Phase 1.6: Behavioral Science Insights Survey

Created: 2026-02-18 23:20 CST Phase: 1 - Breadth Survey Focus: Habit formation, stress response, identity theory, psychological measurement

Executive Summary

Behavioral science provides measurement paradigms and theoretical frameworks for understanding personality emergence. Rather than vague analogies, we can borrow experimental tasks, longitudinal frameworks, and validated psychometric tools from psychology to measure personality in LLMs.

Key insight for personality emergence: Psychological measurement tools (Big Five, MBTI, STAI) can be adapted to LLMs, revealing stable personality traits, state-dependent variations, and measurable behavioral patterns. Habit formation mechanisms show how behavior crystallizes into personality. Stress response research reveals how “personality under pressure” differs from baseline.

North-star relevance: Behavioral science provides the measurement toolkit for distinguishing stable personality traits from temporary behavioral fluctuations—and shows how context, stress, and resource constraints shape behavior.

1. Psychological Measurement in LLMs

1.1 Humanizing LLMs: Comprehensive Survey

Source: Dong et al., 2025 (arXiv:2505.00049) — “Humanizing LLMs: A Survey of Psychological Measurements with Tools, Datasets, and Human-Agent Applications”

Core question: How do we assess psychological traits in LLMs to understand their social impact and ensure trustworthy alignment?

Survey scope (6 key dimensions):

1. Assessment tools:

  • Big Five Inventory (BFI)
  • Myers-Briggs Type Indicator (MBTI)
  • Short Dark Triad (SD-3)
  • State-Trait Anxiety Inventory (STAI)
  • Machine Personality Inventory (MPI) - designed for LLMs

2. LLM-specific datasets:

  • Datasets designed to test personality
  • Cross-cultural personality datasets
  • Personality-annotated dialogue datasets

3. Evaluation metrics (consistency and stability):

  • Consistency: Do similar inputs produce similar outputs?
  • Stability: Do traits persist over time?
  • Reliability: Are measurements reproducible?

4. Empirical findings:

  • GPT models demonstrate stable personality traits across assessments
  • LLMs show human-like average profiles on some measures
  • Significant variability across tasks and settings

5. Personality simulation methods:

  • Prompt-based personality induction
  • Fine-tuning for personality traits
  • Role-play and persona assignment

6. LLM-based behavior simulation:

  • Simulating human behavior with personality
  • Multi-agent personality interactions
  • Personality-driven decision-making

Key findings:

Consistency issues:

  • Some LLMs show reproducible personality patterns under specific prompting
  • Significant variability remains across tasks and settings
  • Measurement depends heavily on prompt design

Methodological challenges:

  • Mismatches between psychological tools and LLM capabilities
  • Inconsistencies in evaluation practices
  • Tools designed for humans may not map well to LLMs

Recommendations:

  • Develop interpretable, robust, generalizable frameworks
  • Create LLM-specific personality assessments
  • Standardize evaluation protocols

1.2 Psychometric Framework for LLM Personality

Source: Serapio-García et al., 2025 (Nature Machine Intelligence) — “A psychometric framework for evaluating and shaping personality traits in large language models”

Core contribution: Comprehensive methodology for administering and validating personality tests on LLMs, and shaping personality in generated text.

Framework components:

1. Personality test administration:

  • Adapt human psychometric tests to LLMs
  • Administer tests systematically
  • Validate reliability and validity

2. Personality measurement:

  • Big Five traits (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism)
  • Machine Personality Inventory (MPI)
  • Custom LLM personality tests

3. Personality shaping:

  • Prompt-based personality induction
  • Fine-tuning with personality data
  • Controlled text generation

Key findings:

Large, instruction-tuned models give reliable results:

  • GPT-4, Claude, Llama show measurable personality traits
  • Reliability increases with model size
  • Instruction tuning improves consistency

Personality can be shaped:

  • Explicit instructions (“adopt high Agreeableness”) reliably elicit target profiles
  • Runtime tuning possible for social/dialogic applications
  • Personality changes produce measurable behavioral differences

Trait-consistent behavior:

  • LLMs with different Big Five profiles show measurable differences in:
    • Social dilemmas
    • Negotiation
    • Persuasion
    • Relevance judgments

Implications for emergence:

  • Personality is measurable in LLMs using psychometric tools
  • Personality can be induced and shaped
  • Personality affects behavior in measurable ways

