IntroductionCognitive behavioral therapy (CBT) is one of the most common interventions for depression and has two key components: Cognitive Restructuring (CR) and Behavioral Activation (BA). However, no evidence-based guidelines exist to help clients and clinicians decide whether CBT would be a good first-line treatment for a given individual based on their personal characteristics, and which CBT intervention would benefit them more. We propose that specific capacities to learn from new information and experiences are prerequisites for response to CBT and that BA and CR require different learning capacities. In this study, we aim to develop predictive models of symptom change based on computationally-derived variables from behavioral tasks, in addition to clinical and demographic self-report data, to identify parameters and variables that can determine which individuals with depressive symptoms would benefit from CBT-based interventions and, ideally, which specific interventions they would benefit from more.Methods and analysisWe plan to recruit at least 1,500 adult participants who report having symptoms of depression and reside in U.S. After completing a series of questionnaires and behavioral tasks to assess their learning propensities, participants will be randomly assigned to a BA or a CR group. Using an online self-help tool, participants will then engage with designated modules according to their assigned group for five weeks. We will assess symptoms 1 week post-intervention (main end point of study) and follow up at 6, 18, and 42 weeks post-intervention. Upon enrolling and consenting into the main study, participants will be randomly assigned to either the training dataset or the held-out test dataset at a ratio of 2:1. This enables a clean separation of training and test datasets and prevent data leakage. We plan to build cross-validated predictive algorithms on the training dataset, and preregister our analysis plan before we validate our models and hypotheses in the held-out, unseen, test dataset. Enrollment of the study started 23rd January, 2024.Study protocol registrationClinicalTrials.gov, identifier (NCT06631183). The protocol follows the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines. Numbers in brackets follow subsection numbers in the guidelines.

Teaching is a foundational social behaviour that can result from mentally effortful reasoning or cognitively frugal heuristics. When do people use these strategies to teach? Here we investigated this using behavioural experiments and computational modelling with adult participants recruited via Prolific. Experiment 1 (N = 100) revealed robust individual differences: some participants taught by reasoning about a learner's knowledge, consistent with an optimal Bayesian pedagogy model, while others relied on simple heuristics that do not require mentalizing. In two preregistered follow-up experiments, we found that people persist in using heuristics even when they are no longer effective (experiment 2, N = 253, P < 0.001, rank-biserial r = 0.287, 95% confidence interval 0.149-0.419) but that this tendency is pre-empted when inference about a learner's knowledge is scaffolded using an auxiliary task (experiment 3, N = 759, P < 0.001, partial η p 2 = 0 . 107 , 95% confidence interval 0.068-0.148). These results demonstrate sophisticated arbitration between planning and heuristics during teaching and elucidate the more general mechanisms involved in adapting mental effort during social interactions.

Importance:There is growing concern about the impact of social media on mental health. Understanding the psychological processes underlying social media use could lead to greater insight. Objective:To investigate individual differences in the effect of social media likes on posting behavior and their association with and specificity to depressive psychopathology. Design, Setting, and Participants:This cross-sectional study used longitudinal Twitter behavior to test the association between depressive and other forms of psychopathology and the tendency to be reinforced by likes. The study involved passively scraped Twitter data (dataset 1) or participants recruited online through Twitter ads (dataset 2) and a recruitment platform (dataset 3). In total, the 3 datasets contained over 17 million posts from 7736 Twitter users. These data were analyzed from March 2023 through January 2026. Main Outcomes and Measures:Reinforcement tendency was estimated as the association between previous-day likes and current-day posting. The depression measure varied between datasets: self-disclosure in dataset 1, past 4-week depression severity in dataset 2, and a validated questionnaire in dataset 3. Dataset 3 also included other psychiatric measures used to estimate scores on transdiagnostic symptom scores. Results:Dataset 1 comprised 1045 users who reported a depression diagnosis on Twitter, and 5001 randomly sampled users. Dataset 2 comprised 601 users (mean [range] age, 45.2 years [18-78]; 251 [41.7%] women, 333 [55.5%] men, and 17 [2.8%] individuals with other gender identities). Dataset 3 comprised 1089 users (mean [range] age, 30.8 years [18-68]; 703 [64.6%] women, 358 [32.8%] men, and 28 [2.6%] individuals with other gender identities). Individuals with a depression diagnosis or higher symptoms of depression tended more to be reinforced by likes in all datasets (dataset 1: β = .013; P = .002; dataset 2: β = .026; P = .03; dataset 3 [preregistered]: β = .008, P = .02), despite substantial differences between datasets in depression measurement and composition. In dataset 3, greater reinforcement tendency was specifically associated with a transdiagnostic anxious-depression factor, whereas behavioral persistence was linked to a compulsivity and intrusive thought factor. Conclusions and Relevance:In this study, depressive psychopathology was associated with a greater tendency to be reinforced by a type of social reward on Twitter, in contrast to typical laboratory-based findings of blunted reinforcement learning in depression. These findings suggest potential mechanisms linking social media use to worse mental health or vice versa, and emphasize the importance of studying real-world behavior.

