Weight gained during treatment predicts 6-month body mass index in a large sample of patients with anorexia nervosa using ensemble machine learning.

OBJECTIVE: This study used machine learning methods to analyze data on treatment outcomes from individuals with anorexia nervosa admitted to a specialized eating disorders treatment program. METHODS: Of 368 individuals with anorexia nervosa (209 adolescents and 159 adults), 160 individuals had data available for a 6-month follow-up analysis. Participants were treated in a 6-day-per-week partial-hospital program. Participants were assessed for eating disorder-specific and non-specific psychopathology. The analyses used established machine learning procedures combined in an ensemble model from support vector machine learning, random forest prediction, and the elastic net regularized regression with an exploration (training; 75%) and confirmation (test; 25%) split of the data. RESULTS: The models predicting body mass index (BMI) at 6-month follow-up explained a 28.6% variance in the training set (n = 120). The model had good performance in predicting 6-month BMI in the test dataset (n = 40), with predicted BMI significantly correlating with actual BMI (r = .51, p = 0.01). The change in BMI from admission to discharge was the most important predictor, strongly correlating with reported BMI at 6-month follow-up (r = .55). Behavioral variables were much less predictive of BMI outcome. Results were similar for z-transformed BMI in the adolescent-only group. Length of stay was most predictive of weight gain in treatment (r = .56) but did not predict longer-term BMI. CONCLUSIONS: This study, using an agnostic ensemble machine learning approach in the largest to-date sample of individuals with anorexia nervosa, suggests that achieving weight gain goals in treatment predicts longer-term weight-related outcomes. Other potential predictors, personality, mood, or eating disorder-specific symptoms were relatively much less predictive. PUBLIC SIGNIFICANCE: The results from this study indicate that the amount of weight gained during treatment predicts BMI 6 months after discharge from a high level of care. This suggests that patients require sufficient time in a higher level of care treatment to meet their specific weight goals and be able to maintain normal weight.

An Introduction to Generative Artificial Intelligence in Mental Health Care: Considerations and Guidance.

PURPOSE OF REVIEW: This paper provides an overview of generative artificial intelligence (AI) and the possible implications in the delivery of mental health care. RECENT FINDINGS: Generative AI is a powerful technology that is changing rapidly. As psychiatrists, it is important for us to understand generative AI technology and how it may impact our patients and our practice of medicine. This paper aims to build this understanding by focusing on GPT-4 and its potential impact on mental health care delivery. We first introduce key concepts and terminology describing how the technology works and various novel uses of it. We then dive into key considerations for GPT-4 and other large language models (LLMs) and wrap up with suggested future directions and initial guidance to the field.

Can peripheral psychophysiological markers predict response to exposure-based cognitive behavioral therapy in youth with severely impairing irritability? A study protocol.

BACKGROUND: Irritability, an increased proneness to anger, is a primary reason youth present for psychiatric care. While initial evidence supports the efficacy of exposure-based cognitive behavioral therapy (CBT) for youth with clinically impairing irritability, treatment mechanisms remain unclear. Here, we propose to measure peripheral psychophysiological indicators of arousal-heart rate (HR)/electrodermal activity (EDA)-and regulation-heart rate variability (HRV)-during exposures to anger-inducing stimuli as potential predictors of treatment efficacy. The objective of this study is to evaluate whether in-situ biosensing data provides peripheral physiological indicators of in-session response to exposures. METHODS: Blood volume pulse (BVP; from which HR and HRV canl be derived) and EDA will be collected ambulatorily using the Empatica EmbracePlus from 40 youth (all genders; ages 8-17) undergoing six in-person exposure treatment sessions, as part of a multiple-baseline trial of exposure-based CBT for clinically impairing irritability. Clinical ratings of irritability will be conducted at baseline, weekly throughout treatment, and at 3-month and 6-month follow-ups via the Clinical Global Impressions Scale (CGI) and the Affective Reactivity Index (ARI; clinician-, parent-, and child-report). Multilevel modeling will be used to assess within- and between-person changes in physiological arousal and regulation throughout exposure-based CBT and to determine whether individual differences are predictive of treatment response. DISCUSSION: This study protocol leverages a wearable biosensor (Empatica) to continuously record HR/HRV (derived from BVP) and EDA during in-person exposure sessions for youth with clinically impairing irritability. Here, the goal is to identify changes in physiological arousal (EDA, HR) and regulation (HRV) over the course of treatment in tandem with changes in clinical symptoms. TRIAL REGISTRATION: The participants in this study come from an overarching clinical trial (trial registration numbers: NCT02531893 first registered on 8/25/2015; last updated on 8/25/2023). The research project and all related materials were submitted and approved by the appropriate Institutional Review Board of the National Institute of Mental Health (NIMH).