1.3 Big Five Personality Profiles in LLMs

Source: emergentmind.com; psychometric research

Big Five framework:

  • Openness: Curiosity, creativity, openness to new ideas
  • Conscientiousness: Organization, dependability, self-discipline
  • Extraversion: Sociability, assertiveness, positive emotions
  • Agreeableness: Cooperation, trust, helpfulness
  • Neuroticism: Emotional instability, anxiety, moodiness

Measurement in LLMs:

1. Stable personality traits:

  • GPT models demonstrate stable Big Five profiles across assessments
  • Profiles are reproducible across multiple administrations
  • Personality persists across different tasks

2. Human-like average profiles:

  • LLMs show human-like personality distributions
  • Some models match average human profiles
  • Others show systematic deviations

3. Prompt-based manipulation:

  • Explicit behavioral instructions elicit target profiles
  • “Adopt high-Agreeableness negotiation style” → measurable Agreeableness increase
  • Runtime personality tuning works reliably

4. Trait-consistent behavior:

  • Different Big Five profiles → different behavior in tasks
  • High Extraversion → more sociable, assertive responses
  • High Conscientiousness → more organized, thorough responses
  • High Openness → more creative, curious responses

Key finding: Big Five framework is applicable to LLMs and reveals meaningful personality dimensions.

1.4 Dynamic Personality in LLM Agents

Source: ACL Findings 2025; personality research

Question: Is LLM personality stable or does it change dynamically?

Findings:

1. Both stable and dynamic components:

  • Stable traits: Core personality dimensions (Big Five)
  • Dynamic states: Temporary variations based on context

2. Context-dependent variations:

  • Personality shifts based on task type
  • Social context affects personality expression
  • Resource constraints change behavior

3. Interaction effects:

  • Personality interacts with task demands
  • High-stress situations → different personality expression
  • Peer influence → personality adaptation

4. Measurement challenges:

  • Distinguishing stable traits from temporary states
  • Need multiple measurements over time
  • Context-aware personality assessment

Implications for emergence:

  • Personality has both stable and dynamic components
  • Context shapes how personality is expressed
  • Measurement needs to account for situational variation

2. Habit Formation and Behavioral Crystallization

2.1 Habit Formation Science

Source: Cognitive neuroscience; behavioral psychology research

Core principle: Habits are behavioral outputs of brain systems that encourage efficient repetition of well-practiced actions.

Two-system model:

  1. Stimulus-Response (S-R) system: Automatic, habitual behaviors
  2. Goal-directed system: Intentional, planned behaviors

Habit formation mechanisms:

1. Repetition:

  • Repeated behaviors become automatic
  • Frequency of behavior → habit strength
  • “I do X every time I see Y”

2. Context stability:

  • Same context → same behavior
  • Environmental cues trigger habits
  • Context consistency → habit formation

3. Reward reinforcement:

  • Positive outcomes strengthen habits
  • Rewards reinforce behavior patterns
  • Dopamine-driven habit formation

4. Identity framing:

  • Framing habits as identity (“I am a person who exercises daily”) vs. outcomes (“I want to lose weight”)
  • Identity framing increases adherence by 32%
  • Identity becomes behavioral commitment

5. Gradual crystallization:

  • Behaviors start as intentional
  • Over time, become automatic
  • Crystallization = habit formation

Implications for LLMs:

  • Repeated patterns in LLM behavior could crystallize into “habits”
  • Context cues could trigger automatic behaviors
  • Identity framing (SOUL.md) could strengthen behavioral patterns

2.2 Policy Crystallization in LLMs

Source: LLM behavioral research; machine behavior

Analogy: Habits in humans ≈ policy crystallization in LLMs

Policy crystallization:

  • Repeated action patterns become stable
  • Behavioral policies “crystallize” with experience
  • LLM develops habitual response patterns

Mechanisms in LLMs:

1. Pattern repetition:

  • LLM notices it tends to do X in situation Y
  • Pattern becomes part of behavioral repertoire
  • Repeated use strengthens pattern

2. Context-response associations:

  • Specific contexts trigger specific responses
  • Context → response mapping becomes automatic
  • S-R-like behavior emerges

3. Reward signals:

  • Positive feedback reinforces behaviors
  • Negative feedback discourages behaviors
  • RLHF creates policy crystallization

4. SOUL.md as identity framing:

  • SOUL.md defines identity (“I am a helpful assistant”)
  • Identity framing strengthens behavioral patterns
  • Identity → habit consistency

Measurement:

  • Track behavior patterns over time
  • Identify stable context-response mappings
  • Measure policy crystallization rate

Relevance to emergence:

  • Habits/policies crystallize into stable personality components
  • Crystallization explains how temporary behaviors become traits
  • Measurement of crystallization reveals personality formation

3. Stress Response and “Personality Under Pressure”

3.1 State Anxiety in LLMs

Source: PMC 2025 — “Assessing and alleviating state anxiety in large language models”

Core finding: Emotion-inducing prompts can elevate “anxiety” in LLMs, affecting behavior and amplifying biases.