The interaction between social media use and mental health is of great public health concern. Studies so far, largely employing self-reported measures of social media use, have produced inconclusive evidence regarding the impact of social media on mental health. Focusing on objective behavioral markers and the psychological mechanisms underlying how users interact with social media platforms could be key to greater insight on this topic. Here we use Twitter data to study how depression modulates a central behavioral process on social media: the response to the social rewards (e.g. likes, shares, views) users receive when they post. Reinforcement learning theory predicts that social media rewards will reinforce posting behavior, such that receiving more likes will lead to posting more frequently and spending more time on the platform. However, laboratory tasks often show blunted reinforcement learning in depression, suggesting a potential attenuation of the effects of social rewards on posting behavior. Across 3 datasets with varied measures of depression and data collection strategies (over 17 million tweets from 7,736 users in total, including a pre-registered replication), we consistently found that depression was associated with a larger reinforcing effect of likes on posting on the next day. In other words, users with depression showed heightened sensitivity to social media rewards, in contrast to findings from laboratory-based tasks. These results identify a psychological mechanism that may link social media use to poor mental health, and underscore the importance of testing the generalizability of in-lab computational psychiatry findings to real-world environments.

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. Applying computational models of reinforcement learning can help to improve psychotherapeutic interventions by helping to identify the therapeutic mechanisms of change and identifying for whom therapies will be most effective.

This study examined how the emotional dimensions of arousal and valence influence effort-based decision-making. Twenty-eight participants were exposed to either low arousal & high valence (-A/+V) or high arousal & low valence (+A/-V) stimuli in a task that required choosing between a more rewarding but also more effortful option versus a less effortful but also less rewarding one. We recorded eye movements, fixation times, and self-reported arousal and valence ratings. Results showed that participants in the +A/-V condition were more likely to select the effortful option (62% vs. 56%), with a significant interaction between arousal and valence (beta = 0.46, 95% CI [0.26, 0.67]; rho <.001). Specifically, positive valence reduced preference for effortful choices, but this effect became null under high arousal. Additionally, participants in the -A/+V condition took longer to make decisions and attended more to reward information than to effort information. These findings suggest complex emotional influences on effort-based decisions.

Humans exhibit a striking tendency to persist with chosen goals. This strong attachment to goals can often appear irrational - a perspective captured by terms such as perseverance or sunk-cost biases. In this review, we explore how goal commitment could stem from several adaptive mechanisms, including those that optimise cognitive resources, shield decisions from interference, and scaffold motivation in the absence of accessible reward signals. We propose that these computational considerations have important implications for algorithmic architectures supporting decision making, including separate algorithms for goal selection and implementation, and for monitoring ongoing goals versus alternative sources of reward. Finally, we discuss how a variety of mechanisms supporting goal commitment and abandonment could relate to dimensions affected in mental health.

Return of fear after exposure poses a significant challenge for treatment of anxiety dis- orders. In this study, we used computational modeling to test competing mechanisms underlying spontaneous recovery of fear over time. We fit computational models of a novel theory of spontaneous recovery—selective maintenance of aversive memories— to behavior from a fear conditioning and extinction task (N=316), and showed that they uniquely captured spontaneous recovery and quantitatively outperformed alternative models embodying theories from the literature. The results were supported across mul- tiple datasets, including a preregistered replication (N=355) and a sample with mental health symptoms (N=520). The selective maintenance modeling framework addition- ally offers mechanistic insights into overgeneralization and the development of anxi- ety. Indeed, in the symptomatic sample we found that symptoms of generalized anxiety disorder correlated with estimates of overgeneralization in the model. Through simu- lations, we further demonstrated that insights from our model can explain how targeted interventions such as retrieval cues and cognitive interventions can prevent the return of fear. These results highlight selective maintenance of aversive events in memory as a critical and testable target for improving anxiety treatments and preventing relapse.