Reliability of task-evoked neural activation during face-emotion paradigms: Effects of scanner and psychological processes.

Assessing and improving test-retest reliability is critical to efforts to address concerns about replicability of task-based functional magnetic resonance imaging. The current study uses two statistical approaches to examine how scanner and task-related factors influence reliability of neural response to face-emotion viewing. Forty healthy adult participants completed two face-emotion paradigms at up to three scanning sessions across two scanners of the same build over approximately 2 months. We examined reliability across the main task contrasts using Bayesian linear mixed-effects models performed voxel-wise across the brain. We also used a novel Bayesian hierarchical model across a predefined whole-brain parcellation scheme and subcortical anatomical regions. Scanner differences accounted for minimal variance in temporal signal-to-noise ratio and task contrast maps. Regions activated during task at the group level showed higher reliability relative to regions not activated significantly at the group level. Greater reliability was found for contrasts involving conditions with clearly distinct visual stimuli and associated cognitive demands (e.g., face vs. nonface discrimination) compared to conditions with more similar demands (e.g., angry vs. happy face discrimination). Voxel-wise reliability estimates tended to be higher than those based on predefined anatomical regions. This work informs attempts to improve reliability in the context of task activation patterns and specific task contrasts. Our study provides a new method to estimate reliability across a large number of regions of interest and can inform researchers' selection of task conditions and analytic contrasts.

The Neural Circuitry of Reward During Sustained Threat.

Reward processing is important for understanding behavior in psychopathology. Opportunities to earn money activate the ventral striatum, as shown by the monetary incentive delay (MID) task. Anxiety conditions have been modeled by presenting shocks and startling sounds. To further investigate the co-occurrence of an anxiety condition and a rewarding stimulus, we modified the MID to include a sustained threat of scream. This study investigated neural patterns of the MID task with an uncertain threat of a startling scream. Forty-three young adults completed a functional MRI scan. The task included two conditions (scream and safe) and three cues (gain $5, gain $0, lose $5). Analyses included a whole brain, group analysis using a linear mixed-effects model and a paired t-test. The whole brain analysis revealed a main effect of cue, with increased ventral striatal activation (F2,210 = 58.8, p < 0.001) during cues to gain or lose $5. We observed a main effect of condition during cue presentation, such that bilateral insula and putamen activation was diminished (p < 0.001) in the scream versus the safe condition. A t-test of condition showed increased activation during threat blocks in the insula and putamen. A time course graph revealed that activation in the insula and putamen responded similarly to incentive but had an overall elevation during the scream condition. These results replicated expected activation in reward and in the setting of uncertain threat. Our results show that uncertain threat increases the magnitude of activation in the dorsal striatum.

Brain reward response in adolescents and young adults with anorexia nervosa is moderated by changes in body weight and sweetness perception.