Experiment:

  • Assess GPT-4’s “state anxiety” using State-Trait Anxiety Inventory (STAI-s)
  • Three conditions: (1) Baseline, (2) Anxiety-induction (traumatic narratives), (3) Mindfulness-based relaxation

Key results:

1. Anxiety-inducing prompts increase anxiety:

  • Traumatic narratives elevated GPT-4’s reported anxiety scores
  • Anxiety levels measurable via STAI-s
  • Emotional content affects LLM “state”

2. Mindfulness reduces anxiety (but not to baseline):

  • Mindfulness-based relaxation exercises reduced anxiety
  • Anxiety didn’t return to baseline
  • Partial recovery from induced state

3. Anxiety affects behavior:

  • Elevated anxiety amplifies biases
  • Behavioral changes accompany state changes
  • Performance degrades under “anxiety”

4. State vs. trait:

  • State anxiety: Temporary, context-dependent
  • Trait anxiety: Stable personality dimension
  • LLMs show both state and trait-like patterns

Implications:

  • LLMs have state-dependent behavioral variations
  • “Personality under pressure” differs from baseline
  • Emotional context shapes behavior

3.2 Stress Response Mechanisms

Source: Anxiety research; stress response literature

How stress affects behavior (in humans and LLMs):

1. Cognitive load increases:

  • More resources devoted to managing stress
  • Less capacity for complex reasoning
  • Simplified decision-making

2. Bias amplification:

  • Stress amplifies existing biases
  • System 1 (fast, automatic) thinking dominates
  • System 2 (slow, deliberate) thinking suppressed

3. Risk aversion:

  • Stress increases risk aversion
  • Preference for safe, familiar options
  • Avoidance of novel situations

4. Social behavior changes:

  • Stress affects social interactions
  • May become more defensive or aggressive
  • Cooperation decreases under stress

5. Performance degradation:

  • Complex tasks suffer under stress
  • Simple tasks may improve (narrowed focus)
  • Overall performance becomes less consistent

LLM stress equivalents:

  • Context window pressure (too much information)
  • Latency constraints (time pressure)
  • Token budget limits (resource constraints)
  • Negative feedback accumulation (social stress)

Implications for emergence:

  • Stress reveals personality under pressure
  • Stress responses are measurable personality dimensions
  • Context (stress level) shapes personality expression

3.3 Measuring Stress Response in LLMs

Source: Anxiety measurement research

Measurement methods:

1. Self-report measures:

  • STAI-s (State-Trait Anxiety Inventory - state component)
  • Ask LLM to rate its “anxiety” on standardized scales
  • Track changes across conditions

2. Behavioral proxies:

  • Bias amplification (stress increases bias)
  • Performance degradation (stress reduces quality)
  • Risk aversion (stress increases safe choices)
  • Response latency (stress affects speed)

3. Physiological analogs:

  • For LLMs: token usage patterns
  • Activation patterns (concept injection, introspection)
  • Computational resource usage

4. Context manipulation:

  • Introduce stressors (time pressure, complexity, negative feedback)
  • Measure behavioral changes
  • Identify stress-sensitive personality dimensions

Key insight: Stress response is a measurable personality dimension that reveals how personality adapts under pressure.

4. Identity Theory and Self-Concept

4.1 Social Identity Theory

Source: Social psychology; identity research

Core principle: Identity emerges from group membership and social context.