Background: The Pavlovian go/no-go task is commonly used to measure individual differences in Pavlovian biases and their interaction with instrumental learning. However, prior research has found suboptimal reliability for computational model-based performance measures for this task, limiting its usefulness in individual-differences research. These studies did not make use of several strategies previously shown to enhance task-measure reliability (e.g., task gamification, hierarchical Bayesian modeling for model estimation). Here we investigated if such approaches could improve the task’s reliability. Methods: Across two experiments, we recruited two independent samples of adult participants (N=103, N=110) to complete a novel, gamified version of the Pavlovian go/no-go task multiple times over several weeks. We used hierarchical Bayesian modeling to derive reinforcement learning model-based indices of participants' task performance, and additionally to estimate the reliability of these measures. Results: In Experiment 1, we observed considerable and unexpected practice effects, with most participants reaching near-ceiling levels of performance with repeat testing. Consequently, the test-retest reliability of some model parameters was unacceptable (range: 0.379–0.973). In Experiment 2, participants completed a modified version of the task designed to lessen these practice effects. We observed greatly reduced practice effects and improved estimates of the test-retest reliability (range: 0.696–0.989). Conclusion: The results demonstrate that model-based measures of performance on the Pavlovian go/no-go task can reach levels of reliability sufficient for use in individual- differences research. However, additional investigation is necessary to validate the modified version of the task in other populations and settings.

Schemas are rich and complex knowledge structures about the typical unfolding of events in a context. For example, a schema of a dinner at a restaurant. In this Perspective, we suggest that reinforcement learning (RL), a computational theory of learning the structure of the world and relevant goal-oriented behavior, underlies schema learning. We synthesize literature about schemas and RL to offer that three RL principles might govern the learning of schemas: learning via prediction errors, constructing hierarchical knowledge using hierarchical RL and dimensionality reduction through learning a simplified and abstract representation of the world. We then suggest that the orbito-medial prefrontal cortex is involved in both schemas and RL due to its involvement in dimensionality reduction and in guiding memory reactivation through interactions with posterior brain regions. Last, we hypothesize that the amount of dimensionality reduction might underlie gradients of involvement along the ventral-dorsal and posterior-anterior axes of the orbito-medial prefrontal cortex. More specific and detailed representations might engage the ventral and posterior parts, whereas abstraction might shift representations toward the dorsal and anterior parts of the medial prefrontal cortex.

Human goal selection is simultaneously flexible and structured. Existing accounts explain thisdichotomy in terms of a biological reward function that may have proved adaptive on evolutionarytimescales. However, such explanations struggle to capture the human ability to quickly and flexiblyreevaluate goals and stimuli in response to changing contexts. In this article, we consider howhuman goal selection may be shaped by cognitive limitations. We structure this perspective aroundthree computational-style problems that people face when selecting their own goals: consideration,evaluation, and feasibility. Overall, we offer a brief survey of relevant literature alongside a resource-rational account of structured goal selection without appealing to intrinsic rewards.

A core strength of computational psychiatry is its focus on theory-driven research, in which cognitive processes are precisely quantified using computational models that formalize specific theoretical mechanisms. However, the data used in these studies often come from traditional laboratory-based cognitive tasks, which have unclear ecological validity. In this review we propose that the same theoretical frameworks and computational models can be applied to real-world data such as experience sampling, passive data, and digital-behavior data (e.g., online activity such as on social media). In turn, modeling real-world data can benefit from a theory-driven computational approach to move from purely predictive to explanatory power. We illustrate these points using emerging studies and discuss the challenges and opportunities of using real-world data in computational psychiatry.

Anxiety is one of the most prevalent mental health concerns. Current theories suggest that anxiety may arise due to deficits in segmentation of continuous experience into discrete context representations (‘event segmentation’), which leads to overgeneralization of fear across contexts, or conversely, overly rigid segmentation that prevents safety learning. Here, in two segmentation tasks (N=1109), we found novel and direct evidence that anxiety is associated with changes in event segmentation. Individuals with higher anxiety symptoms responded more slowly to transitions between events (event boundaries). They also segmented movies into discrete events more typically and more hierarchically, two hallmarks of precise segmentation. This precision was linked to self-reporting fewer context changes in daily life, suggesting individuals with anxiety prefer stable and predictable environments. These findings challenge overgeneralization theories of anxiety, revealing instead that individuals with anxiety exhibit precise, and potentially overly rigid segmentation. Such segmentation could maintain fear by preventing generalization from safe to fearful contexts, which has important implications for interventions like exposure therapy.

