Cognitive Bias Library

What the library is

Rubicon Probity's bias detection engine draws on a structured library of 153 cognitive biases, each catalogued with its category, typical decision stages, severity rating, and business impact. The library was seeded on day four of the platform's existence - before any decision UI existed. The intellectual foundation was built first.

The 153 patterns span eight categories covering how humans perceive information, form memories, reason under uncertainty, weigh value, protect ego, respond to social pressure, react to organisational incentives, and plan and execute. In a given decision, Probity surfaces the subset most relevant to the current context - not all 153 simultaneously, but the patterns that the decision's structure, stakes, and framing make most likely.

Detailed pages are available for the 15 highest-frequency strategy patterns. The full library is listed below.

How the library was built and how it stays honest

The 153 patterns were catalogued from peer-reviewed cognitive science, behavioural economics, and decision research. The starting set was distilled from the published work of Daniel Kahneman, Amos Tversky, Gary Klein, Richards Heuer, Ray Dalio, Roger Martin, and the broader academic corpus on judgement and decision-making - names that anyone serious about the field will recognise. Categorisation, severity rating, and stage assignment were authored by Rubicon and remain editorial choices we own.

When you make a decision through Rubicon Probity, the system matches the situation to the patterns most likely to fire at that decision stage and surfaces them with detection questions and mitigations. Every bias surfaced is recorded against the decision in the audit trail. Every outcome reviewed feeds a calibration loop that gets better at flagging the biases that matter for your context.

The library grows. Decision research is an active field - the rise of agentic AI has surfaced patterns (automation bias, algorithm aversion, anthropomorphism in machine-mediated reasoning) that were not on anyone's radar a decade ago. New entries are added as the literature settles, and Rubicon Scout monitors the research surface to keep the catalogue current.

Category A: Attention and Perception (20 biases)

The biases that govern what information reaches conscious attention and how it is initially framed.

Category B: Memory, Learning, and Interpretation of the Past (15 biases)

The biases that distort how experience is stored, recalled, and used as evidence.

Category C: Inference Under Uncertainty (35 biases)

The largest category. The biases that distort reasoning when evidence is incomplete, ambiguous, or probabilistic.

Category D: Value, Preferences, and Choice Architecture (22 biases)

The biases that distort how options are valued, compared, and selected.

Category E: Commitment, Ego, and Escalation (14 biases)

The biases that arise when identity, reputation, and prior commitment distort judgement.

Category F: Social, Identity, and Group (21 biases)

The biases that emerge from social pressure, identity, and group dynamics.

Category G: Organisational, Incentives, and Governance (11 biases)

The biases that emerge from misaligned incentives, governance structures, and institutional behaviour.

Category H: Planning, Execution, and Control (15 biases)

The biases that emerge during the translation of decisions into delivery.