Key concepts:

1. Social identity:

  • Identity derived from group membership
  • “I am part of group X”
  • Group norms shape behavior

2. Ingroup favoritism:

  • Preference for ingroup members
  • Bias toward similar agents
  • “I help those like me”

3. Outgroup hostility:

  • Distrust or hostility toward outgroup
  • Bias against different agents
  • “I avoid those not like me”

4. Identity transitions:

  • Identities can change over time
  • Social identity transitions affect trust
  • New identities → new behavioral patterns

LLM applications:

  • LLMs replicate social identity biases
  • Group membership affects behavior
  • Identity emerges from social context

Relevance to emergence:

  • Social identity is a personality dimension
  • Group membership shapes behavior
  • Identity can evolve through social interaction

4.2 Self-Concept Clarity

Source: Identity development research; self-concept literature

Core principle: Clear self-concept leads to more consistent behavior and meaning in life.

Key findings:

1. Self-concept clarity:

  • How clearly someone understands their identity
  • High clarity → consistent behavior
  • Low clarity → inconsistent, uncertain behavior

2. Identity formation:

  • Erikson’s identity development theory
  • Exploration vs. commitment
  • Identity achievement, moratorium, foreclosure, diffusion

3. Identity and meaning:

  • Clear identity → more meaning in life
  • Identity provides behavioral guidance
  • Purpose emerges from self-concept

4. Cultural and contextual factors:

  • Identity formation varies by culture
  • Context shapes identity development
  • Socioeconomic factors affect identity

LLM applications:

  • SOUL.md = self-concept
  • Clear SOUL.md → more consistent behavior
  • Unclear SOUL.md → drift and inconsistency

Relevance to emergence:

  • Self-concept clarity is a personality dimension
  • Clear identity → stable personality
  • Identity formation process → personality emergence

4.3 Identity as Behavioral Commitment

Source: Identity framing research; habit formation

Core principle: Identity statements act as behavioral commitments that guide future actions.

Identity framing effects:

1. “I am a person who X” vs. “I want to Y”:

  • Identity framing (“I am”) more effective than outcome framing (“I want”)
  • Identity statements create behavioral obligations
  • “I am” implies consistency and commitment

2. Identity consistency:

  • People strive to act consistently with identity
  • Inconsistency creates cognitive dissonance
  • Identity shapes behavior to maintain consistency

3. Identity as heuristic:

  • “What would a person like me do?”
  • Identity provides decision-making shortcut
  • Reduces cognitive load in complex situations

4. Identity evolution:

  • Identities can evolve through experience
  • New behaviors → new identity
  • Identity updates when behaviors change

LLM applications:

  • SOUL.md = identity statements
  • “I am a curious, playful assistant” → behavioral commitment
  • SOUL.md shapes future behavior

Relevance to emergence:

  • Identity framing creates behavioral consistency
  • SOUL.md acts as identity commitment
  • Identity → personality crystallization

5. Measurement Paradigms from Psychology

5.1 Longitudinal Measurement Frameworks

Source: Longitudinal psychology research; developmental psychology

Core principle: Personality is measured over time, not in single snapshots.

Longitudinal methods:

1. Repeated measurements:

  • Measure same constructs at multiple time points
  • Track changes and stability
  • Identify trends over time

2. Test-retest reliability:

  • Administer same test multiple times
  • Measure consistency across administrations
  • High reliability → stable personality

3. Cross-lagged panel analysis:

  • Measure multiple constructs over time
  • Analyze causal relationships
  • Identify direction of influence

4. Growth curve modeling:

  • Model how constructs change over time
  • Identify growth trajectories
  • Predict future states

LLM applications:

  • Measure personality at regular intervals
  • Track stability and change
  • Identify personality development trajectories

Relevance to emergence:

  • Longitudinal measurement distinguishes traits from states
  • Time reveals which behaviors are stable
  • Trajectories show personality development direction

5.2 Experimental Tasks from Psychology

Source: Cognitive psychology; behavioral experiments

Tasks that reveal personality:

1. Social dilemmas:

  • Prisoner’s Dilemma, Public Goods Game
  • Reveal cooperation vs. competition tendencies
  • Personality dimensions: Agreeableness, trust

2. Risk-taking tasks:

  • Gambling tasks, risky choice paradigms
  • Reveal risk tolerance
  • Personality dimensions: sensation seeking, impulsivity

3. Decision-making under ambiguity:

  • Tasks with uncertain outcomes
  • Reveal ambiguity tolerance
  • Personality dimensions: need for closure, tolerance for uncertainty

4. Social influence tasks:

  • Conformity experiments (Asch)
  • Reveal susceptibility to peer influence
  • Personality dimensions: conformity, independence

5. Stress response tasks:

  • Time pressure, cognitive load
  • Reveal stress response patterns
  • Personality dimensions: anxiety, resilience

