Methodologyhow this codex is built.

Bias Codex covers 200 cognitive biases, heuristics, and closely related psychological effects. Inclusion requires that the phenomenon is described in the peer-reviewed psychology, behavioral economics, or decision-science literature, or is an established named concept whose omission would leave a gap in the map.

A small number of entries are not cognitive biases in the strict sense — reasoning principles like Occam's razor, adages like Murphy's law, psychophysical laws like Weber–Fechner, or adaptive functions like memory inhibition. These are kept because they are frequently searched alongside biases and because they underlie or interact with genuine biases. Each one is explicitly labeled as a related concept on its page, with a note explaining the distinction.

Every entry cites the canonical primary literature: the paper (or book) that introduced the phenomenon, plus, where available, a major review or meta-analysis. Citations list real authors, years, and journals, and link to a Google Scholar search for the exact title so any reader can verify the source and find an accessible copy in one click.

We do not cite sources we could not verify. Where a named effect has little or no direct empirical literature of its own (for example, the Delmore effect), the entry says so plainly in its critiques section, and cites the nearest relevant research instead.

Psychology's replication crisis changed how much trust individual findings deserve, and a serious reference should say so. Entries carry an evidence label where we have assessed the literature: Robust (replicates reliably across labs and methods), Mixed (real but conditional — significant moderators or inconsistent replications), Contested (the phenomenon or its interpretation is actively disputed, e.g. the Dunning–Kruger effect's statistical-artifact debate), and Weak (limited evidence or failed large-scale replication, e.g. the backfire effect).

These labels are editorial judgments grounded in meta-analyses and registered replication reports cited on each page. They are revisited as the literature moves.

Each bias belongs to one of seven functional groups (perception, reasoning, attribution, decision-making, memory, social influence, motivation) based on the primary cognitive process involved. Many biases straddle groups; we assign by center of gravity and use cross-links to express the rest.

Related-bias links are hand-curated. Where two effects are commonly confused — gambler's fallacy and hot-hand fallacy, groupthink and the Abilene paradox, endowment and mere ownership — the link includes a short note explaining how they differ.

Case studies on bias pages are teaching narratives that illustrate how a bias plays out in realistic business settings. They are composites, not reports of specific documented events, and should be read as illustrations rather than evidence.

Every entry displays the date it was last reviewed. A review covers naming, definition accuracy, citation verification, classification, and cross-links.

If you find an error — a miscited paper, an outdated replication status, a wrong definition — use the contact form. Substantiated corrections are applied and the review date updated.