Depression is characterized by abnormalities in emotional processing, but the specific drivers of such emotional abnormalities are unknown. Computational work indicates that both surprising outcomes (prediction errors; PEs) and outcomes (values) themselves drive emotional responses, but neither has been consistently linked to affective disturbances in depression. As a result, the computational mechanisms driving emotional abnormalities in depression remain unknown.

Here, in 687 individuals, one-third of whom qualify as depressed via a standard self-report measure (the PHQ-9), we use high-stakes, naturalistic events – the reveal of midterm exam grades – to test whether individuals with heightened depression display a specific reduction in emotional response to positive PEs.

Using Bayesian mixed effects models, we find that individuals with heightened depression do not affectively benefit from surprising, good outcomes – that is, they display reduced affective responses to positive PEs. These results were highly specific: effects were not observed to negative PEs, value signals (grades), and were not related to generalized anxiety. This suggests that the computational drivers of abnormalities in emotion in depression may be specifically due to positive PE-based emotional responding.

Affective abnormalities are core depression symptoms, but the computational mechanisms underlying such differences are unknown. This work suggests that blunted affective reactions to positive PEs are likely mechanistic drivers of emotional dysregulation in depression.

Cognitive distancing is an emotion regulation strategy commonly used in psychological treatment of various mental health disorders, but its therapeutic mechanisms are unknown.

935 participants completed an online reinforcement learning task involving choices between pairs of symbols with differing reward contingencies. Half (49.1%) of the sample was randomised to a cognitive self-distancing intervention and were trained to regulate or ‘take a step back’ from their emotional response to feedback throughout. Established computational (Q-learning) models were then fit to individuals' choices to derive reinforcement learning parameters capturing clarity of choice values (inverse temperature) and their sensitivity to positive and negative feedback (learning rates).

Cognitive distancing improved task performance, including when participants were later tested on novel combinations of symbols without feedback. Group differences in computational model-derived parameters revealed that cognitive distancing resulted in clearer representations of option values (estimated 0.17 higher inverse temperatures). Simultaneously, distancing caused increased sensitivity to negative feedback (estimated 19% higher loss learning rates). Exploratory analyses suggested this resulted from an evolving shift in strategy by distanced participants: initially, choices were more determined by expected value differences between symbols, but as the task progressed, they became more sensitive to negative feedback, with evidence for a difference strongest by the end of training.

Adaptive effects on the computations that underlie learning from reward and loss may explain the therapeutic benefits of cognitive distancing. Over time and with practice, cognitive distancing may improve symptoms of mental health disorders by promoting more effective engagement with negative information.

Mood instability and risk-taking are hallmarks of borderline personality disorder (BPD). Schema modes are combinations of self-reflective evaluations, negative emotional states, and destructive coping strategies common in BPD. When activated, they can push patients with BPD into emotional turmoil and a dissociative state of mind. Our knowledge of the underlying neurocognitive mechanisms driving these changes is incomplete. We hypothesized that in patients with BPD, affective instability is more influenced by reward expectation, outcomes, and reward prediction errors (RPEs) during risky decision-making than in healthy controls. Additionally, we expected that these alterations would be related to schema modes.

Thirty-two patients with BPD and thirty-one healthy controls were recruited. We used an established behavioral paradigm to measure mood fluctuations during risky decision-making. The impact of expectations and RPEs on momentary mood was quantified by a computational model, and its parameters were estimated with hierarchical Bayesian analysis. Model parameters were compared using High-Density Intervals.

We found that model parameters capturing the influence of RPE and Certain Rewards on mood were significantly higher in patients with BPD than in controls. These model parameters correlated significantly with schema modes, but not with depression severity.

BPD is coupled with altered associations between mood fluctuation and reward processing under uncertainty. Our findings seem to be BPD-specific, as they stand in contrast with the correlates of depressive symptoms. Future studies should establish the clinical utility of these alterations, such as predicting or assessing therapeutic response in BPD.

No significant relationships.

Major depressive disorder (MDD) is associated with impaired reward processing and reward learning. The literature is inconclusive regarding whether these impairments persist after remission. The current study examined reward processing during a probabilistic learning task in individuals in remission from MDD (n = 19) and never depressed healthy controls (n = 31) matched for age and sex. The outcome measures were pupil dilation (an indirect index of noradrenergic activity and arousal) and computational modeling parameters.