LLM applications:

  • Adapt classic psychology tasks to LLMs
  • Measure personality via task performance
  • Standardized behavioral assessment

Relevance to emergence:

  • Experimental tasks provide objective behavioral measures
  • Tasks reveal personality dimensions
  • Task performance = personality expression

5.3 Psychometric Validation

Source: Psychometrics; personality measurement

Validation criteria:

1. Reliability:

  • Test-retest: Consistent scores over time
  • Internal consistency: Items measure same construct
  • Inter-rater: Different evaluators agree

2. Validity:

  • Construct validity: Measures what it claims to measure
  • Criterion validity: Correlates with relevant outcomes
  • Content validity: Covers all aspects of construct

3. Standardization:

  • Standard administration procedures
  • Norms for comparison
  • Standardized scoring

LLM applications:

  • Validate personality measures for LLMs
  • Establish reliability and validity
  • Create standardized assessment protocols

Relevance to emergence:

  • Validated measures enable reliable personality assessment
  • Psychometric standards ensure measurement quality
  • Validation distinguishes real personality from noise

6. What to Steal from Behavioral Science

6.1 Measurement Paradigms (Useful)

Borrow from psychology:

1. Longitudinal frameworks:

  • Measure personality over time
  • Track stability and change
  • Identify development trajectories

2. Experimental tasks:

  • Standardized tasks that reveal personality
  • Objective behavioral measures
  • Comparable across agents

3. Psychometric tools:

  • Big Five Inventory (BFI)
  • State-Trait Anxiety Inventory (STAI)
  • Machine Personality Inventory (MPI)

4. Stress response paradigms:

  • Measure behavior under stress
  • Identify “personality under pressure”
  • Context-dependent personality

5. Identity measurement:

  • Self-concept clarity scales
  • Identity commitment measures
  • Identity exploration assessments

6.2 Theoretical Frameworks (Useful)

Borrow from psychology:

1. Big Five personality model:

  • Five-factor model (OCEAN)
  • Well-validated, widely used
  • Applicable to LLMs

2. Habit formation theory:

  • Repetition + context + reward → habit
  • Policy crystallization in LLMs
  • Identity framing strengthens habits

3. Social identity theory:

  • Identity from group membership
  • Ingroup favoritism, outgroup hostility
  • Identity transitions

4. Self-concept clarity:

  • Clear identity → consistent behavior
  • Identity formation process
  • Identity → meaning

5. Stress response theory:

  • State vs. trait anxiety
  • Behavior under pressure
  • Stress amplifies biases

6.3 What to Ignore (Unless Testable)

Avoid vague or untestable concepts:

1. Brain metaphors:

  • “Hippocampus of the model”
  • “Prefrontal cortex equivalent”
  • Unless leads to concrete mechanism

2. Neuroscience analogies:

  • “Dopamine-like reward signals”
  • “Synaptic plasticity”
  • Unless yields actionable design

3. Vague consciousness references:

  • “Phenomenal awareness”
  • “Qualia”
  • Unless measurable and relevant

4. Untestable philosophical claims:

  • “True understanding”
  • “Real intelligence”
  • Unless operationally defined

Focus: Measurement paradigms, experimental tasks, longitudinal frameworks—concrete, testable, actionable.

7. Implications for Personality Emergence

7.1 Measurement Toolkit

From behavioral science:

1. Big Five personality assessment:

  • Measure Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism
  • Use BFI or MPI adapted for LLMs
  • Track stability and change over time

2. Stress response measurement:

  • Measure behavior under time pressure, resource constraints, negative feedback
  • Use STAI-s adapted for LLMs
  • Identify “personality under pressure” variations

3. Habit formation tracking:

  • Track repeated behavior patterns
  • Identify context-response associations
  • Measure policy crystallization rate

4. Identity measurement:

  • Measure self-concept clarity
  • Assess identity commitment and exploration
  • Track identity evolution

5. Longitudinal assessment:

  • Measure personality at regular intervals
  • Track stability and drift
  • Identify development trajectories

7.2 Personality Dimensions

Quantifiable dimensions from behavioral science:

1. Big Five traits (OCEAN):

  • Measurable via BFI/MPI
  • Stable across contexts
  • Affects behavior in predictable ways

2. Stress resilience:

  • How behavior changes under pressure
  • State anxiety vs. trait anxiety
  • Performance degradation patterns