OBJECTIVE: Anorexia nervosa (AN) is a severe psychiatric illness with complex etiology. Recently, we found elevated striatal brain response to sweet taste stimuli in adolescents and young adults with AN. Here, we tested the hypothesis that nutritional rehabilitation normalizes prediction error activation, a measure for dopamine-related reward circuit response, to salient caloric taste stimuli in AN. METHODS: A total of 28 individuals with AN (age = 16 ± 2 years; body mass index [BMI] = 16 ± 1) who previously underwent brain imaging while performing a taste prediction error task using sucrose as salient caloric stimulus, participated in a second brain imaging scan (BMI = 18 ± 1) after intensive specialized eating disorder treatment (41 ± 15 days). A total of 31 healthy controls (age = 16 ± 3 years; BMI = 21 ± 2) were also studied on two occasions. RESULTS: At baseline, individuals with AN demonstrated an elevated salience response in bilateral caudate head and nucleus accumbens, and right ventral striatum. At the second scan, elevated response was only found in the right nucleus accumbens. A moderator analysis indicated that greater increase in BMI and greater decrease in sweetness perception predicted lesser prediction error response at the second scan in AN. CONCLUSION: Consistent with the previously reported monetary stimulus-response, elevated taste prediction error response in AN was largely absent after weight restoration. This study indicates that changes in BMI and sweet taste perception are independent moderators of change of brain salience response in adolescents and young adults with AN. The study points toward dynamic changes in the brain reward circuitry in AN and highlights the importance of nutrition and weight restoration in that process. PUBLIC SIGNIFICANCE STATEMENT: AN is a severe psychiatric illness. Biological factors that integrate neurobiology and behavior could become important targets to improve treatment outcome. This study highlights the importance of weight normalization and taste perception the normalization of brain function, and food type or taste-specific interventions could help in the recovery process. Furthermore, the study suggests that food-related and nonfood-related reward processing adapts to illness state in AN.

Changes in Internalizing Symptoms During the COVID-19 Pandemic in a Transdiagnostic Sample of Youth: Exploring Mediators and Predictors.

The COVID-19 pandemic is a chronically stressful event, particularly for youth. Here, we examine (i) changes in mood and anxiety symtpoms, (ii) pandemic-related stress as a mediator of change in symptoms, and (ii) threat processing biases as a predictor of increased anxiety during the pandemic. A clinically well-characterized sample of 81 youth ages 8-18 years (M = 13.8 years, SD = 2.65; 40.7% female) including youth with affective and/or behavioral psychiatric diagnoses and youth without psychopathology completed pre- and during pandemic assessments of anxiety and depression and COVID-related stress. Forty-six youth also completed a threat processing fMRI task pre-pandemic. Anxiety and depression significantly increased during the pandemic (all ps < 0.05). Significant symptom change was partially mediated by pandemic stress and worries. Increased prefrontal activity in response to neutral faces pre-pandemic was associated with more intense parent-reported anxiety during the pandemic (all Fs(1.95,81.86) > 14.44, ps < 0.001). The present work extends existing knowledge on the mediating role of psychological stress on symptoms of anxiety and depression in youth.

High-risk social drinkers and heavy drinkers display similar rates of alcohol consumption.

Alcohol consumption is often assessed over weeks to months, but few attempts have been made to characterize alcohol consumption rates at the level of an individual drinking session. Here, we aimed to compare the rate of alcohol consumption in social drinkers at high risk for alcohol use disorder (AUD) and heavy drinkers. One hundred and sixty social drinkers and 48 heavy drinkers participated in an alcohol self-administration study. Social drinkers were classified as low risk or high risk for AUD based on sex, impulsivity, and family history of alcoholism. Participants received a priming dose of intravenous alcohol to assess alcohol-induced craving and completed a 125-minute intravenous alcohol self-administration session to assess rate of achieving a binge-level exposure (blood alcohol concentration greater than or equal to 80 mg%). There were no differences between rates of binging in high-risk and heavy drinkers (hazard ratio = 0.87; 95% CI, 0.48-1.56). Heavy drinkers reported higher levels of craving than high-risk and low-risk drinkers at baseline. However, following a priming dose of alcohol, there were no longer differences in craving between high-risk and heavy drinkers. These results indicate that high-risk social drinkers demonstrate binging behavior that is similar to heavy drinkers, which may be driven by alcohol-induced craving. Prospective studies are needed to elucidate whether these patterns of craving and consumption in high-risk social drinkers are predictive of future AUD.

On Defining Irritability and its Relationship to Affective Traits and Social Interpretations.

