From Broken Models to Treatment Selection: Active Inference as a Tool to Guide Clinical Research and Practice Editorial From Broken Models to Treatment Selection: Active Inference as a Tool to Guide Clinical Research and Practice Lukas Kirchner 1, Anna-Lena Eckert 2, Max Berg 1 [1] Department of Psychology, Clinical Psychology and Psychotherapy, Philipps-University of Marburg, Marburg, Germany. [2] Department of Psychology, Theoretical Cognitive Science, Philipps-University of Marburg, Marburg, Germany. Clinical Psychology in Europe, 2022, Vol. 4(2), Article e9697, https://doi.org/10.32872/cpe.9697 Published (VoR): 2022-06-30 Corresponding Author: Lukas Kirchner, Department of Psychology, Clinical Psychology and Psychotherapy, Philipps-University of Marburg, Gutenbergstraße 18, 35037 Marburg, Germany. E-mail: lukas.kirchner@uni- marburg.de Computational theories have fundamentally changed the scientific understanding of how the mind works for both healthy and pathological experiences and behaviours. In this context, the active inference framework has gained considerable attention within the scientific community (Heins et al., 2022; Smith et al., 2022). As a process theory, it integrates complex phenomena, such as perception, learning, and action under a unified theory of Bayesian inference (Da Costa et al., 2020; Friston et al., 2017). Active inference has proven useful in modelling data from heterogeneous fields ranging from cognitive neuroscience to biology and general psychology (e.g., Friston et al., 2016, 2017). Its com­ putational tractability and biological plausibility have also made it increasingly relevant to clinical psychology in recent years (e.g., Smith, Badcock, et al., 2021). In active inference and related, Bayesian neurocomputational theories, it is assumed that individuals do not have direct access to the circumstances in their surroundings. Instead, they have to infer the (probabilistic) properties of their environment through action and perception by integrating prior information about their environment with ambiguous sensory input in a rational (i.e., Bayes-optimal) manner (Friston et al., 2016; Hohwy et al., 2008). The resulting “internal model of the world” (i.e., the agent’s beliefs about how certain sensory information relates to environmental conditions) shapes future perception (Friston, 2010) and enables agents to leverage the past to predict the future in an ever-changing environment (Badcock et al., 2017). In accordance with this perspective, perception, action, and learning are all subject to inferential process­ This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 International License, CC BY 4.0, which permits unrestricted use, distribution, and reproduction, provided the original work is properly cited. https://crossmark.crossref.org/dialog/?doi=10.32872/cpe.9697&domain=pdf&date_stamp=2022-06-30 https://cpe.psychopen.eu/ https://www.psychopen.eu/ https://creativecommons.org/licenses/by/4.0/ https://creativecommons.org/licenses/by/4.0/ https://creativecommons.org/licenses/by/4.0/ es on different timescales (Smith et al., 2022). Individuals thus take an active role in constructing their experiences, which they can further alter by actively changing their environment. Whereas accurate internal models provide good predictions about which action se­ quences lead to preferred sensory observations, distorted internal models can lead to aberrant experiences and behaviours that may hinder the organism from achieving its goals (Badcock et al., 2017; Schwartenbeck et al., 2015). There is evidence that the internal models of individuals with mental disorders show substantial deviations from each other and from the models of healthy individuals. For example, recent accounts of depression have conceptualised patients' tendency to reappraise or disregard posi­ tive information in terms of highly precise and hence tenacious negative prior beliefs (Kube et al., 2020). This computational perspective has inspired novel ideas for the treat­ ment of depression (e.g., Chekroud, 2015). Similar reconceptualisations with relevance for psychological treatments have been suggested for numerous mental disorders and health conditions, including psychosis (Sterzer et al., 2018), persistent somatic symptoms (Paulus et al., 2019), and eating disorders (Barca & Pezzulo, 2020). In this context, the active inference framework offers the opportunity to formalise deviations in a person’s internal models, thus enabling a detailed description and op­ erationalisation of relevant experiential and behavioural distortions (e.g., Montague et al., 2012). From our point of view, this could improve clinical research, diagnostics, and the treatment of mental disorders in several ways. Clinical research may benefit from a finely grained formalisation of deviant experiences and behaviours