Computation in Psychotherapy, or How Computational Psychiatry Can Aid Learning-Based Psychological Therapies

Using computational models of learning to advance cognitive behavioral therapy

Many psychotherapy interventions have a large evidence base and can help a substantial number of people with symptoms of mental health conditions. However, we still have little understanding of why treatments work. Early advances in psychotherapy, such as the development of exposure therapy, built on theoretical and experimental evidence from Pavlovian and instrumental conditioning. More generally, all psychotherapy achieves change through learning. The past 25 years have seen substantial developments in computational models of learning, with increased computational precision and a focus on multiple learning mechanisms and their interaction. Now might be a good time to formalize psychotherapy interventions as computational models of learning to improve our understanding of mechanisms of change in psychotherapy. To advance research and help bring together a new joint field of theory-driven computational psychotherapy, we first review literature on cognitive behavioral therapy (exposure therapy and cognitive restructuring) and introduce computational models of reinforcement learning and representation learning. We then suggest a mapping of these learning algorithms on change processes presumably underlying the effects of exposure therapy and cognitive restructuring. Finally, we outline how the understanding of interventions through the lens of learning algorithms can inform intervention research.

Learning training as a cognitive restructuring intervention

BACKGROUND

A core part of cognitive therapy for low mood is learning to identify and challenge negative beliefs. However, it is currently unclear whether improved ability to recognise such beliefs, and the biased interpretations of events which may maintain them, is a mechanism of symptom change during treatment.

METHODS

We investigated the effects of completing a learning task (training to identify and select self-enhancing interpretations of events) and a brief cognitive restructuring intervention (how exploring alternative explanations of events may result in improved mood) on causal attribution tendencies. Studies were conducted online using randomized-controlled experimental designs (N=200 &N=164), and data were analysed using hierarchical Bayesian models.

RESULTS

We found that both learning training and the restructuring intervention decreased tendencies to make unhelpful attributions and increased tendencies to make self-enhancing attributions. Across two studies, changes in attribution tendencies were associated with higher learning rates during learning training, an effect specific to learning about different kinds of event attribution. Contrary to expectation, we found no evidence that faster learning was associated specifically to changes in attribution tendencies following cognitive restructuring. Since participants with higher learning rate estimates also provided explicit ratings and free-text descriptions of event causes which were closer to the ground truth, we interpret this as representing a greater benefit of learning training in individuals who were better able to understand the task state space.

CONCLUSIONS

We suggest that personalized training, in conjunction with feedback based on interpretable computational model output, may provide a useful form of augmentation or learning support tool during therapy.

Computational Psychotherapy System for Mental Health Prediction and Behavior Change with a Conversational Agent

Background

The importance of computational psychotherapy is increasing due to the record high prevalence of mental health issues worldwide. Despite advancements, current computational psychotherapy systems lack advanced prediction and behavior change mechanisms using conversational agents.

Purpose

This work presents a computational psychotherapy system for mental health prediction and behavior change using a conversational agent. It makes two major contributions. First, we introduce a novel, golden standard dataset, comprising panel data with 1495 instances of quantitative stress, anxiety, and depression (SAD) symptom scores from diagnostic-level questionnaires and qualitative daily diary entries. Second, we present the computational psychotherapy system itself.

Hypothesis

We hypothesize that simulating a theory of mind - the human cognitive ability to understand others - in a conversational agent enhances its effectiveness in relieving mental health issues.

Methods

The system simulates theory of mind with a cognitive architecture comprising an ensemble of computational models, using cognitive modelling and machine learning models trained on the novel dataset, and novel ontologies. The system was evaluated through a computational experiment on mental health phenomena prediction from text, and an empirical interventional study on relieving mental health issues in 42 participants.

Results

The system outperformed state-of-the-art systems in terms of the number of detected categories and detection accuracy (highest accuracy: 91.41% using k-nearest neighbors (kNN); highest accuracy of other systems: 84% using long-short term memory network (LSTM)). The highest accuracy for 7-day forecasting was 87.68%, whereas the other systems were not able to forecast trends. In the study, the system outperformed Woebot, the current state-of-the-art, in reducing stress (p = 0.004) and anxiety (p = 0.008) levels.

Conclusion

The confirmation of our hypothesis indicates that incorporating theory of mind simulation in conversational agents significantly enhances their efficacy in computational psychotherapy, offering a promising advancement for mental health interventions and support compared to current state-of-the-art systems.

The neuroscience of mental illness: Building toward the future

Mental illnesses arise from dysfunction in the brain. Although numerous extraneural factors influence these illnesses, ultimately, it is the science of the brain that will lead to novel therapies. Meanwhile, our understanding of this complex organ is incomplete, leading to the oft-repeated trope that neuroscience has yet to make significant contributions to the care of individuals with mental illnesses. This review seeks to counter this narrative, using specific examples of how neuroscientific advances have contributed to progress in mental health care in the past and how current achievements set the stage for further progress in the future.

