Where Bias Enters: A Differentiated Realist Map of Cognitive Distortion.

Research on cognitive bias has established a heterogeneous body of findings concerning recurrent distortions of judgment and decision-making, including availability, representativeness, anchoring, framing, confirmation processes, and sunk-cost effects. The same literature also shows that heuristic economy and cognitive selectivity are not intrinsically irrational: simple strategies can succeed when the information they use is appropriately related to the target and environment. Experimental psychology investigates the effects, causal psychological processes, boundary conditions, and interventions associated with these phenomena. A further philosophical question concerns the epistemic and practical status of the cognition thereby produced: what has been understood, what has been judged, what is being decided, and which standards govern the relations by which those answers are supported?This article develops that question through Differentiated Realism. Attention opens and orients a finite field of access toward what is encountered; Primary Apprehension first receives what is encountered as something intelligible before essence is clarified or the object is judged, classified under a system, measured, formalized, or used; abstraction seeks constitutive intelligibility through relevant variation; and formalization selects and stabilizes aspects or relations according to a proper operative end. These differentiated relations can contribute to later judgment and practical life together with memory, testimony, evidence, reasons, prior judgments, and other mediations. Judgment, reasoning, deliberation, practical determination, decision, and action remain functionally distinct rather than forming a necessary sequence. Deliberation can concern different questions; when the question is what is to be done, practical deliberation may occur, may be absent, or may remain internally proportionate to materials already distorted elsewhere.Differentiated Realist diagnosis therefore distinguishes the standards appropriate to the relations involved. Where understanding is at stake, the question is whether the relevant variation through which abstraction seeks what is constitutive has been adequate. Where formalization is at stake, the form must be assessed relative to what it selects or represents, its defined scope, and its proper operative end: a form may be inadequate to that task, or later be overextended despite being adequate within it. Where judgment is at stake, the target question must be specified and the epistemic contribution of each consideration differentiated. Epistemic role concerns what a consideration is entitled to contribute; weight concerns how much force that contribution receives. The scope of warrant constrains how broadly a conclusion may extend, while its strength constrains how confidently the conclusion may be treated as established. Where practical determination is at stake, the question is what genuinely counts for or against the available courses of action and what practical significance those considerations should have in light of circumstances, consequences, obligations, and ends.Kahneman and Frederick's attribute substitution names a specific psychological mechanism in which an accessible heuristic attribute is evaluated in place of a more difficult target attribute. It can instantiate substitution in the broader Differentiated Realist sense, but the two concepts operate at different explanatory levels. In judgment, weighting distorts force within a legitimate bearing, whereas substitution transfers epistemic authority or sufficiency beyond the question, attribute, relation, epistemic role, scope, or criterion that warrants it. More broadly, substitution can also concern an unwarranted transfer across practical criteria or ends. It is not defined by ease or accessibility.Across availability, representativeness, anchoring, framing, confirmation processes, and sunk-cost effects, the relevant disproportion can become operative in different relations. Availability and representativeness may affect the comparative field of understanding or the evidential treatment of a later judgment; anchoring may distort the force of a starting value; confirmation processes may become disproportionate in judgment or reasoning; framing may alter evaluation or practical significance; and sunk-cost effects concern the bearing of irrecoverable prior investment on present choice. Formalized representations can participate in such cases either by being inadequate to their own scope and operative end or by acquiring later authority beyond what an otherwise adequate form supports. The phenomena remain causally heterogeneous even where their philosophical structures can be compared.The article also distinguishes architectural entry, affected relation, and diagnostic point. Architectural entry names the first differentiated relation within a case at which a case-specific disproportion becomes operative; affected relation identifies a later cognitive, practical, or socially mediated relation shaped by that disproportion; diagnostic point identifies where a relation becomes assessable according to its proper standard. These may coincide, but they need not. Decisions and actions can extend the history of a disproportion by altering evidence, opportunity, records, expectations, and social representations available to later cognition without thereby relocating its original architectural entry.These distinctions also clarify corrective leverage. Where a field, question, role, scope, criterion, or end has been narrowed or displaced, correction requires reopening or restoration; where a legitimate contribution has received disproportionate force, correction requires recalibration. A distorted judgment or decision may therefore require correction at a relation earlier than the point at which its consequence becomes visible, while recursive effects may require attention to altered evidential or social conditions. Philosophical diagnosis can identify what would have to be restored or recalibrated without establishing that a particular intervention will be empirically effective.