Active Inference Therapy: A First-Principles Framework for Psychotherapeutic Intervention

9 min readMay 26, 2025

An early academic outline

Abstract

This paper introduces Active Inference Therapy (AIT), a novel psychotherapeutic framework grounded in the free energy principle and active inference theory. AIT conceptualizes the therapeutic process as a collaborative exploration of "psychoceptive space," wherein patient and therapist form a coupled system engaged in mutual model updating. Drawing on Bayesian epistemology, computational psychiatry, and neuroplasticity research, AIT offers a metaframework that transcends traditional therapeutic modalities while providing specific mechanisms for understanding and facilitating psychological change. The framework delineates five distinct phases of treatment, incorporates hierarchical goal-setting, and emphasizes the development of metacognitive capacities for ongoing self-evidencing. This theoretical exposition presents AIT's foundational principles, process model, and potential applications across various psychopathological conditions, while acknowledging its current status as an exploratory framework requiring empirical validation.

Keywords: active inference, free energy principle, psychotherapy integration, computational psychiatry, Bayesian brain, neuroplasticity

Introduction

The search for unifying principles in psychotherapy has long occupied theorists and clinicians seeking to understand the mechanisms of psychological change (Wampold & Imel, 2015). While numerous therapeutic modalities have demonstrated efficacy, the field lacks a comprehensive framework that can account for both the diversity of approaches and their common factors. This paper proposes Active Inference Therapy (AIT) as a potential solution—a metaframework grounded in first principles derived from computational neuroscience and Bayesian mechanics.

Active inference, as articulated by Friston and colleagues (2010, 2014), posits that biological systems maintain their integrity by minimizing prediction error through either updating internal models or acting upon the environment. This principle, when applied to psychotherapy, offers a precise mathematical framework for understanding how therapeutic change occurs through the iterative process of model updating within the patient-therapist dyad.

Theoretical Foundations

The Free Energy Principle in Psychological Context

At its core, AIT builds upon the free energy principle, which states that self-organizing systems resist entropy by maintaining probabilistic models of their environment (Friston, 2010). In the psychological domain, this translates to individuals maintaining predictive models of self, others, and world that guide perception and action. Psychopathology, from this perspective, can be understood as maladaptive priors or inefficient model updating processes.

The Therapeutic Dyad as a Coupled System

AIT conceptualizes the therapeutic relationship as a coupled dynamical system, where patient and therapist engage in mutual active inference. This "analytic unity" operates within a bounded space—the therapeutic frame—which functions as a Markov blanket containing all necessary information for causal inference within the system. This formulation provides mathematical precision to long-standing psychoanalytic concepts such as the "analytic third" (Ogden, 1994) and intersubjective field theory (Stolorow & Atwood, 1992).

Psychoceptive Space: A Novel Conceptual Domain

The paper introduces the concept of "psychoceptive space"—a liminal domain between interoceptive and exteroceptive experience where therapeutic exploration occurs. This is within the Markov blanket, where causal influences are strongest, and is one proposed mechanism by which actions within psycheceptive space lead to developmental changes. This space is neither purely internal nor external but represents the co-constructed reality of the therapeutic encounter. Within this space, both participants engage in what we term "co-self-evidencing," utilizing data from within and outside the therapeutic frame to update predictive models.

The AIT Process Model

Phase Structure

AIT delineates five distinct phases of treatment, each characterized by specific objectives and processes:

Phase 1: Capture - Initial engagement and assessment phase incorporating conventional diagnostic methods alongside AIT-specific evaluations of self-modeling capabilities, personality structure, and baseline self-governance. This phase establishes the therapeutic coupling and determines appropriate vocabulary levels for technical communication.

Phase 2: Initial Evidencing - Characterized by higher model uncertainty and experimental implementation of selected interventions. This phase alternates between structured inquiry based on active inference principles and open exploration, with continuous tracking of responses using precision-weighted prediction error metrics.