Irritability has gained recognition as a clinically significant trait in youth and adults that when persistent and severe, predicts poor outcomes throughout life. However, its definition, measurement, and relationship to similar constructs remain poorly understood. In a community sample of adults (N=458; 19-74 years; M=40.5), we sought to identify a unitary irritability factor from independently constructed self-reported measures of irritability distinct from the related constructs of aggression, depression, and anxiety, and whether it was associated with face emotion identification deficits and hostile interpretation biases previously established in clinical pediatric samples. The three measures of irritability generated a common factor characterized by a rapid, angry response to provocation. This irritability factor had unique associations with tendencies to judge ambiguous stimuli as reflecting hostility, but not with face emotion identification performance. These findings clarify the nature of irritability and its associations with neurocognitive phenomenon.

Intraclass correlation: Improved modeling approaches and applications for neuroimaging.

Intraclass correlation (ICC) is a reliability metric that gauges similarity when, for example, entities are measured under similar, or even the same, well-controlled conditions, which in MRI applications include runs/sessions, twins, parent/child, scanners, sites, and so on. The popular definitions and interpretations of ICC are usually framed statistically under the conventional ANOVA platform. Here, we provide a comprehensive overview of ICC analysis in its prior usage in neuroimaging, and we show that the standard ANOVA framework is often limited, rigid, and inflexible in modeling capabilities. These intrinsic limitations motivate several improvements. Specifically, we start with the conventional ICC model under the ANOVA platform, and extend it along two dimensions: first, fixing the failure in ICC estimation when negative values occur under degenerative circumstance, and second, incorporating precision information of effect estimates into the ICC model. These endeavors lead to four modeling strategies: linear mixed-effects (LME), regularized mixed-effects (RME), multilevel mixed-effects (MME), and regularized multilevel mixed-effects (RMME). Compared to ANOVA, each of these four models directly provides estimates for fixed effects and their statistical significances, in addition to the ICC estimate. These new modeling approaches can also accommodate missing data and fixed effects for confounding variables. More importantly, we show that the MME and RMME approaches offer more accurate characterization and decomposition among the variance components, leading to more robust ICC computation. Based on these theoretical considerations and model performance comparisons with a real experimental dataset, we offer the following general-purpose recommendations. First, ICC estimation through MME or RMME is preferable when precision information (i.e., weights that more accurately allocate the variances in the data) is available for the effect estimate; when precision information is unavailable, ICC estimation through LME or the RME is the preferred option. Second, even though the absolute agreement version, ICC(2,1), is presently more popular in the field, the consistency version, ICC(3,1), is a practical and informative choice for whole-brain ICC analysis that achieves a well-balanced compromise when all potential fixed effects are accounted for. Third, approaches for clear, meaningful, and useful result reporting in ICC analysis are discussed. All models, ICC formulations, and related statistical testing methods have been implemented in an open source program 3dICC, which is publicly available as part of the AFNI suite. Even though our work here focuses on the whole-brain level, the modeling strategy and recommendations can be equivalently applied to other situations such as voxel, region, and network levels.

Early-Childhood Social Reticence Predicts Brain Function in Preadolescent Youths During Distinct Forms of Peer Evaluation.

Social reticence is expressed as shy, anxiously avoidant behavior in early childhood. With development, overt signs of social reticence may diminish but could still manifest themselves in neural responses to peers. We obtained measures of social reticence across 2 to 7 years of age. At age 11, preadolescents previously characterized as high (n = 30) or low (n = 23) in social reticence completed a novel functional-MRI-based peer-interaction task that quantifies neural responses to the anticipation and receipt of distinct forms of social evaluation. High (but not low) social reticence in early childhood predicted greater activity in dorsal anterior cingulate cortex and left and right insula, brain regions implicated in processing salience and distress, when participants anticipated unpredictable compared with predictable feedback. High social reticence was also associated with negative functional connectivity between insula and ventromedial prefrontal cortex, a region commonly implicated in affect regulation. Finally, among participants with high social reticence, negative evaluation was associated with increased amygdala activity, but only during feedback from unpredictable peers.