Lack of optimistic bias during social evaluation learning reflects reduced positive self-beliefs in depression and social anxiety, but via distinct mechanisms

Processing social feedback optimistically may maintain positive self-beliefs and stable social relationships. Conversely, a lack of this optimistic bias in depression and social anxiety may perpetuate negative self-beliefs and maintain symptoms. Research investigating this mechanism is scarce, however, and the mechanisms by which depressed and socially anxious individuals respond to social evaluation may also differ. Using a range of computational approaches in two large datasets (mega-analysis of previous studies, n = 450; pre-registered replication study, n = 807), we investigated how depression (PHQ-9) and social anxiety (BFNE) symptoms related to social evaluation learning in a computerized task. Optimistic bias (better learning of positive relative to negative evaluations) was found to be negatively associated with depression and social anxiety. Structural equation models suggested this reflected a heightened sensitivity to negative social feedback in social anxiety, whereas in depression it co-existed with a blunted response to positive social feedback. Computational belief-based learning models further suggested that reduced optimism was driven by less positive trait-like self-beliefs in both depression and social anxiety, with some evidence for a general blunting in belief updating in depression. Recognizing such transdiagnostic similarities and differences in social evaluation learning across disorders may inform approaches to personalizing treatment.

Understanding Cognitive Behavioral Therapy for Psychosis Through the Predictive Coding Framework

Psychological treatments for persecutory delusions, particularly cognitive behavioral therapy for psychosis, are efficacious; however, mechanistic theories explaining why they work rarely bridge to the level of cognitive neuroscience. Predictive coding, a general brain processing theory rooted in cognitive and computational neuroscience, has increasing experimental support for explaining symptoms of psychosis, including the formation and maintenance of delusions. Here, we describe recent advances in cognitive behavioral therapy for psychosis-based psychotherapy for persecutory delusions, which targets specific psychological processes at the computational level of information processing. We outline how Bayesian learning models employed in predictive coding are superior to simple associative learning models for understanding the impact of cognitive behavioral interventions at the algorithmic level. We review hierarchical predictive coding as an account of belief updating rooted in prediction error signaling. We examine how this process is abnormal in psychotic disorders, garnering noisy sensory data that is made sense of through the development of overly strong delusional priors. We argue that effective cognitive behavioral therapy for psychosis systematically targets the way sensory data are selected, experienced, and interpreted, thus allowing for the strengthening of alternative beliefs. Finally, future directions based on these arguments are discussed.

Pain perception as hierarchical Bayesian inference: A test case for the theory of constructed emotion

An intriguing perspective about human emotion, the theory of constructed emotion considers emotions as generative models according to the Bayesian brain hypothesis. This theory brings fresh insight to existing findings, but its complexity renders it challenging to test experimentally. We argue that laboratory studies of pain could support the theory because although some may not consider pain to be a genuine emotion, the theory must at minimum be able to explain pain perception and its dysfunction in pathology. We review emerging evidence that bear on this question. We cover behavioral and neural laboratory findings, computational models, placebo hyperalgesia, and chronic pain. We conclude that there is substantial evidence for a predictive processing account of painful experience, paving the way for a better understanding of neuronal and computational mechanisms of other emotions.

Different components of cognitive-behavioral therapy affect specific cognitive mechanisms

Psychological therapies are among the most effective treatments for common mental health problems-however, we still know relatively little about how exactly they improve symptoms. Here, we demonstrate the power of combining theory with computational methods to parse effects of different components of cognitive-behavioral therapies onto underlying mechanisms. Specifically, we present data from a series of randomized-controlled experiments testing the effects of brief components of behavioral and cognitive therapies on different cognitive processes, using well-validated behavioral measures and associated computational models. A goal setting intervention, based on behavioral activation therapy activities, reliably and selectively reduced sensitivity to effort when deciding how to act to gain reward. By contrast, a cognitive restructuring intervention, based on cognitive therapy materials, reliably and selectively reduced the tendency to attribute negative everyday events to self-related causes. The effects of each intervention were specific to these respective measures. Our approach provides a basis for beginning to understand how different elements of common psychotherapy programs may work.

Neural Response to Expecting a Caloric Sweet Taste Stimulus Predicts Body Mass Index Longitudinally Among Young Adult Women With Anorexia Nervosa

BACKGROUND

Anorexia nervosa (AN) is an often-chronic illness, and we lack biomarkers to predict long-term outcome. Recent neuroimaging studies using caloric taste stimuli suggest that paradigms that have tested conditioned neural responses to expectation or salient stimulus receipt may underpin behaviors. However, whether activation of those neural circuits can predict long-term outcome has not been studied.