Phase 3: Working Phase - Marked by increased baseline model certainty and greater mastery of new predictive frameworks. This phase consolidates understanding, develops specific action plans, and monitors for necessary returns to earlier phases during developmental transitions.

Phase 4: Release - The termination phase focusing on consolidating gains, strengthening self-monitoring abilities, and ensuring internalization of model-updating skills. This phase plans for optimal decoupling of the therapeutic dyad.

Phase 5: Post-termination - Ongoing monitoring phase with structured parameters for tracking potential recurrence and maintaining readiness for re-engagement if needed.

Assessment and Intervention Framework

AIT employs a multi-level assessment approach evaluating:
- Baseline self-evidencing capabilities
- Structural personality organization (unitary vs. plural self-states)
- Model complexity and updating efficiency
- Precision-weighting of interoceptive vs. exteroceptive signals
- Tolerance for surprise and uncertainty

Interventions are tailored to address specific deficits in predictive processing, incorporating techniques from various therapeutic modalities reframed through active inference principles.

Putative Clinical Applications

Psychopathology Through an Active Inference Lens [needs development]

AIT reconceptualizes major psychiatric conditions as disorders of predictive processing:

Major Depressive Disorder - Characterized by overly precise negative priors resistant to updating, leading to anhedonia and behavioral withdrawal.

Anxiety Disorders - Manifest as hyperactive prediction error signaling with impaired precision-weighting of threat-related stimuli.

Trauma-Related Disorders - Understood as maladaptive neuroplasticity with overgeneralized threat detection and impaired model flexibility.

Personality Disorders - Conceptualized as rigid, poorly integrated predictive models of self and others with limited updating capacity.

Temperature or Entropy Mediated Plasticity (TEMP)

AIT introduces TEMP as a mechanism for facilitating model updating by modulating the "temperature" of neural processing—increasing flexibility during critical therapeutic moments while maintaining stability during consolidation phases.

Integration with Emerging Technologies

AIT explicitly incorporates potential AI-assisted tools for:
- Enhanced modeling of patient predictive processes
- Real-time tracking of therapeutic progress
- Simulation of personalized developmental sequences
- Augmented reality interventions for exposure-based treatments

Discussion

Theoretical Implications

AIT offers several theoretical advances:
1. Mathematical formalization of therapeutic process
2. Integration of neuroscientific and psychodynamic principles
3. Precision medicine approach to psychotherapy
4. Framework for understanding common factors across modalities

Limitations and Future Directions

As an exploratory framework, AIT requires:
- Empirical validation of core propositions
- Development of standardized assessment tools
- Clinical trials comparing AIT to established treatments
- Refinement of phase-specific interventions

Ethical Considerations

The use of AI tools and computational modeling in psychotherapy raises important ethical questions regarding privacy, agency, and the nature of therapeutic relationship that require careful consideration.

Conclusion

Active Inference Therapy represents an ambitious attempt to ground psychotherapeutic practice in first principles derived from computational neuroscience. By conceptualizing therapy as a process of collaborative model updating within a coupled dynamical system, AIT offers both theoretical coherence and practical flexibility. While remaining respectful of established therapeutic traditions, it provides a metaframework capable of integrating diverse approaches under a unified mathematical formalism. Future research will determine whether this theoretical promise translates into enhanced clinical outcomes.

References

Carhart-Harris, R. L., & Friston, K. J. (2019). REBUS and the anarchic brain: Toward a unified model of the brain action of psychedelics. *Pharmacological Reviews*, 71(3), 316-344.

Friston, K. (2010). The free-energy principle: A unified brain theory? *Nature Reviews Neuroscience*, 11(2), 127-138.

Friston, K. J., Stephan, K. E., Montague, R., & Dolan, R. J. (2014). Computational psychiatry: The brain as a phantastic organ. *The Lancet Psychiatry*, 1(2), 148-158.

