ARM-3 separates cognitive performance from the epistemic direction in which cognition is used. The model asks not only whether a person can think, but whether thinking serves model correction, model protection, or no deeper examination at all.
The terms do not describe biological or moral categories of people. They denote dominant epistemic modes between which the same person may shift depending on context and pressure.
The capacity for reflective thinking is present but is not reliably activated. Judgments may arise primarily through heuristic, intuitive, habitual, or socially adopted processing.
Cognitive resources are primarily used to form a model that is as accurate as possible. Counterevidence can change the weight of a belief; uncertainty is allowed to remain.
Under increasing Rationalization Pressure, cognitive resources are increasingly used to preserve an existing or desired model. High intelligence may enable more complex justifications.
The existing test dimensions remain as observable primary scales. Above them sits a new theoretical layer intended to explain how model revision and model protection may arise.
How strongly cognitive capacity is actually mobilized for active analysis. In v1.1, primarily conceptualized from Cognitive Performance and Need for Cognition.
The ability to monitor one’s own assumptions, confidence, reasoning errors, and affective influences, and to regulate information processing accordingly.
The ability to keep competing explanations open, shift perspectives, tolerate uncertainty, and revise a model when better evidence is available.
The degree to which a belief is cognitively, emotionally, biographically, or socially anchored in the existing self-model and world model.
The subjective costs of fundamental revision, for example to self-image, biography, status, group membership, or prior decisions.
How sensitive a system is to threat and how strongly the observed mode actually shifts between neutral and stressful conditions.
The older ARM-3 v1.0 model combined Cognitive Activation and Defensive Pressure into Rationalization Power. v1.1 instead separates the pressure to defend a model from the cognitive ability to implement that pressure argumentatively.
A highly activated cognitive system may therefore be either especially revisable or especially elaborate in its rationalization. What matters is not only the amount of thinking, but its epistemic direction.
This is not a validated measurement equation. It visualizes a hypothesis. Weights and functional form would have to be determined empirically.
Strong cognitive activation combined with sufficient flexibility and metacognitive control favors evidence-oriented model revision.
The same cognitive resources may be used to search for counterarguments, auxiliary hypotheses, and model-protective explanations.
The capacity for rational examination may be present without being sufficiently mobilized.
A generally rational mode may shift under ego, status, group, commitment, affective, or uncertainty stress.
10 tasks covering logic, conditional reasoning, Bayes, percentage reasoning, causality, confounding, reflection, and falsification – including Confidence Calibration.
60 self-report items on Need for Cognition, metacognition, Intellectual Humility, Open-Minded Thinking, uncertainty, and ego, status, group, commitment, and affective dimensions.
24 scenarios compare neutral and threat conditions and assess knowledge, likely behavior, rationalization tendencies, and withdrawal tendencies.
The digital instrument currently uses scoring v1.0.x. The v1.1 mechanism layer is a theoretical development and will only be integrated into scoring after item reclassification and methodological review.
The result does not reduce ARM-3 to a single label. It combines Animal State, mechanisms, prototype similarity, and State Transition into a multidimensional profile. The values below are deliberately fictional and are shown only to demonstrate the result format.
84% prototype similarity
In this fictional profile, high cognitive activation is paired with comparatively strong metacognitive control and flexibility. Within the model, this configuration favors model revision over model protection. Under stronger personal commitment, however, the state may still shift.
Animal State · mechanisms · 10 primary dimensions · defensive profile · State Transition · archetypes · data quality · Critical Question · overall assessment
Note: All scores shown here are illustrative. They are not taken from a real participant, are not normative values, and do not indicate population distributions.
View the complete ARM-3 taxonomy in Research →