METHODS

We followed women treated for AN (n = 35, mean age [SD] = 23 [7] years) and tested whether functional imaging brain response during a taste conditioning paradigm could predict posttreatment body mass index (BMI). We anticipated greater neural activity relative to caloric stimulus expectation and that dopamine-related receipt conditions would predict lower posttreatment BMI, indicating fear-associated arousal.

RESULTS

Follow-up occurred at mean (SD) = 1648 (1216) days after imaging. Stimulus expectation in orbitofrontal and striatal regions and BMI and BMI change at follow-up were negatively correlated, and these correlations remained significant for the right superior orbitofrontal cortex and BMI change after multiple comparison correction (r = -0.484, p = .003). This relationship remained significant after including time between brain scanning and follow-up in the model. Reward prediction error response did not predict long-term BMI.

CONCLUSIONS

The orbitofrontal cortex is involved in learning and conditioning, and these data implicate this region in learned caloric stimulus expectation and long-term prediction of weight outcomes in AN. Thus, conditioned elevated brain response to the anticipation of receiving a caloric stimulus may drive food avoidance, suggesting that breaking such associations is central for long-term recovery from AN.

Evaluating Change in Transference, Interpersonal Functioning, and Trust Processes in the Treatment of Borderline Personality Disorder: A Single-Case Study Using Ecological Momentary Assessment

Transference-focused psychotherapy (TFP) is an empirically supported treatment for borderline personality disorder (BPD) that improves functioning via targeting representations of self affectively relating to others, particularly as evoked in the therapeutic relationship. If change in TFP operates as theorized, then shifts in patterns of "self affectively relating to others" should be observed in the transference prior to shifts in daily relationships. Using ecological momentary assessment (EMA), a patient with BPD rated daily interpersonal events for 2-week periods during 18 months of TFP; at 9 and 18 months these ratings included interactions with the therapist. Results suggest that positive perceptions of her therapist that ran counter to her negatively biased perception in other relationships preceded changes in her perceptions of others. EMA shifts corresponded to improvements in self-reported symptoms, interview-based personality functioning, and therapist assessments. Implications for assimilation of a trusting experience with the therapist as a mechanism of change in TFP are discussed.

Changes in Learning From Social Feedback After Web-Based Interpretation Bias Modification: Secondary Analysis of a Digital Mental Health Intervention Among Individuals With High Social Anxiety Symptoms

BACKGROUND

Biases in social reinforcement learning, or the process of learning to predict and optimize behavior based on rewards and punishments in the social environment, may underlie and maintain some negative cognitive biases that are characteristic of social anxiety. However, little is known about how cognitive and behavioral interventions may change social reinforcement learning in individuals who are anxious.

OBJECTIVE

This study assessed whether a scalable, web-based cognitive bias modification for interpretations (CBM-I) intervention changed social reinforcement learning biases in participants with high social anxiety symptoms. This study focused on 2 types of social reinforcement learning relevant to social anxiety: learning about other people and learning about one's own social performance.

METHODS

Participants (N=106) completed 2 laboratory sessions, separated by 5 weeks of ecological momentary assessment tracking emotion regulation strategy use and affect. Approximately half (n=51, 48.1%) of the participants completed up to 6 brief daily sessions of CBM-I in week 3. Participants completed a task that assessed social reinforcement learning about other people in both laboratory sessions and a task that assessed social reinforcement learning about one's own social performance in the second session. Behavioral data from these tasks were computationally modeled using Q-learning and analyzed using mixed effects models.

RESULTS

After the CBM-I intervention, participants updated their beliefs about others more slowly (P=.04; Cohen d=-0.29) but used what they learned to make more accurate decisions (P=.005; Cohen d=0.20), choosing rewarding faces more frequently. These effects were not observed among participants who did not complete the CBM-I intervention. Participants who completed the CBM-I intervention also showed less-biased updating about their social performance than participants who did not complete the CBM-I intervention, learning similarly from positive and negative feedback and from feedback on items related to poor versus good social performance. Regardless of the intervention condition, participants at session 2 versus session 1 updated their expectancies about others more from rewarding (P=.003; Cohen d=0.43) and less from punishing outcomes (P=.001; Cohen d=-0.47), and they became more accurate at learning to avoid punishing faces (P=.001; Cohen d=0.20).

CONCLUSIONS

Taken together, our results provide initial evidence that there may be some beneficial effects of both the CBM-I intervention and self-tracking of emotion regulation on social reinforcement learning in individuals who are socially anxious, although replication will be important.

Belief Updating and Paranoia in Individuals With Schizophrenia

BACKGROUND

Persecutory delusions are among the most common delusions in schizophrenia and represent the extreme end of the paranoia continuum. Paranoia is accompanied by significant worry and distress. Identifying cognitive mechanisms underlying paranoia is critical for advancing treatment. We hypothesized that aberrant belief updating, which is related to paranoia in human and animal models, would also contribute to persecutory beliefs in individuals with schizophrenia.