Spontaneous recovery of fear after extinction is a well- established behavioral phenomenon. Different theories in psychology account for spontaneous recovery by proposing that it may result from temporal weighting, reduced processing of stimuli over time, enhanced salience of adverse events or re- turn of the acquisition context. We propose a novel mechanism of spontaneous recovery: selective maintenance of adverse events, and ground this mechanism in a computational model of latent cause inference. To investigate the proposed mechanism, we collected behavioral data with an aversive conditioning and extinction task (N=280) and fit the data with computational models formalizing our and others’ theories. Quantitative and qualitative model comparisons indicated that selective maintenance of adverse events accounts for spontaneous recovery better than alternative theories. As spontaneous recovery of fear after extinction can serve as a model of relapse after exposure therapy, we use this mechanistic understanding of spontaneous recovery to propose and simulate the effect of add-on interventions to prevent relapse after exposure therapy.

Here, we describe the efforts we dedicated to the challenge of modifying entrenched emotionally laden memories. In recent years, through a number of collaborations and using a combination of behavioral, molecular, and computational approaches, we: (a) developed novel approaches to fear attenuation that engage mechanisms that differ from those engaged during extinction (Monfils), (b) examined whether our approaches can generalize to other reinforcers (Lee, Gonzales, Chaudhri, Cofresi, and Monfils), (c) derived principled explanations for the differential outcomes of our approaches (Niv, Gershman, Song, and Monfils), (d) developed better assessment metrics to evaluate outcome success (Shumake and Monfils), (e) identified biomarkers that can explain significant variance in our outcomes of interest (Shumake and Monfils), and (f) developed better basic research assays and translated efforts to the clinic (Smits, Telch, Otto, Shumake, and Monfils). We briefly highlight each of these milestones and conclude with final remarks and extracted principles. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Humans have an outstanding ability to generalize from past experiences, which requires parsing continuously experienced events into discrete, coherent units, and relating them to similar past experiences. Time is a key element in this process; however, how temporal information is used in generalization remains unclear. Latent-cause inference provides a Bayesian framework for clustering experiences, by building a world model in which related experiences are generated by a shared cause. Here, we examine how temporal information is used in latent-cause inference, using a novel task in which participants see "microbe" stimuli and explicitly report the latent cause ("strain") they infer for each microbe. We show that humans incorporate time in their inference of latent causes, such that recently inferred latent causes are more likely to be inferred again. In particular, a "persistent" model, in which the latent cause inferred for one observation has a fixed probability of continuing to cause the next observation, explains the data significantly better than two other time-sensitive models, although extensive individual differences exist. We show that our task and this model have good psychometric properties, highlighting their potential use for quantifying individual differences in computational psychiatry or in neuroimaging studies.

Computational models of addiction often rely on a model-free reinforcement learning (RL) formulation, owing to the close associations between model-free RL, habitual behavior and the dopaminergic system. However, such formulations typically do not capture key recurrent features of addiction phenomena such as craving and relapse. Moreover, they cannot account for goal-directed aspects of addiction that necessitate contrasting, model-based formulations. Here we synthesize a growing body of evidence and propose that a latent-cause framework can help unify our understanding of several recurrent phenomena in addiction, by viewing them as the inferred return of previous, persistent “latent causes”. We demonstrate that applying this framework to Pavlovian and instrumental settings can help account for defining features of craving and relapse such as outcome-specificity, generalization, and cyclical dynamics. Finally, we argue that this framework can bridge model-free and model-based formulations, and account for individual variability in phenomenology by accommodating the memories, beliefs, and goals of those living with addiction, motivating a centering of the individual, subjective experience of addiction and recovery.

Our goals fundamentally shape how we experience the world. For example, when we are hungry, we tend to view objects in our environment according to whether or not they are edible (or tasty). Alternatively, when we are cold, we may view the very same objects according to their ability to produce heat. Computational theories of learning in cognitive systems, such as reinforcement learning, use the notion of "state-representation" to describe how agents decide which features of their environment are behaviorally-relevant and which can be ignored. However, these approaches typically assume "ground-truth" state representations that are known by the agent, and reward functions that need to be learned. Here we suggest an alternative approach in which state-representations are not assumed veridical, or even pre-defined, but rather emerge from the agent's goals through interaction with its environment. We illustrate this novel perspective by inferring the goals driving rat behavior in an odor-guided choice task and discuss its implications for developing, from first principles, an information-theoretic account of goal-directed state representation learning and behavior.