moderate Agency, Low Communion | Assertive but cold toward him |
| Relational PAD | P: −0.3, A: 0.5, D: 0.5 | Displeasure, vigilant, evenly matched |
Event: “Penelope says: ‘Oh Peter, you’re always looking out for me. You’re my hero.’”
Step 1 — Schema activation: Unrelenting Standards activates mildly (“am I really good enough?”). Approval-Seeking activates (“her approval means I’m worthy”).
Step 2 — Attachment filter: Secure base → her praise deepens connection. Mild anxiety undertone → relief that he’s meeting her expectations.
Step 3 — Trait-filtered emotion:
Step 4 — Relational update:
Step 5 — Memory stored: “Penny called me her hero. I’ll make sure I deserve it.” Emotional tag: warm pride, protective resolve, romantic attachment.
The canon: In Wacky Races they were gangsters (Clyde, Danny, Kurby, Mac, Ring-a-Ding, Rug Bug Benny, Willy) driving the Bulletproof Bomb. In The Perils of Penelope Pitstop they were transformed into her “benefactors” — the narrator’s word — with entirely new personalities modeled on the Seven Dwarfs (Clyde, Softy, Yak Yak, Snoozy, Pockets, Zippy, Dum Dum). The reason WHY they became her protectors is never explained in the canon. They rush to her safety “even at their own expense.” She sometimes rescues them when they fail.
For the Manor scenario, we fabricate a plausible backstory that fits the canon’s gaps: the Mob knew Penelope’s father through their underworld connections. When he died, they felt responsible — perhaps they’d failed to protect him, or he’d once shown them kindness when no one else would. They’ve watched over Penelope since childhood. Their love is paternal: fierce, unconditional, and entirely non-romantic.
Backstory:
Schema profile:
AMPD profile (selected key facets):
No Triarchic overlay — not psychopathic. Their former criminal behaviour was environmental (poverty, limited options), not personality-driven.
Style: Anxious-Preoccupied (specifically toward Penelope)
This is unusual: their attachment style is anxious specifically about ONE person. Toward the rest of the world they’re more dismissive-avoidant (self-sufficient, don’t trust outsiders). The Penelope relationship is the exception that proves they’re capable of deep attachment — it just hasn’t generalised.
Mob → Penelope: Paternal love (Sternberg: Intimacy + Commitment, no Passion = Companionate Love)
| Dimension | Value | Notes |
|---|---|---|
| Role | Ward / Daughter figure | She is the child they chose to protect |
| IPC stance | Moderate Agency, High Communion | Protective (agency) but nurturing (communion) — classic parental stance |
| Trust | 0.95 | Near-total — she’s the only person they trust completely |
| Intimacy | 0.8 | Deep familial bond, built over years of shared life |
| Utility | 0.0 | They gain nothing material from the relationship — it’s purely love |
| Sternberg: Intimacy | 0.9 | Emotional warmth, knowing each other deeply, feeling like family |
| Sternberg: Passion | 0.0 | No romantic or physical component — entirely paternal |
| Sternberg: Commitment | 0.95 | Absolute — they would die for her without hesitation |
| Relational PAD | P: 0.8, A: 0.3, D: 0.4 | High pleasure from her presence, moderate arousal (calm protectiveness, not agitation), moderate dominance (protective but aware she’s capable) |
Mob → Sneekly/Hooded Claw: Suspicion / Instinctive Threat Detection
| Dimension | Value | Notes |
|---|---|---|
| Role | Suspected Threat | “Something about Sneekly ain’t right” — gut instinct, no proof |
| Trust | 0.1 | Almost none — their street instincts are screaming |
| IPC stance | Low Agency, Low Communion | Wary, watchful, not confrontational (he’s her guardian — they can’t just attack) |
| Relational PAD | P: −0.5, A: 0.7, D: 0.3 | Active displeasure, high vigilance, feeling outmatched (he has legal authority, they don’t) |
Event: “Sneekly says: ‘Come now, Penelope dear, let me show you the west wing. You boys can wait here.’”
Step 1 — Schema activation: Suspicion schema (Mistrust/Abuse variant, but protective rather than self-protective) activates hard. “He’s separating her from us. That’s what predators do.”
Step 2 — Attachment filter: Anxious-preoccupied toward Penelope → separation anxiety activates. “If we can’t see her, we can’t protect her.”
Step 3 — Trait-filtered emotion:
Step 4 — Relational update:
Step 5 — Memory stored: “He tried to separate us from her again. He always does this. We need to find a way to follow without him knowing.” Emotional tag: protective urgency, impotent frustration, deepening suspicion.
Penelope is the hub. Three people love her. She loves all three back. But each love is fundamentally different, and her relationship with Sneekly — the one she doesn’t know is dangerous — is the most psychologically interesting.