METHODS

Belief updating was assessed in 42 participants with schizophrenia and 44 healthy control participants using a 3-option probabilistic reversal learning task. Hierarchical Gaussian Filter was used to estimate computational parameters of belief updating. Paranoia was measured using the Positive and Negative Syndrome Scale and the revised Green et al. Paranoid Thoughts Scale. Unusual thought content was measured with the Psychosis Symptom Rating Scale and the Peters et al. Delusions Inventory. Worry was measured using the Dunn Worry Questionnaire.

RESULTS

Paranoia was significantly associated with elevated win-switch rate and prior beliefs about volatility both in schizophrenia and across the whole sample. These relationships were specific to paranoia and did not extend to unusual thought content or measures of anxiety. We observed a significant indirect effect of paranoia on the relationship between prior beliefs about volatility and worry.

CONCLUSIONS

This work provides evidence that relationships between belief updating parameters and paranoia extend to schizophrenia, may be specific to persecutory beliefs, and contribute to theoretical models implicating worry in the maintenance of persecutory delusions.

Computational Methods in Psychotherapy: A Scoping Review

BACKGROUND

The study of complex systems, such as the psychotherapeutic encounter, transcends the mechanistic and reductionist methods for describing linear processes and needs suitable approaches to describe probabilistic and scarcely predictable phenomena.

OBJECTIVE

The present study undertakes a scoping review of research on the computational methods in psychotherapy to gather new developments in this field and to better understand the phenomena occurring in psychotherapeutic interactions as well as in human interaction more generally.

DESIGN

Online databases were used to identify papers published 2011-2022, from which we selected 18 publications from different resources, selected according to criteria established in advance and described in the text. A flow chart and a summary table of the articles consulted have been created.

RESULTS

The majority of publications (44.4%) reported combined computational and experimental approaches, so we grouped the studies according to the types of computational methods used. All but one of the studies collected measured data. All the studies confirmed the usefulness of predictive and learning models in the study of complex variables such as those belonging to psychological, psychopathological and psychotherapeutic processes.

CONCLUSIONS

Research on computational methods will benefit from a careful selection of reference methods and standards. Therefore, this review represents an attempt to systematise the empirical literature on the applications of computational methods in psychotherapy research in order to offer clinicians an overview of the usefulness of these methods and the possibilities of their use in the various fields of application, highlighting their clinical implications, and ultimately attempting to identify potential opportunities for further research.

Functional neuroimaging in psychiatry and the case for failing better

Psychiatric disorders encompass complex aberrations of cognition and affect and are among the most debilitating and poorly understood of any medical condition. Current treatments rely primarily on interventions that target brain function (drugs) or learning processes (psychotherapy). A mechanistic understanding of how these interventions mediate their therapeutic effects remains elusive. From the early 1990s, non-invasive functional neuroimaging, coupled with parallel developments in the cognitive neurosciences, seemed to signal a new era of neurobiologically grounded diagnosis and treatment in psychiatry. Yet, despite three decades of intense neuroimaging research, we still lack a neurobiological account for any psychiatric condition. Likewise, functional neuroimaging plays no role in clinical decision making. Here, we offer a critical commentary on this impasse and suggest how the field might fare better and deliver impactful neurobiological insights.

Altered predictive control during memory suppression in PTSD

Aberrant predictions of future threat lead to maladaptive avoidance in individuals with post-traumatic stress disorder (PTSD). How this disruption in prediction influences the control of memory states orchestrated by the dorsolateral prefrontal cortex is unknown. We combined computational modeling and brain connectivity analyses to reveal how individuals exposed and nonexposed to the 2015 Paris terrorist attacks formed and controlled beliefs about future intrusive re-experiencing implemented in the laboratory during a memory suppression task. Exposed individuals with PTSD used beliefs excessively to control hippocampal activity during the task. When this predictive control failed, the prediction-error associated with unwanted intrusions was poorly downregulated by reactive mechanisms. This imbalance was linked to higher severity of avoidance symptoms, but not to general disturbances such as anxiety or negative affect. Conversely, trauma-exposed participants without PTSD and nonexposed individuals were able to optimally balance predictive and reactive control during the memory suppression task. These findings highlight a potential pathological mechanism occurring in individuals with PTSD rooted in the relationship between the brain’s predictive and control mechanisms.

It remains unclear how predictions of future threat affect memory recall, specifically in the case of post-traumatic stress disorder (PTSD). Here, the authors combined computational modeling and brain connectivity analyses to show that individuals with PTSD have exaggerated predictive control and reduced reactive control in a memory suppression task.