Participants completed two versions (facial/nonfacial feedback) of probabilistic reward learning task with changing contingencies. Pupil dilation was measured with a corneal reflection eye tracker. The hypotheses and analysis plan were preregistered.

Healthy controls had larger pupil dilation following losses than gains (p <.001), whereas no significant difference between outcomes was found in individuals with a history of MDD, resulting in an interaction between group and outcome (β = 0.81, SE = 0.34, t = 2.37, p = .018). The rMDD group also achieved lower mean score at the last trial (t[46.77] = 2.12, p = .040) as well as a smaller proportion of correct choices (t[46.70] = 2.09, p = .041) compared with healthy controls.

Impaired reward processing may persist after remission from MDD and could constitute a latent risk factor for relapse. Measuring pupil dilation in a reward learning task is a promising method for identifying reward processing abnormalities linked to MDD. The task is simple and noninvasive, which makes it feasible for clinical research.

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.

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.

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.

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.

Psychiatric disorders, including eating disorders (EDs), have clinical outcomes that range widely in severity and chronicity. The ability to predict such outcomes is extremely limited. Machine-learning (ML) approaches that model complexity may optimize the prediction of multifaceted psychiatric behaviors. However, the investigations of many psychiatric concerns have not capitalized on ML to improve prognosis. This study conducted the first comparison of an ML approach (elastic net regularized logistic regression) to traditional regression to longitudinally predict ED outcomes.

Females with heterogeneous ED diagnoses completed demographic and psychiatric assessments at baseline (n = 415) and Year 1 (n = 320) and 2 (n = 277) follow-ups. Elastic net and traditional logistic regression models comprising the same baseline variables were compared in ability to longitudinally predict ED diagnosis, binge eating, compensatory behavior, and underweight BMI at Years 1 and 2.

Elastic net models had higher accuracy for all outcomes at Years 1 and 2 [average Area Under the Receiving Operating Characteristics Curve (AUC) = 0.78] compared to logistic regression (average AUC = 0.67). Model performance did not deteriorate when the most important predictor was removed or an alternative ML algorithm (random forests) was applied. Baseline ED (e.g. diagnosis), psychiatric (e.g. hospitalization), and demographic (e.g. ethnicity) characteristics emerged as important predictors in exploratory predictor importance analyses.

ML algorithms can enhance the prediction of ED symptoms for 2 years and may identify important risk markers. The superior accuracy of ML for predicting complex outcomes suggests that these approaches may ultimately aid in advancing precision medicine for serious psychiatric disorders.

Mood and anxiety disorders are ubiquitous but current treatment options are ineffective for many sufferers. Moreover, a number of promising pre-clinical interventions have failed to translate into clinical efficacy in humans. Improved treatments are unlikely without better animal–human translational pipelines. Here, we translate a rodent measure of negative affective bias into humans, exploring its relationship with (1) pathological mood and anxiety symptoms and (2) transient induced anxiety.

Adult participants (age = 29 ± 11) who met criteria for mood or anxiety disorder symptomatology according to a face-to-face neuropsychiatric interview were included in the symptomatic group. Study 1 included N = 77 (47 = asymptomatic [female = 21]; 30 = symptomatic [female = 25]), study 2 included N = 47 asymptomatic participants (25 = female). Outcome measures were choice ratios, reaction times and parameters recovered from a computational model of reaction time – the drift diffusion model (DDM) – from a two-alternative-forced-choice task in which ambiguous and unambiguous auditory stimuli were paired with high and low rewards.

Both groups showed over 93% accuracy on unambiguous tones indicating intact discrimination, but symptomatic individuals demonstrated increased negative affective bias on ambiguous tones [proportion high reward = 0.42 (s.d. = 0.14)] relative to asymptomatic individuals [0.53 (s.d. = 0.17)] as well as a significantly reduced DDM drift rate. No significant effects were observed for the within-subjects anxiety-induction.

Humans with pathological anxiety symptoms directly mimic rodents undergoing anxiogenic manipulation. The lack of sensitivity to transient anxiety suggests the paradigm might be more sensitive to clinically relevant symptoms. Our results establish a direct translational pipeline (and candidate therapeutics screen) from negative affective bias in rodents to pathological mood and anxiety symptoms in humans.