Capacity
Cognitive Capacity (C) concerns available task-relevant reasoning operations. Capacity is not the same construct as momentary activation.
Anthropological Rationality Model · Foundation v1.1.3
Understand epistemic
processing.
ARM-3 examines which cognitive resources are available, how strongly they are mobilized in a situation, and how evidence-sensitive that mobilized cognition remains when revision becomes motivationally costly.
Canonical Core: Foundation Version 1.1.3Three ideal-typical, context-sensitive processing modesLegacy Instrument v1.0 separated from the Core
ARM-3 Research · Pilot phase
Research approach
Knowledge through integration
ARM-3 does not seek to reinvent human rationality; it is an integrative research approach. It brings together insights from philosophy, cognitive science, and psychology to better understand how established mechanisms interact and to derive empirically testable hypotheses. Scientific perspectives are invited into dialogue rather than set against one another.
Activation
Cognitive Activation (CA) is the state-dependent mobilization of available resources — not ability, Need for Cognition, or response accuracy.
Protection conditions
MC, MRC, and TA are measured separately. MPP is only the theoretical envelope for their protection-relevant configuration — not a score.
Epistemic behavior
Model correction and model protection are inferred from theory-neutral outcomes such as weighting, search, standards, confidence, and posterior beliefs.
Three ideal-typical modes
Not three kinds of people.
Context-sensitive processing descriptions.
Mode language is secondary to the dimensional variables from which it is inferred. The same person may show different patterns across domains, evidence conditions, and motivational contexts.
Modes in detail →01
Low mobilization
Animal rationabile
Reflective capacity appears available, but is comparatively weakly mobilized.
It cannot be inferred from poor performance alone and does not imply low intelligence.
02
Evidence-sensitive regulation
Animal rationale
Active cognition remains responsive to evidence quality and stable evaluative standards.
The criterion is proportionality to evidence, not maximal belief change or chronic doubt.
03
Preferential model protection
Animal rationalizans
Active cognition shows preferential model protection under motivationally relevant stakes.
The mode is not inferred from disagreement or persistence alone, but from motivationally contingent epistemic asymmetry under controlled evidence.
Minimal ARM-3 claim · v1.1.3
Available cognition ≠ mobilized cognition ≠ epistemically well-regulated cognition.
Under protection-relevant conditions, greater mobilization of available cognitive resources may support either evidence-sensitive correction or directionally selective model protection.
MPP · no formula
MC · MRC · TA
Model-Protection Pressure is a theoretical envelope around separately measured antecedents. CF and EMC remain outside that envelope as candidate regulators.
State dependence in the Core
Modes are context-sensitive,
not permanent assignments.
ARM-3 separates stable between-person tendencies from situational variation. Broader attractor, developmental, and lifecycle models are theoretical extensions, not part of Foundation v1.1.3.
Priors, source reliability, and task structure are taken into account.
Revision can become motivationally consequential.
Only observed epistemic behavior licenses a higher-level interpretation.
Core state sensitivity · extensions separateView dynamic extension →
Risky core test
Activation × Protection.
The strongest specific prediction in v1.1.3 is a positive preregistered interaction between randomized activation encouragement and randomized protection pressure on Evidence Weighting Asymmetry (EWA), conditional on successful independent first-stage checks.
Hypotheses & falsification →Normative identification
Model protection is not inferred from non-revision alone, but relative to evidence, priors, and source reliability.
Adverse models
ARM-3 must compete against simpler models and strong engagement × directional-motivation rivals.
Revisability
If constructs or the core interaction fail, the model should simplify rather than reinterpret failure as support.
Legacy Research Instrument v1.0
Exploratory instrument — not Foundation v1.1.3.
The existing 94-item assessment remains available for research and backward compatibility, but it is not a validated operationalization of the frozen Core Paper.