within the active inference framework because it opens up a possibility for studying aetiological mechanisms from a computational perspective (Stephan, Binder, et al., 2016). Because of their generative structure, computational theories enable the derivation of well-opera­ tionalised hypotheses about pathological processes in mental disorders and the computer simulation of aberrant experience and behaviour. In comparison with empirical data, researchers could thus rigorously formalise, simulate, and compare different mechanis­ tic models of patients’ experiential and behavioural symptoms. Moreover, the active inference perspective could inform the diagnostics of mental disorders (or rather the diagnostics of patients’ implicit belief systems) by providing practitioners with individual estimates of their patients’ internal model parameters in disorder-relevant situations (Stephan, Bach, et al., 2016). For example, using probabilistic gambling tasks that dis­ tinguish between goal-directed information seeking and random exploration behaviour could provide clinicians with individual parameter diagnostics regarding the relationship between information seeking, reward sensitivity, and psychopathology in substance abusers (Smith, Schwartenbeck, et al., 2021). Such applications could not only strengthen a more transdiagnostic perspective on mental disorders, but also have tangible implica­ tions for treatment development and treatment selection. If we assume that therapeutic interventions may have different effects on patients' model parameters, finely grained Active Inference – A Tool to Guide Clinical Research and Practice 2 Clinical Psychology in Europe 2022, Vol. 4(2), Article e9697 https://doi.org/10.32872/cpe.9697 https://www.psychopen.eu/ operationalisation in the context of active inference could contribute to tailored interven­ tions that target specifically these parameters. From the practitioner's perspective, it would be particularly important to investigate which tasks are likely to diagnose internal models that inform treatment selection and guide psychotherapy. The active inference approach affords an improved mechanistic understanding of pathological processes in mental disorders. As a unifying process theory of brain and mind function, it brings together perception, learning, action, and decision making un­ der the umbrella of a Bayesian principle, which predestines it for clinical application. Because of its high degree of formalisation and its flexibility, we believe that the active inference approach is well suited to functionally link heterogeneous clinical phenomena to patients’ internal belief systems. This will enable researchers to better differentiate and operationalise underlying mechanisms and tailor the diagnosis, aetiology, and treat­ ment of mental disorders. Funding: This work was funded by the Hessian Ministry of Higher Education, Research, Science, and the Arts. Acknowledgments: The authors have no additional (i.e., non-financial) support to report. Competing Interests: This work was realised within the cluster initiative “The Adaptive Mind” (TAM) which brings together scientists from experimental psychology, clinical psychology, and artificial intelligence to improve the understanding of how the human mind successfully adapts to changing conditions. TAM is funded by the Hessian Ministry of Higher Education, Research, Science, and the Arts. R e f e r e n c e s Badcock, P. B., Davey, C. G., Whittle, S., Allen, N. B., & Friston, K. J. (2017). The depressed brain: An evolutionary systems theory. Trends in Cognitive Sciences, 21(3), 182–194. https://doi.org/10.1016/j.tics.2017.01.005 Barca, L., & Pezzulo, G. (2020). Keep your interoceptive streams under control: An active inference perspective on anorexia nervosa. Cognitive, Affective & Behavioral Neuroscience, 20(2), 427–440. https://doi.org/10.3758/s13415-020-00777-6 Chekroud, A. M. (2015). Unifying treatments for depression: An application of the Free Energy Principle. Frontiers in Psychology, 6, Article 153. https://doi.org/10.3389/fpsyg.2015.00153 Da Costa, L., Parr, T., Sajid, N., Veselic, S., Neacsu, V., & Friston, K. (2020). Active inference on discrete state-spaces: A synthesis. Journal of Mathematical Psychology, 99, Article 102447. https://doi.org/10.1016/j.jmp.2020.102447 Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138. https://doi.org/10.1038/nrn2787 Kirchner, Eckert, & Berg 3 Clinical Psychology in Europe 2022, Vol. 4(2), Article e9697 https://doi.org/10.32872/cpe.9697 https://doi.org/10.1016/j.tics.2017.01.005 https://doi.org/10.3758/s13415-020-00777-6 https://doi.org/10.3389/fpsyg.2015.00153 https://doi.org/10.1016/j.jmp.2020.102447 https://doi.org/10.1038/nrn2787 https://www.psychopen.eu/ Friston, K., FitzGerald, T., Rigoli, F., Schwartenbeck, P., O’Doherty, J., & Pezzulo, G. (2016). Active