Self-esteem depends on beliefs about the rate of change of social approval

A major challenge in understanding the neurobiological basis of psychiatric disorders is rigorously quantifying subjective metrics that lie at the core of mental illness, such as low self-esteem. Self-esteem can be conceptualized as a 'gauge of social approval' that increases in response to approval and decreases in response to disapproval. Computational studies have shown that learning signals that represent the difference between received and expected social approval drive changes in self-esteem. However, it is unclear whether self-esteem based on social approval should be understood as a value updated through associative learning, or as a belief about approval, updated by new evidence depending on how strongly it is held. Our results show that belief-based models explain self-esteem dynamics in response to social evaluation better than associative learning models. Importantly, they suggest that in the short term, self-esteem signals the direction and rate of change of one's beliefs about approval within a group, rather than one's social position.

Computational Psychiatry Needs Time and Context

Why has computational psychiatry yet to influence routine clinical practice? One reason may be that it has neglected context and temporal dynamics in the models of certain mental health problems. We develop three heuristics for estimating whether time and context are important to a mental health problem: Is it characterized by a core neurobiological mechanism? Does it follow a straightforward natural trajectory? And is intentional mental content peripheral to the problem? For many problems the answers are no, suggesting that modeling time and context is critical. We review computational psychiatry advances toward this end, including modeling state variation, using domain-specific stimuli, and interpreting differences in context. We discuss complementary network and complex systems approaches. Novel methods and unification with adjacent fields may inspire a new generation of computational psychiatry.

Mental distress through the prism of predictive processing theory

We review the predictive processing theory's take on goals and affect, to shed new light on mental distress and how it develops into psychopathology such as in affective and motivational disorders. This analysis recovers many of the classical factors known to be important in those disorders, like uncertainty and control, but integrates them in a mechanistic model of adaptive and maladaptive cognition and behavior. We derive implications for treatment that have so far remained underexposed in existing predictive processing accounts of mental disorder, specifically with regard to the model-dependent construction of value, the importance of model validation (evidence), and the introduction and learning of new, adaptive beliefs that relieve suffering.

Simulating the computational mechanisms of cognitive and behavioral psychotherapeutic interventions: insights from active inference

Cognitive-behavioral therapy (CBT) leverages interactions between thoughts, feelings, and behaviors. To deepen understanding of these interactions, we present a computational (active inference) model of CBT that allows formal simulations of interactions between cognitive interventions (i.e., cognitive restructuring) and behavioral interventions (i.e., exposure) in producing adaptive behavior change (i.e., reducing maladaptive avoidance behavior). Using spider phobia as a concrete example of maladaptive avoidance more generally, we show simulations indicating that when conscious beliefs about safety/danger have strong interactions with affective/behavioral outcomes, behavioral change during exposure therapy is mediated by changes in these beliefs, preventing generalization. In contrast, when these interactions are weakened, and cognitive restructuring only induces belief uncertainty (as opposed to strong safety beliefs), behavior change leads to generalized learning (i.e., "over-writing" the implicit beliefs about action-outcome mappings that directly produce avoidance). The individual is therefore equipped to face any new context, safe or dangerous, remaining in a content state without the need for avoidance behavior-increasing resilience from a CBT perspective. These results show how the same changes in behavior during CBT can be due to distinct underlying mechanisms; they predict lower rates of relapse when cognitive interventions focus on inducing uncertainty and on reducing the effects of automatic negative thoughts on behavior.

Computational approaches and machine learning for individual-level treatment predictions

RATIONALE

The impact of neuroscience-based approaches for psychiatry on pragmatic clinical decision-making has been limited. Although neuroscience has provided insights into basic mechanisms of neural function, these insights have not improved the ability to generate better assessments, prognoses, diagnoses, or treatment of psychiatric conditions.

OBJECTIVES

To integrate the emerging findings in machine learning and computational psychiatry to address the question: what measures that are not derived from the patient's self-assessment or the assessment by a trained professional can be used to make more precise predictions about the individual's current state, the individual's future disease trajectory, or the probability to respond to a particular intervention?

RESULTS

Currently, the ability to use individual differences to predict differential outcomes is very modest possibly related to the fact that the effect sizes of interventions are small. There is emerging evidence of genetic and neuroimaging-based heterogeneity of psychiatric disorders, which contributes to imprecise predictions. Although the use of machine learning tools to generate clinically actionable predictions is still in its infancy, these approaches may identify subgroups enabling more precise predictions. In addition, computational psychiatry might provide explanatory disease models based on faulty updating of internal values or beliefs.

CONCLUSIONS

There is a need for larger studies, clinical trials using machine learning, or computational psychiatry model parameters predictions as actionable outcomes, comparing alternative explanatory computational models, and using translational approaches that apply similar paradigms and models in humans and animals.

Latent cause inference during extinction learning in trauma-exposed individuals with and without PTSD

BACKGROUND

Problems in learning that sights, sounds, or situations that were once associated with danger have become safe (extinction learning) may explain why some individuals suffer prolonged psychological distress following traumatic experiences. Although simple learning models have been unable to provide a convincing account of why this learning fails, it has recently been proposed that this may be explained by individual differences in beliefs about the causal structure of the environment.