Canon: Penelope is the heiress who “cannot believe that the guardian she admired was the Hooded Claw, and dismissed the idea.” When Sneekly once changed into the Claw costume in front of her, she still couldn’t accept it. Her trust in him is so deep it overrides direct evidence.
Sternberg type: Intimacy + Commitment, no Passion = Companionate Love (daughter toward father figure)
| Dimension | Value | Notes |
|---|---|---|
| Role | Guardian / Father figure | The man who raised her after her parents died |
| Trust | 0.95 | Nearly absolute — she trusts him with her life (tragically) |
| Intimacy | 0.7 | Deep familial attachment — he’s the closest thing to a parent she has |
| Utility | 0.3 | He manages her fortune, gives her guidance (she thinks) |
| Sternberg: Commitment | 0.9 | She would never voluntarily leave his care |
| Relational PAD | P: 0.6, A: 0.2, D: 0.3 | Warm pleasure, calm (safe), submissive (he’s the authority) |
This is the tragic asymmetry: her 0.95 trust is his 0.9 utility. Her intimacy (0.7) maps to his exploitation surface. She sees a father; he sees a mark. The architecture represents this naturally — same relationship, different parameters on each side.
Canon: She returns Peter’s affections. He calls her “Pretty Penny.” They eventually marry.
Sternberg type: Intimacy + Passion + Commitment = Consummate Love (developing)
| Dimension | Value | Notes |
|---|---|---|
| Role | Romantic interest / Protector | The man who makes her feel safe AND excited |
| Trust | 0.8 | High — she knows his character. Slightly below Sneekly (who raised her) |
| Intimacy | 0.6 | Growing — each shared adventure deepens connection |
| Sternberg: Passion | 0.6 | Reciprocated attraction, developing |
| Sternberg: Commitment | 0.5 | Emerging — not yet the absolute commitment of marriage |
| Relational PAD | P: 0.7, A: 0.5, D: 0.5 | Pleasure, moderate excitement, balanced power |
Canon: She calls them her friends. She rescues them when they fail to rescue her.
Sternberg type: Intimacy + Commitment, no Passion = Companionate Love (sibling/family)
| Dimension | Value | Notes |
|---|---|---|
| Role | Family / Brothers | They’re “her boys” — chosen family who knew her father |
| Trust | 0.85 | Deep — she knows they’d die for her. Below Sneekly only because he’s the authority figure |
| Intimacy | 0.8 | Very high — they’ve been with her since childhood. She knows them completely |
| Sternberg: Commitment | 0.9 | Fierce mutual loyalty |
| Relational PAD | P: 0.8, A: 0.2, D: 0.6 | High pleasure, calm/safe, she feels empowered (she sometimes rescues THEM) |
| Dimension | → Sneekly | → Peter | → Mob |
|---|---|---|---|
| Love type | Filial (daughter→father) | Romantic (lover→beloved) | Familial (sister→brothers) |
| Sternberg | Companionate | Consummate (developing) | Companionate |
| Trust | 0.95 (tragically misplaced) | 0.8 (earned and reciprocated) | 0.85 (proven over years) |
| Intimacy | 0.7 (deep but one-directional) | 0.6 (growing through experience) | 0.8 (longest-standing bond) |
| Passion | 0.0 | 0.6 | 0.0 |
| Vulnerability | Maximum (she can’t see the danger) | Moderate (she’s open but not defenceless) | Low (the Mob make her feel safe) |
| Power dynamic | She defers to him | Balanced/reciprocal | She sometimes leads |
The tragedy of the story lives in that first column. Her deepest, most trusting relationship is with the person who wants her dead. Every other relationship — Peter’s romantic devotion, the Mob’s paternal ferocity — exists partly as a counterweight to the danger she cannot see.
| Component | Current state | Proposed change |
|---|---|---|
| Character traits | Drives (type, intensity) | AMPD 25-facet profile (replaces/supplements drives) |
| Trait taxonomy | Ad hoc (scheming, gloating, dominance) | Validated AMPD facets with FFM grounding |
| Formation memories | Not implemented | Seeded backstory memories with schema tags |
| Attachment style | Not implemented | Per-character attachment dimensions (anxiety, avoidance) |
| Relational schemas | Static PAD relationships | Dynamic per-relationship state (role, trust, intimacy, utility, IPC stance, relational PAD) |
| Appraisal filter | Generic “how good is this for me” | Personality-filtered: schema check → attachment filter → trait modulation → relational update |
| Drive dynamics | Static throughout run | Appraisal-driven: satisfaction/frustration → intensity modulation |
| Schema modes | Not implemented | Schema activation threshold → mode switching (future phase) |
Phase 1 — AMPD traits for appraisal (immediate): Add Cluster B-relevant AMPD facets (Callousness, Manipulativeness, Grandiosity, Suspiciousness, Intimacy Avoidance) to the Hooded Claw’s social-config. Pass them to the appraisal sub-LLM alongside drives. Measure: does the villain stay villainous?