Holmes, J., & Nolte, T. (2019). "Surprise" and the Bayesian brain: Implications for psychotherapy theory and practice. *Frontiers in Psychology*, 10, 592.

[Additional references would continue in standard academic format...]

---

*Corresponding Author*: Grant H. Brenner, MD, DFAPA
*Email*: [contact information]
*Conflicts of Interest*: None declared
*Funding*: This theoretical work received no specific funding

Organized by Claude Opus 4 from auther draft

Selected References: Active Inference and Closely Related Fields

I. Core Active Inference Theory

Foundational Papers

Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138.
Friston, K. J., Stephan, K. E., Montague, R., & Dolan, R. J. (2014). Computational psychiatry: the brain as a phantastic organ. The Lancet Psychiatry, 1(2), 148–158.
Parr, T., Pezzulo, G., & Friston, K. J. (2022). Active Inference: The Free Energy Principle in Mind, Brain, and Behavior. MIT Press.
Kirchhoff, M., Parr, T., Palacios, E., Friston, K., & Kiverstein, J. (2018). The Markov blankets of life: autonomy, active inference and the free energy principle. Journal of The Royal Society Interface, 15(138), 20170792.

Active Inference and Cognition

Parr, T., & Friston, K. J. (2017). Working memory, attention, and salience in active inference. Scientific Reports, 7(1), 1–21.
Seth, A. K., & Friston, K. J. (2016). Active interoceptive inference and the emotional brain. Philosophical Transactions of the Royal Society B, 371(1708), 20160007.
Allen, M., & Friston, K. J. (2018). From cognitivism to autopoiesis: towards a computational framework for the embodied mind. Synthese, 195(6), 2459–2482.

II. Active Inference in Psychotherapy

Direct Applications

Holmes, J. (2024). Friston, Free Energy, and Psychoanalytic Psychotherapy. Entropy, 26(4), 343.
Holmes, J. (2022). Friston’s free energy principle: new life for psychoanalysis? BJPsych Bulletin, 46(3), 164–168.
Holmes, J., & Nolte, T. (2019). “Surprise” and the Bayesian brain: Implications for psychotherapy theory and practice. Frontiers in Psychology, 10, 592.
Smith, R., Moutoussis, M., & Bilek, E. (2021). Simulating the computational mechanisms of cognitive and behavioral psychotherapeutic interventions: insights from active inference. Scientific Reports.

Computational Approaches

Hopkins, J. (2016). Free energy and virtual reality in neuroscience and psychoanalysis: A complexity theory of dreaming and mental disorder. Frontiers in Psychology, 7, 922.
Hopkins, J. (2015). Psychoanalysis, representation and neuroscience: The Freudian unconscious and the Bayesian brain. In From the couch to the lab.
Connolly, P., & van Deventer, V. (2017). Hierarchical recursive organization and the free energy principle: From biological self-organization to the psychoanalytic mind. Frontiers in Psychology, 8, 1695.

III. Active Inference and Touch/Manual Therapy

McParlin, Z., Cerritelli, F., Rossettini, G., Friston, K. J., & Esteves, J. E. (2022). Therapeutic Alliance as Active Inference: The Role of Therapeutic Touch and Biobehavioural Synchrony in Musculoskeletal Care. Frontiers in Behavioral Neuroscience, 16, 897247.
Kim, J., Esteves, J. E., Cerritelli, F., & Friston, K. (2022). An Active Inference Account of Touch and Verbal Communication in Therapy. Frontiers in Psychology, 13, 828952.
Esteves, J. E., Cerritelli, F., Kim, J., & Friston, K. J. (2022). Osteopathic care as (En)active inference: A theoretical framework for developing an integrative hypothesis in osteopathy. Frontiers in Psychology, 13, 812926.
Bohlen, L., Shaw, R., Cerritelli, F., & Esteves, J. E. (2021). Osteopathy and mental health: An embodied, predictive, and interoceptive framework. Frontiers in Psychology, 12, 767005.