METHODS

Here, we tested two competing hypotheses as to how differences in causal inference might be related to trauma-related psychopathology, using extinction learning data collected from clinically well-characterised individuals with varying degrees of post-traumatic stress (N = 56). Model parameters describing individual differences in causal inference were related to multiple post-traumatic stress disorder (PTSD) and depression symptom dimensions via network analysis.

RESULTS

Individuals with more severe PTSD were more likely to assign observations from conditioning and extinction stages to a single underlying cause. Specifically, greater re-experiencing symptom severity was associated with a lower likelihood of inferring that multiple causes were active in the environment.

CONCLUSIONS

We interpret these results as providing evidence of a primary deficit in discriminative learning in participants with more severe PTSD. Specifically, a tendency to attribute a greater diversity of stimulus configurations to the same underlying cause resulted in greater uncertainty about stimulus-outcome associations, impeding learning both that certain stimuli were safe, and that certain stimuli were no longer dangerous. In the future, better understanding of the role of causal inference in trauma-related psychopathology may help refine cognitive therapies for these disorders.

[From the Ethics of AI to the Ethics of Consciousness: Ethical Aspects of Computational Psychiatry]

OBJECTIVE

Identifying ethical problems arising from AI research and Computational Psychiatry for psychiatric research and practice.

METHODS

Conceptual analysis and discussion of ethically relevant projects within Computational Psychiatry.

RESULTS

Computational Psychiatry promises a contribution to improving diagnostics and therapy (prediction). Ethical problems include dealing with data protection, consequences for our self-image, as well as the risk of biologization and the neglect of conscious experience.

CONCLUSION

It is necessary to consider possible applications of AI and Computational Psychiatry now in order to create the conditions for responsible use in the future. This requires a basic understanding of how AI applications work and of the associated ethical problems.

A Bayesian computational model reveals a failure to adapt interoceptive precision estimates across depression, anxiety, eating, and substance use disorders

Recent neurocomputational theories have hypothesized that abnormalities in prior beliefs and/or the precision-weighting of afferent interoceptive signals may facilitate the transdiagnostic emergence of psychopathology. Specifically, it has been suggested that, in certain psychiatric disorders, interoceptive processing mechanisms either over-weight prior beliefs or under-weight signals from the viscera (or both), leading to a failure to accurately update beliefs about the body. However, this has not been directly tested empirically. To evaluate the potential roles of prior beliefs and interoceptive precision in this context, we fit a Bayesian computational model to behavior in a transdiagnostic patient sample during an interoceptive awareness (heartbeat tapping) task. Modelling revealed that, during an interoceptive perturbation condition (inspiratory breath-holding during heartbeat tapping), healthy individuals (N = 52) assigned greater precision to ascending cardiac signals than individuals with symptoms of anxiety (N = 15), depression (N = 69), co-morbid depression/anxiety (N = 153), substance use disorders (N = 131), and eating disorders (N = 14)-who failed to increase their precision estimates from resting levels. In contrast, we did not find strong evidence for differences in prior beliefs. These results provide the first empirical computational modeling evidence of a selective dysfunction in adaptive interoceptive processing in psychiatric conditions, and lay the groundwork for future studies examining how reduced interoceptive precision influences visceral regulation and interoceptively-guided decision-making.

Moving beyond Ordinary Factor Analysis in Studies of Personality and Personality Disorder: A Computational Modeling Perspective

Almost all forms of psychopathology, including personality disorders, are arrived at through complex interactions among neurobiological vulnerabilities and environmental risk factors across development. Yet despite increasing recognition of etiological complexity, psychopathology research is still dominated by searches for large main effects causes. This derives in part from reliance on traditional inferential methods, including ordinary factor analysis, regression, ANCOVA, and other techniques that use statistical partialing to isolate unique effects. In principle, some of these methods can accommodate etiological complexity, yet as typically applied they are insensitive to interactive functional dependencies (modulating effects) among etiological influences. Here, we use our developmental model of antisocial and borderline traits to illustrate challenges faced when modeling complex etiological mechanisms of psychopathology. We then consider how computational models, which are rarely used in the personality disorders literature, remedy some of these challenges when combined with hierarchical Bayesian analysis.

Why Do We Need Computational Models of Psychological Change and Recovery, and How Should They Be Designed and Tested?

