The pessimistic meta-induction (PMI) is an argument in the philosophy of science holding that the historical record of successful but later-abandoned scientific theories gives inductive grounds for doubting that current theories are true, or that their theoretical posits genuinely refer to features of the world. The argument proceeds by analogy: just as past scientists accepted theories that were predictively successful yet are now considered false or radically incomplete, present scientists may be similarly situated with respect to today's best theories. The precise formulation, scope, and force of the argument are contested; what counts as a “successful” theory, what failure amounts to, and whether the induction is well-formed are all subjects of active debate.

The PMI was given its canonical modern formulation by Larry Laudan in his 1981 paper “A Confutation of Convergent Realism,” where he assembled a list of historically successful theories - including the phlogiston theory of combustion, the caloric theory of heat, the crystalline-sphere model of celestial motion, and the ether theory of light propagation - whose central theoretical posits are no longer accepted.¹⁾ Laudan used this list as an inductive base to argue against Scientific Realism, and in particular against the inference from predictive success to approximate truth.

Scientific realists have responded with several strategies. Stathis Psillos and others advanced the divide et impera (selectivity) response, arguing that what was genuinely responsible for a past theory's predictive success - its “working posits” - was often retained or approximately preserved through theory change, even when the theory's broader framework was abandoned.²⁾ Phillip Kitcher similarly distinguished between idle theoretical posits and those that actively contributed to successful predictions.³⁾

A separate line of realist response, associated with Hasok Chang and others, questions the composition of Laudan's historical sample, arguing that it cherry-picks dramatic episodes of discontinuity while underweighting the many cases of genuine theoretical continuity across scientific revolutions.⁴⁾

Anti-realists - including proponents of Scientific Instrumentalism and Constructive Empiricism - have generally welcomed the PMI as support for restricting scientific knowledge claims to observable phenomena, or for treating theories instrumentally rather than as candidate descriptions of unobservable reality. Some have noted, however, that the PMI is itself an inductive argument, and its standing depends on contested assumptions about how induction from historical cases to present ones is licensed.

Debates about the PMI connect closely to questions in Entity Realism, which concedes the argument's force against high-level theoretical frameworks while defending realism about particular entities that can be experimentally manipulated. Structural Realism, particularly in its ontic form, represents another realist response, proposing that structural relationships rather than object-level posits are what theories preserve across change.

- Scientific realism (selectivity response): Current theories, or the portions of them bearing on predictive success, are approximately true and refer to real features of the world; the PMI misidentifies what in past theories was doing explanatory work. See pessimistic-meta-induction-scientific-realism-selectivity-viewpoint.

- Structural realism: The PMI undermines object-level realism but is compatible with realism about the mathematical or structural content of theories, which is typically retained through theory change.⁵⁾ See pessimistic-meta-induction-structural-realism-viewpoint.

- Entity realism: The PMI applies to theoretical laws and frameworks but not to entities that scientists can manipulate and deploy experimentally; realism is warranted at the level of causally active entities.⁶⁾ See pessimistic-meta-induction-entity-realism-viewpoint.

- Scientific instrumentalism: The PMI vindicates an instrumentalist reading of theories; theories need not be true to be useful, and success is no guide to truth about unobservables. See pessimistic-meta-induction-scientific-instrumentalism-viewpoint.

- Constructive empiricism: In line with Bas van Fraassen's anti-realism, the PMI supports limiting the epistemic aim of science to empirical adequacy rather than truth about the unobservable world.⁷⁾ See pessimistic-meta-induction-constructive-empiricism-viewpoint.

- Base-rate / no-miracles counter-response: Some philosophers argue the PMI is a poorly constructed induction, either because the reference class of “successful theories” is under-defined, or because a proper base-rate analysis would not undermine confidence in current well-confirmed theories. See pessimistic-meta-induction-base-rate-objection-viewpoint.

- pessimistic-meta-induction-laudan-list-debate - Whether Laudan's historical examples constitute a representative or well-chosen inductive base - pessimistic-meta-induction-theory-change-continuity-debate - The extent to which scientific theory change involves preservation of theoretical content - pessimistic-meta-induction-inductive-strength-debate - Whether the argument is structurally sound as an induction

- pessimistic-meta-induction-realism-vs-instrumentalism-viewpoint-debate - pessimistic-meta-induction-structural-realism-vs-entity-realism-viewpoint-debate

- pessimistic-meta-induction-history - Development of the argument from Poincaré and Duhem through Putnam and Laudan to contemporary responses

- Scientific Realism - Scientific Instrumentalism - Entity Realism - Structural Realism - Constructive Empiricism - Theory Change - underdetermination-of-theory