IV. Active Inference and Psychopathology

Interoception and Active Inference

Paulus, M. P., Feinstein, J. S., & Khalsa, S. S. (2019). An Active Inference Approach to Interoceptive Psychopathology. Annual Review of Clinical Psychology, 15, 97–122.
Pezzulo, G., Maisto, D., Barca, L., & Van den Bergh, O. (2019). Symptom perception from a predictive processing perspective. Clinical Psychology in Europe, 1(4), 1–14.
Fotopoulou, A., & Tsakiris, M. (2017). Mentalizing homeostasis: The social origins of interoceptive inference. Neuropsychoanalysis, 19(1), 3–28.

Specific Disorders

Fabry, R. E. (2020). Into the dark room: A predictive processing account of major depressive disorder. Phenomenology and the Cognitive Sciences, 19(4), 685–704.
Linson, A., Parr, T., & Friston, K. J. (2020). Active inference, stressors, and psychological trauma: A neuroethological model of (mal)adaptive explore-exploit dynamics in ecological context. Behavioural Brain Research, 380, 112421.
Chekroud, A. M. (2015). Unifying treatments for depression: an application of the free energy principle. Frontiers in Psychology, 6, 153.

V. Consciousness and Psychedelics

Carhart-Harris, R. L., & Friston, K. J. (2019). REBUS and the anarchic brain: toward a unified model of the brain action of psychedelics. Pharmacological Reviews, 71(3), 316–344.
Carhart-Harris, R. L. (2018). The entropic brain — revisited. Neuropharmacology, 142, 167–178.
Solms, M. (2019). The hard problem of consciousness and the free energy principle. Frontiers in Psychology, 9, 2714.

VI. Social Active Inference

Vasil, J., Badcock, P. B., Constant, A., Friston, K., & Ramstead, M. J. (2020). A world unto itself: Human communication as active inference. Frontiers in Psychology, 11, 417.
Veissière, S. P., Constant, A., Ramstead, M. J., Friston, K. J., & Kirmayer, L. J. (2020). Thinking through other minds: A variational approach to cognition and culture. Behavioral and Brain Sciences, 43, e90.
Constant, A., Ramstead, M. J., Veissière, S. P., & Friston, K. (2019). Regimes of expectations: An active inference model of social conformity and human decision making. Frontiers in Psychology, 10, 679.

VII. Hierarchical Processing and Mind

Badcock, P. B., Friston, K. J., & Ramstead, M. J. (2019). The hierarchically mechanistic mind: A free-energy formulation of the human psyche. Physics of Life Reviews, 31, 104–121.
Ramstead, M. J., Kirchhoff, M. D., & Friston, K. J. (2020). A tale of two densities: Active inference is enactive inference. Adaptive Behavior, 28(4), 225–239.

VIII. Predictive Processing Foundations

Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36(3), 181–204.
Bruineberg, J., Kiverstein, J., & Rietveld, E. (2018). The anticipating brain is not a scientist: The free-energy principle from an ecological-enactive perspective. Synthese, 195(6), 2417–2444.

IX. Recent Developments

Biddell, H., Solms, M., Slagter, H., & Laukkonen, R. (2024). Arousal coherence, uncertainty, and well-being: an active inference account. Neuroscience of Consciousness, 2024(1), niae011.
Wilkinson, S., Deane, G., Suarez, K., & Rushton, J. A. (2019). Active inference and epistemic value in psychiatry. In Computational Psychiatry (pp. 27–52). Academic Press.

X. Related Computational Approaches

Miller, M., & Clark, A. (2018). Happily entangled: Prediction, emotion, and the embodied mind. Synthese, 195(6), 2559–2575.
Hutchinson, J. B., & Barrett, L. F. (2019). The power of predictions: an emerging paradigm for psychological research. Current Directions in Psychological Science, 28(3), 280–291.