Traditional research methodologies typically assume that humans operate on the basis of an "open loop" stimulus-process-response rather than the "closed loop" control of internal state. They also average behavioral data across repeated measures rather than assess it continuously, and they draw inferences about the working of an individual from statistical group effects. As such, we propose that they are limited in their capacity to accurately identify and test for the mechanisms of change within psychological therapies. As a solution, we explain the advantages of using a closed loop functional architecture, based on an extended homeostatic model of the brain, to construct working computational models of individual clients that can be tested against real-world data. Specifically, we describe tests of a perceptual control theory (PCT) account of psychological change that combines the components of negative feedback control, hierarchies, conflict, reorganization, and awareness into a working model of psychological function, and dysfunction. In brief, psychopathology is proposed to be the loss of control experienced due to chronic, unresolved conflict between important personal goals. The mechanism of change across disorders and different psychological therapies is proposed to be the capacity for the therapist to help the client shift and sustain their awareness on the higher level goals that are driving goal conflict, for sufficiently long enough to permit a trial-and-error learning process, known as reorganization, to "stumble" upon a solution that regains control. We report on data from studies that have modeled these components both separately and in combination, and we describe the parallels with human data, such as the pattern of early gains and sudden gains within psychological therapy. We conclude with a description of our current research program that involves the following stages: (1) construct a model of the conflicting goals that are held by people with specific phobias; (2) optimize a model for each individual using their dynamic movement data from a virtual reality exposure task (VRET); (3) construct and optimize a learning parameter (reorganization) within each model using a subsequent VRET; (3) validate the model of each individual against a third VRET. The application of this methodology to robotics, attachment dynamics in childhood, and neuroimaging is discussed.

Under the Hood: Using Computational Psychiatry to Make Psychological Therapies More Mechanism-Focused

Psychological therapies, such as CBT, are an important part of the treatment of a range of psychiatric disorders such as depression and anxiety. There is a growing desire to understand the mechanisms by which such therapies effect change so as to improve treatment outcomes. Here we argue that adopting a computational framework may be one such approach. Computational psychiatry aims to provide a theoretical framework for moving between higher-level psychological states (like emotions, decisions and beliefs) to neural circuits, by modeling these constructs mathematically. These models are explicit hypotheses that contain quantifiable variables and parameters derived from each individual's behavior. This approach has two advantages. Firstly, some of the variables described by these models appears to reflect the neural activity of specific brain regions. Secondly, the parameters estimated by these models may offer a unique description of a patient's symptoms which can be used to both tailor therapy and track its effect. In doing so this approach may offer some additional granularity in understanding how psychological therapies, such as CBT, are working. Although this field shows significant promise, we also highlight several of the key hurdles that must first be overcome before clinical translation of computational insights can be realized.

Anxiety, avoidance, and sequential evaluation

Anxiety disorders are characterized by a range of aberrations in the processing of and response to threat, but there is little clarity what core pathogenesis might underlie these symptoms. Here we propose that a particular set of unrealistically pessimistic assumptions can distort an agent's behavior and underlie a host of seemingly disparate anxiety symptoms. We formalize this hypothesis in a decision theoretic analysis of maladaptive avoidance and a reinforcement learning model, which shows how a localized bias in beliefs can formally explain a range of phenomena related to anxiety. The core observation, implicit in standard decision theoretic accounts of sequential evaluation, is that the potential for avoidance should be protective: if danger can be avoided later, it poses less threat now. We show how a violation of this assumption - via a pessimistic, false belief that later avoidance will be unsuccessful - leads to a characteristic, excessive propagation of fear and avoidance to situations far antecedent of threat. This single deviation can explain a range of features of anxious behavior, including exaggerated threat appraisals, fear generalization, and persistent avoidance. Simulations of the model reproduce laboratory demonstrations of abnormal decision making in anxiety, including in situations of approach-avoid conflict and planning to avoid losses. The model also ties together a number of other seemingly disjoint phenomena in anxious disorders. For instance, learning under the pessimistic bias captures a hypothesis about the role of anxiety in the later development of depression. The bias itself offers a new formalization of classic insights from the psychiatric literature about the central role of maladaptive beliefs about control and self-efficacy in anxiety. This perspective also extends previous computational accounts of beliefs about control in mood disorders, which neglected the sequential aspects of choice.

[Innovative psychotherapy research: towards an evidence-based and process-based individualized and modular psychotherapy]

Psychotherapy has been proven to be effective; however, this statement applies in particular to the "average patient" in randomized controlled trials. As a considerable proportion of patients do not show any benefits despite the constant development of new therapy methods and the mechanisms of action are still too little understood, innovative psychotherapy research has to address both problems. In addition, the idea of personalization that originated in somatic medicine or - from our point of view more appropriately - individualization or person-centering should be taken up. After providing an overview of further developments in psychotherapy beyond disorder-specific methods, this article presents an evidence- and process-based individualized and modular psychotherapy as a visionary goal of psychotherapeutic research: Beyond syndromes and disorders, as many biopsychosocial characteristics as possible and the processes and mechanisms underlying the mental problems should be analyzed and bundled in an individual comprehensive functional analysis. Based on this functional analysis, evidence-based techniques and modules should be selected. The individual response during the course of therapy should be continuously documented, so that feedback helps to determine the further therapeutic procedure. In order to pursue this vision, studies are needed that are oriented towards the individual patient, investigate the central mechanisms of action and generate large translational datasets. These should be analyzed by ideographic analyses and reduce the gap between research and practice, thus contributing to the paradigm of a practice research network, which is now consistently moving to the centre of research.