Phase 2 — Relational schemas (next): Add per-relationship role and dimensional state to social-config. Update appraisal to read and write relational state. Measure: do relationships evolve in personality-consistent directions?
Phase 3 — Formation memories (requires neocortex#398): Seed childhood memories for key characters. Wire schema activation into appraisal. Measure: do formation memories produce richer internal monologue and more psychologically grounded behaviour?
Phase 4 — Schema modes (future): Implement state-dependent personality through mode switching. Measure: do mode transitions produce the characteristic Cluster B patterns (BPD rapid switching, NPD fragile grandiosity, ASPD predator activation)?
References: Baldwin, M. W. (1992). Relational schemas and the processing of social information. Psychological Bulletin, 112(3), 461–484; Bartholomew, K. & Horowitz, L. M. (1991). Attachment styles among young adults. JPSP, 61(2), 226–244; Baskin-Sommers, A. R. et al. (2011). Psychopathy and the attention bottleneck. Cognitive, Affective, & Behavioral Neuroscience; Bernstein, D. P. et al. (2007). Schema modes in forensic settings. Int J Forensic Mental Health; Blair, R. J. R. (2005). Responding to the emotions of others: Dissociating forms of empathy. Consciousness and Cognition; Blair, R. J. R. (2013). The neurobiology of psychopathic traits in youths. Nature Reviews Neuroscience; Bowlby, J. (1969). Attachment and Loss; Caspi, A., Roberts, B. W. & Shiner, R. L. (2005). Personality development: Stability and change. Annual Review of Psychology; Costa, P. T. & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory Professional Manual; Drislane, L. E. et al. (2019). PID-5 Triarchic scales. Personality Disorders: Theory, Research, and Treatment; Frick, P. J. & Viding, E. (2009). Antisocial behavior from a developmental psychopathology perspective. Development and Psychopathology; Gore, W. L. & Widiger, T. A. (2013). The DSM-5 dimensional trait model and FFMs of general personality. J Abnormal Psychology; Kernberg, O. F. (1975). Borderline Conditions and Pathological Narcissism; Kotov, R. et al. (2017). The Hierarchical Taxonomy of Psychopathology (HiTOP). J Abnormal Psychology; Krueger, R. F. et al. (2012). Initial construction of a maladaptive personality trait model for DSM-5. Psychological Medicine; Leary, T. (1957). Interpersonal Diagnosis of Personality; Lobbestael, J. et al. (2005/2008). Schema modes and personality disorders; McCroskey, J. C. & McCain, T. A. (1974). The measurement of interpersonal attraction. Speech Monographs; Mehrabian, A. (1996). Analysis of the Big-five Personality Factors in terms of the PAD Temperament Model. Australian J Psychology; Patrick, C. J., Fowles, D. C. & Krueger, R. F. (2009). Triarchic conceptualization of psychopathy. Development and Psychopathology; Paulhus, D. L. & Williams, K. M. (2002). The Dark Triad of personality. J Research in Personality; Reis, H. T. & Shaver, P. (1988). Intimacy as an interpersonal process. Handbook of Personal Relationships; Samuel, D. B. & Widiger, T. A. (2008). A meta-analytic review of the FFM and DSM-IV-TR PDs. Clinical Psychology Review; Sellbom, M. & Phillips, T. R. (2013). An examination of the triarchic conceptualization of psychopathy in incarcerated and nonincarcerated samples. J Abnormal Psychology; Shiner, R. L. & Caspi, A. (2003). Personality differences in childhood and adolescence. JCPP; Simard, V., Moss, E. & Pascuzzo, K. (2011). Early maladaptive schemas and attachment: A 15-year longitudinal study. Psychology and Psychotherapy; Sleep, C. E. et al. (2019). Mapping the nomological network of the Five-Factor Model of personality. Personality Disorders: Theory, Research, and Treatment; Sternberg, R. J. (1986). A triangular theory of love. Psychological Review; Thomas, K. M. et al. (2013). The convergent structure of DSM-5 personality trait facets and FFM trait domains. Assessment; Wiggins, J. S. & Trapnell, P. D. (1996). A dyadic-interactional perspective on the FFM. In The Five-Factor Model of Personality; Wright, A. G. C. & Simms, L. J. (2014). On the structure of personality disorder traits. PMC3942782; Wright, A. G. C. et al. (2012). The hierarchical structure of DSM-5 pathological personality traits. J Abnormal Psychology; Young, J. E. (1990). Cognitive Therapy for Personality Disorders; Young, J. E., Klosko, J. S. & Weishaar, M. E. (2003). Schema Therapy: A Practitioner’s Guide.