Narrative as active inference: an integrative account of cognitive and social functions in adaptation

The paper addresses the lack of an integrative mechanistic framework linking narrative to human adaptation. It proposes that many cognitive and social functions of narratives can be modeled via active inference, a predictive processing framework where organisms minimize prediction error through perception, learning, and action. The authors focus on narrative structure (e.g., coherence, temporal and causal organization, master narratives) rather than content. They motivate their account by re-examining work that treats narratives as cognitive schemas, cultural tools, and storytelling practices, and argue that narratives evolved partly to enhance individual prediction and coordinate group action. They outline how narratives, by organizing events and their relations, enrich generative models for adaptive behavior across individual and sociocultural scales.

The paper synthesizes several literatures: (1) Event narratives and episodic memory/future projection—narratives segment experience into events, aiding encoding, recall, and construction of future scenarios; episodic memories often follow narrative templates and support adaptive future simulation (Labov & Waletzky; Tulving; Schacter & Addis). Empirical work links narrative segmentation and hippocampal processes to prediction-error and memory updating. (2) Narrative identity—life stories provide coherence and meaning, linking past experiences to future goals; coherence and meaning-making correlate with well-being and support self-continuity and goal pursuit (McAdams; Habermas; Adler). (3) Socio-cultural narratives—Narrative Practice Hypothesis treats storytelling as social practice that scaffolds folk-psychological reasoning and joint attention; developmental work shows caregiver scaffolding shapes autobiographical memory and narrative skills (Nelson & Fivush; Hutto). (4) Master Narrative Framework—master narratives are ubiquitous, often implicit cultural templates guiding life course, social norms, and identity; they are useful, ubiquitous, invisible, compulsory, and rigid (Hammack; McLean & Syed). (5) Evolutionary and cultural perspectives—narratives as information exchange enabling coordination, planning, and knowledge transmission; contemporary media amplify narrative dynamics in coordination and identity formation. Across these strands, the authors argue narrative functions dovetail with active inference processes of prediction, uncertainty reduction, and model updating.

This is a conceptual and integrative theoretical paper. The authors: (1) provide a brief, non-technical overview of active inference (predictive coding, generative models, perception-learning-action/policy selection, epistemic foraging); (2) conduct a selective narrative-focused review across psychology, anthropology, philosophy, and cultural studies (event narratives, episodic memory and future projection, narrative identity, storytelling practices, enculturation, master narratives); (3) map these narrative phenomena onto an active inference framework, proposing a multi-level account spanning event-level cognition, identity-level organization, and socio-cultural practices; (4) formalize the iterative dynamics conceptually via a POMDP-style cycle (schema updating after surprises; construction of coherent identity; social self-presentation; environmental change yielding deontic cues/master narratives); and (5) outline empirical research designs to test predictions (e.g., longitudinal prediction-accuracy tasks contrasting narrative vs non-narrative representations; intention-only vs intention-narrative planning paradigms; narrative-assisted model updating after prediction failures; examining predictive utility of master vs alternative narratives). No new empirical data are collected; the paper advances testable hypotheses and suggested designs.

- Core thesis: Narratives function as tools for active inference by creating and managing predictions across cognitive and social levels; narratives are primarily for the future, even when recounting the past. - Event narratives: Narrative segmentation supports episodic memory and episodic future thinking; narratives provide a pool of past event information for recognizing regularities and recombining elements to simulate future events, balancing conservative (habit/routine) and constructive (recombination/counterfactual) processes for optimal prediction. Surprises increase memory for context and drive model updating (e.g., hippocampal involvement). - Narrative identity: Coherence reduces internal contradictions and prediction error, supporting stable expectations about life trajectories and goals for self and others; meaning-making serves epistemic foraging, updating models after unexpected events. Coherent identities enhance predictability (self and social), aiding adaptation and well-being; negative but coherent narratives can also persist by reducing prediction error, explaining maladaptive stability (e.g., learned helplessness, trauma reenactment). - Socio-cultural narratives: Storytelling practices scaffold attention, salience, and culturally normative behaviors, enabling mutual prediction and coordination. Master narratives synchronize group members’ predictions about behaviors and life courses, explaining invisibility, ubiquity, compulsivity, and rigidity via predictive efficiency and social pressure to reduce others’ prediction errors. Adoption of alternative narratives can occur when they better predict an individual’s reality despite social costs. - Integrative contribution: A parsimonious active inference model spans individual cognition, identity construction, and socio-cultural coordination, generating empirically testable predictions and design suggestions (e.g., narrative descriptions should improve predictive accuracy and planning vs non-narrative representations).

The proposed framework addresses the central question of how narratives contribute to human adaptation by modeling their cognitive and social functions as prediction-oriented processes under active inference. By treating narratives as structures that segment events, connect causes over time, and coordinate social expectations, the account explains why narrative coherence and meaning-making facilitate adaptive functioning: they minimize prediction error and support effective policy selection for individuals and groups. The model clarifies mechanisms behind both beneficial and harmful narrative dynamics—how coherent negative narratives can persist by accurately forecasting adverse outcomes, and how master narratives stabilize social coordination while resisting change. It integrates internalist (cognitive schemas, memory) and externalist (storytelling practices, deontic cues, master narratives) perspectives into a single mechanistic loop that can be probed with empirical paradigms. This reframing shifts narrative research toward future-oriented functions (e.g., “Who am I becoming?”), with implications for identity development, mental health, social cohesion, and cultural change. The discussion highlights trade-offs between predictive efficiency and flexibility, the role of exploration in model updating, and social consequences of deviating from master narratives, setting the stage for targeted empirical tests and potential clinical interventions informed by predictive processing.

The paper advances an active inference account of narratives that integrates cognitive and social functions: narratives segment and organize event information for memory and future projection; support meaning-making and identity coherence to reduce prediction error; and operate through storytelling practices and master narratives to synchronize social expectations and coordination. It emphasizes that narratives are fundamentally future-oriented tools for prediction at individual and collective scales. The framework provides computationally grounded, testable hypotheses and suggests empirical designs to assess the predictive validity of narratives and their role in planning and model updating. While not a comprehensive theory of all narrative phenomena (e.g., aesthetics, entertainment), it offers a parsimonious mechanistic basis to study narrative dynamics in adaptation and psychopathology, and points to future research on the linkage between past and future narratives, predictive accuracy, goal progress, and the conditions under which alternative narratives emerge and propagate.

- Scope: The account does not aim to explain all narrative functions (e.g., full aesthetics of art/entertainment, persuasion) and only partially addresses maladaptive narrative dynamics. - Empirical status: The model is theoretical; proposed mechanisms and predictions require targeted empirical validation and formal modeling beyond the presented conceptual POMDP sketch. - Content vs structure: While focusing on structure (coherence, temporal/causal organization), outcomes can depend critically on narrative content (e.g., negative coherent narratives can harm well-being despite predictive efficiency). - Cultural variability and change: The framework acknowledges but does not fully model frequencies of updating under different niche dynamics, or the complex power structures maintaining master narratives. - Psychopathology: It suggests pathways (e.g., learned helplessness, trauma reenactment) but does not provide a full typology or clinical validation of narrative pathologies.