inference and learning. Neuroscience and Biobehavioral Reviews, 68, 862–879. https://doi.org/10.1016/j.neubiorev.2016.06.022 Friston, K., FitzGerald, T., Rigoli, F., Schwartenbeck, P., & Pezzulo, G. (2017). Active inference: A process theory. Neural Computation, 29(1), 1–49. https://doi.org/10.1162/NECO_a_00912 Heins, C., Millidge, B., Demekas, D., Klein, B., Friston, K., Couzin, I. D., & Tschantz, A. (2022). pymdp: A Python library for active inference in discrete state spaces. Journal of Open Source Software, 7(73), Article 4098. https://doi.org/10.21105/joss.04098 Hohwy, J., Roepstorff, A., & Friston, K. (2008). Predictive coding explains binocular rivalry: An epistemological review. Cognition, 108(3), 687–701. https://doi.org/10.1016/j.cognition.2008.05.010 Kube, T., Schwarting, R., Rozenkrantz, L., Glombiewski, J. A., & Rief, W. (2020). Distorted cognitive processes in major depression: A predictive processing perspective. Biological Psychiatry, 87(5), 388–398. https://doi.org/10.1016/j.biopsych.2019.07.017 Montague, P. R., Dolan, R. J., Friston, K. J., & Dayan, P. (2012). Computational psychiatry. Trends in Cognitive Sciences, 16(1), 72–80. https://doi.org/10.1016/j.tics.2011.11.018 Paulus, M. P., Feinstein, J. S., & Khalsa, S. S. (2019). An active inference approach to interoceptive psychopathology. Annual Review of Clinical Psychology, 15(1), 97–122. https://doi.org/10.1146/annurev-clinpsy-050718-095617 Schwartenbeck, P., FitzGerald, T. H. B., Mathys, C., Dolan, R., Wurst, F., Kronbichler, M., & Friston, K. (2015). Optimal inference with suboptimal models: Addiction and active Bayesian inference. Medical Hypotheses, 84(2), 109–117. https://doi.org/10.1016/j.mehy.2014.12.007 Smith, R., Badcock, P., & Friston, K. J. (2021). Recent advances in the application of predictive coding and active inference models within clinical neuroscience. Psychiatry and Clinical Neurosciences, 75(1), 3–13. https://doi.org/10.1111/pcn.13138 Smith, R., Friston, K. J., & Whyte, C. J. (2022). A step-by-step tutorial on active inference and its application to empirical data. Journal of Mathematical Psychology, 107, Article 102632. https://doi.org/10.1016/j.jmp.2021.102632 Smith, R., Schwartenbeck, P., Stewart, J., Kulplicki, R., Ekhtiari, H., & Paulus, M. (2021). Reduced action precision and biased learning in substance users when solving the explore-exploit dilemma: An active inference modelling approach. Biological Psychiatry, 89(9), S55–S56. https://doi.org/10.1016/j.biopsych.2021.02.155 Stephan, K. E., Bach, D. R., Fletcher, P. C., Flint, J., Frank, M. J., Friston, K. J., Heinz, A., Huys, Q. J. M., Owen, M. J., Binder, E. B., Dayan, P., Johnstone, E. C., Meyer-Lindenberg, A., Montague, P. R., Schnyder, U., Wang, X.-J., & Breakspear, M. (2016). Charting the landscape of priority problems in psychiatry, part 1: Classification and diagnosis. The Lancet Psychiatry, 3(1), 77–83. https://doi.org/10.1016/S2215-0366(15)00361-2 Stephan, K. E., Binder, E. B., Breakspear, M., Dayan, P., Johnstone, E. C., Meyer-Lindenberg, A., Schnyder, U., Wang, X.-J., Bach, D. R., Fletcher, P. C., Flint, J., Frank, M. J., Heinz, A., Huys, Q. J. M., Montague, P. R., Owen, M. J., & Friston, K. J. (2016). Charting the landscape of priority Active Inference – A Tool to Guide Clinical Research and Practice 4 Clinical Psychology in Europe 2022, Vol. 4(2), Article e9697 https://doi.org/10.32872/cpe.9697 https://doi.org/10.1016/j.neubiorev.2016.06.022 https://doi.org/10.1162/NECO_a_00912 https://doi.org/10.21105/joss.04098 https://doi.org/10.1016/j.cognition.2008.05.010 https://doi.org/10.1016/j.biopsych.2019.07.017 https://doi.org/10.1016/j.tics.2011.11.018 https://doi.org/10.1146/annurev-clinpsy-050718-095617 https://doi.org/10.1016/j.mehy.2014.12.007 https://doi.org/10.1111/pcn.13138 https://doi.org/10.1016/j.jmp.2021.102632 https://doi.org/10.1016/j.biopsych.2021.02.155 https://doi.org/10.1016/S2215-0366(15)00361-2 https://www.psychopen.eu/ problems in psychiatry, part 2: Pathogenesis and aetiology. The Lancet Psychiatry, 3(1), 84–90. https://doi.org/10.1016/S2215-0366(15)00360-0 Sterzer, P., Adams, R. A., Fletcher, P., Frith, C., Lawrie, S. M., Muckli, L., Petrovic, P., Uhlhaas, P., Voss, M., & Corlett, P. R. (2018). The predictive coding account of psychosis. Biological Psychiatry, 84(9), 634–643. https://doi.org/10.1016/j.biopsych.2018.05.015 Clinical Psychology in Europe (CPE) is the official journal of the European Association of Clinical Psychology and Psychological Treatment (EACLIPT). PsychOpen GOLD is a publishing service by Leibniz Institute for Psychology (ZPID), Germany. Kirchner, Eckert, & Berg 5 Clinical Psychology in Europe 2022, Vol. 4(2), Article e9697 https://doi.org/10.32872/cpe.9697 https://doi.org/10.1016/S2215-0366(15)00360-0 https://doi.org/10.1016/j.biopsych.2018.05.015 https://www.psychopen.eu/