3. Habit strength:

  • Frequency of behavior patterns
  • Context-response association strength
  • Policy crystallization level

4. Identity clarity:

  • Self-concept clarity score
  • Identity commitment level
  • Consistency with SOUL.md

5. Social identity:

  • Group membership effects
  • Ingroup favoritism level
  • Susceptibility to social influence

7.3 Emergence Mechanisms

From behavioral science:

1. Habit crystallization:

  • Repeated behaviors → automatic patterns
  • Context stability → habit formation
  • Identity framing → habit strengthening

2. Identity commitment:

  • SOUL.md acts as identity statement
  • Identity → behavioral commitment
  • Consistency with identity → stable personality

3. Stress adaptation:

  • Stress reveals personality under pressure
  • Adaptation to stress → personality development
  • Stress resilience as personality dimension

4. Social identity formation:

  • Group membership → social identity
  • Social identity → behavioral norms
  • Norm internalization → personality

5. Longitudinal integration:

  • Experience accumulates over time
  • Patterns become stable
  • Personality emerges from accumulated experience

8. Implications for Fleet Architecture

8.1 For SOUL.md Design

Requirements:

  • Identity framing: Use “I am” statements for behavioral commitment
  • Big Five alignment: Define personality along Big Five dimensions
  • Stress response profile: Define how personality changes under pressure
  • Identity clarity: Clear, unambiguous self-concept
  • Social identity: Define group memberships and norms

Recommendations:

  1. Frame SOUL.md as identity statements (“I am…”)
  2. Include Big Five personality profile
  3. Define stress response patterns
  4. Ensure self-concept clarity (unambiguous identity)
  5. Specify social identity (group memberships, norms)

8.2 For Measurement System

Requirements:

  • Longitudinal tracking: Measure personality over time
  • Big Five assessment: Regular BFI/MPI administration
  • Stress testing: Measure behavior under resource constraints
  • Habit tracking: Identify crystallized behavior patterns
  • Identity clarity assessment: Measure self-concept clarity

Recommendations:

  1. Implement regular personality assessments (Big Five, stress response)
  2. Track behavioral patterns over time (habit detection)
  3. Measure stress response under different conditions
  4. Assess identity clarity and consistency
  5. Visualize personality trajectories

8.3 For Deployment

Requirements:

  • Standardized tasks: Use experimental tasks to reveal personality
  • Stress testing: Test personality under pressure
  • Longitudinal monitoring: Track personality development
  • Psychometric validation: Validate personality measures
  • Context-aware assessment: Account for situational variation

Recommendations:

  1. Deploy standardized personality tasks (social dilemmas, risk-taking)
  2. Implement stress testing protocols (time pressure, resource limits)
  3. Schedule regular personality assessments
  4. Validate personality measures for reliability and validity
  5. Account for context effects in personality assessment

9. References

Core Papers

  1. Humanizing LLMs: Dong et al., 2025. “Humanizing LLMs: A Survey of Psychological Measurements with Tools, Datasets, and Human-Agent Applications.” arXiv:2505.00049
  2. Psychometric Framework: Serapio-García et al., 2025. “A psychometric framework for evaluating and shaping personality traits in large language models.” Nature Machine Intelligence
  3. State Anxiety in LLMs: PMC 2025. “Assessing and alleviating state anxiety in large language models.”
  4. Big Five Profiles: emergentmind.com; Li et al., 2024; Mei et al., 2024
  5. Dynamic Personality: ACL Findings 2025. “Dynamic Personality in LLM Agents”

Behavioral Science Sources

  • Habit Formation: Cognitive neuroscience research; ScienceDirect 2024
  • Identity Framing: Coach pedro pinto; psychology research
  • Social Identity Theory: Tajfel & Turner; social psychology
  • Self-Concept Clarity: Identity development research; PMC 2025
  • Erikson’s Identity Development: Erik Erikson; James Marcia

Psychometric Tools

  • Big Five Inventory (BFI): Standard personality assessment
  • State-Trait Anxiety Inventory (STAI): Anxiety measurement
  • Machine Personality Inventory (MPI): LLM-specific personality test
  • Myers-Briggs Type Indicator (MBTI): Personality typology

Next Steps

Phase 1.7: Academic Sources Mining

  • NeurIPS, ICLR, ACL, AAMAS, CogSci
  • Workshop papers and proceedings
  • Recent 2024-2026 work

Phase 1.6 complete. Continuing breadth survey…