Neurocognitive Adaptation and Mental Health Vulnerability Following Maltreatment: The Role of Social Functioning

Childhood maltreatment is associated with a lifetime increase in risk of mental health disorder. We propose that such vulnerability may stem in large part from altered patterns of social functioning. Here, we highlight key findings from the psychological and epidemiological literature indicating that early maltreatment experience compromises social functioning and attenuates social support in ways that increase mental health vulnerability. We then review the extant neuroimaging studies of children and adolescents, focusing on three domains implicated in social functioning: threat processing, reward processing, and emotion regulation. We discuss how adaptations in these domains may increase latent vulnerability to mental health problems by impacting on social functioning via increased stress susceptibility as well as increased stress generation. Finally, we explore how computational psychiatry approaches, alongside systematically reported measures of social functioning, can complement studies of neural function in the creation of a mechanistic framework aimed at informing approaches to prevention and intervention.

Explanatory integration, computational phenotypes, and dimensional psychiatry: The case of alcohol use disorder

We compare three theoretical frameworks for pursuing explanatory integration in psychiatry: a new dimensional framework grounded in the notion of computational phenotype, a mechanistic framework, and a network of symptoms framework. Considering the phenomenon of alcoholism, we argue that the dimensional framework is the best for effectively integrating computational and mechanistic explanations with phenomenological analyses.

Motivation and cognitive control in depression

Depression is linked to deficits in cognitive control and a host of other cognitive impairments arise as a consequence of these deficits. Despite of their important role in depression, there are no mechanistic models of cognitive control deficits in depression. In this paper we propose how these deficits can emerge from the interaction between motivational and cognitive processes. We review depression-related impairments in key components of motivation along with new cognitive neuroscience models that focus on the role of motivation in the decision-making about cognitive control allocation. Based on this review we propose a unifying framework which connects motivational and cognitive control deficits in depression. This framework is rooted in computational models of cognitive control and offers a mechanistic understanding of cognitive control deficits in depression.

Annual Research Review: Developmental computational psychiatry

Most psychiatric disorders emerge during childhood and adolescence. This is also a period that coincides with the brain undergoing substantial growth and reorganisation. However, it remains unclear how a heightened vulnerability to psychiatric disorder relates to this brain maturation. Here, we propose 'developmental computational psychiatry' as a framework for linking brain maturation to cognitive development. We argue that through modelling some of the brain's fundamental cognitive computations, and relating them to brain development, we can bridge the gap between brain and cognitive development. This in turn can lead to a richer understanding of the ontogeny of psychiatric disorders. We illustrate this perspective with examples from reinforcement learning and dopamine function. Specifically, we show how computational modelling deepens an understanding of how cognitive processes, such as reward learning, effort learning, and social learning might go awry in psychiatric disorders. Finally, we sketch the promises and limitations of a developmental computational psychiatry.

"Surprise" and the Bayesian Brain: Implications for Psychotherapy Theory and Practice

The free energy principle (FEP) has gained widespread interest and growing acceptance as a new paradigm of brain function, but has had little impact on the theory and practice of psychotherapy. The aim of this paper is to redress this. Brains rely on Bayesian inference during which “bottom-up” sensations are matched with “top-down” predictions. Discrepancies result in “prediction error.” The brain abhors informational “surprise,” which is minimized by (1) action enhancing the statistical likelihood of sensory samples, (2) revising inferences in the light of experience, updating “priors” to reality-aligned “posteriors,” and (3) optimizing the complexity of our generative models of a capricious world. In all three, free energy is converted to bound energy. In psychopathology energy either remains unbound, as in trauma and inhibition of agency, or manifests restricted, anachronistic “top-down” narratives. Psychotherapy fosters client agency, linguistic and practical. Temporary uncoupling bottom-up from top-down automatism and fostering scrutinized simulations sets a number of salutary processes in train. Mentalising enriches Bayesian inference, enabling experience and feeling states to be “metabolized” and assimilated. “Free association” enhances more inclusive sensory sampling, while dream analysis foregrounds salient emotional themes as “attractors.” FEP parallels with psychoanalytic theory are outlined, including Freud’s unpublished project, Bion’s “contact barrier” concept, the Fonagy/Target model of sexuality, Laplanche’s therapist as “enigmatic signifier,” and the role of projective identification. The therapy stimulates patients to become aware of and revise the priors’ they bring to interpersonal experience. In the therapeutic “duet for one,” the energy binding skills and non-partisan stance of the analyst help sufferers face trauma without being overwhelmed by psychic entropy. Overall, the FEP provides a sound theoretical basis for psychotherapy practice, training, and research.