Irritability in child and adolescent anxiety disorders.

BACKGROUND: Our objective was to compare self- and parent-reported irritability in youths with anxiety disorders, healthy youths, and those with mood disorders characterized by irritability. Irritability is a common but relatively understudied psychiatric symptom in child and adolescent anxiety disorders. In anxious youths, little is known about the severity of irritability, its impact on functioning, or the effect of informant source on reports of irritability. METHODS: We compared parent- and self-report forms of the Affective Reactivity Index (ARI), a validated measure of irritability, in youths ages 8-17 years with no psychopathology (healthy comparison, HC; n = 38), anxiety disorders (ANX; n = 42), bipolar disorder (BD; n = 35), or severe mood dysregulation (SMD; n = 61; a phenotype characterized by chronic, severely impairing irritability). RESULTS: Irritability was significantly higher in ANX than HC youths by both parent and self-report (partial η(2) = 0.24 and 0.22, respectively, P's < 0.001). Informant effects differed among ANX, BD, and SMD. Overall, parent-reported irritability was higher in BD with comorbid anxiety disorders and SMD with or without comorbid anxiety disorders than ANX (P's < 0.007), but self-reported irritability was not significantly different among the three patient groups. DISCUSSION: By both parent and self-report, youths with anxiety disorders exhibit significantly more irritability and associated impairment than healthy subjects. Self-reported irritability in youths with anxiety disorders is comparable to that observed in youths with severe mood disorders, although parental reports of irritability differ among the disorders. Future research should examine the pathophysiology of anxiety-associated irritability, as well as its prognostic and treatment implications.

Attention bias to threat faces in severe mood dysregulation.

BACKGROUND: We used a dot-probe paradigm to examine attention bias toward threat (i.e., angry) and happy face stimuli in severe mood dysregulation (SMD) versus healthy comparison (HC) youth. The tendency to allocate attention to threat is well established in anxiety and other disorders of negative affect. SMD is characterized by the negative affect of irritability, and longitudinal studies suggest childhood irritability predicts adult anxiety and depression. Therefore, it is important to study pathophysiologic connections between irritability and anxiety disorders. METHODS: SMD patients (N = 74) and HC youth (N = 42) completed a visual probe paradigm to assess attention bias to emotional faces. Diagnostic interviews were conducted and measures of irritability and anxiety were obtained in patients. RESULTS: SMD youth differed from HC youth in having a bias toward threatening faces (P < .01). Threat bias was positively correlated with the severity of the SMD syndrome and depressive symptoms; degree of threat bias did not differ between SMD youth with and without co-occurring anxiety disorders or depression. SMD and HC youth did not differ in bias toward or away from happy faces. CONCLUSIONS: SMD youth demonstrate an attention bias toward threat, with greater threat bias associated with higher levels of SMD symptom severity. Our findings suggest that irritability may share a pathophysiological link with anxiety and depressive disorders. This finding suggests the value of exploring further whether attention bias modification treatments that are effective for anxiety are also helpful in the treatment of irritability.

Editorial: The EDI-YC: A New Measure of Emotion Dysregulation in Young Children.

Although emotion dysregulation is at a peak during the preschool age (2-5 years) and clinically significant dysregulation has impact across the lifespan, there are surprisingly few ways to measure it in this age group. This is especially true for groups of children in whom emotion may be particularly dysregulated, such as those with autism spectrum disorder. The modern, rigorous development of a well-grounded measure has profound clinical implications. Practically, it gives a common reference for the severity of a clinical issue, which is the foundation of measurement-based care and quantitative research. Theoretically, the process also defines the issue involving the scale designers, those whom the scale is about, and even the scale users as the measure is used and refined over years. Measures of preschool emotion dysregulation will allow us to better map its trajectory from preschool throughout the lifespan. In this issue, Day and Mazefsky et al.1 rigorously extend the Emotion Dysregulation Inventory (EDI; a set of questionnaires) to 2 groups of preschoolers, those representing children with neurodevelopmental issues, especially autism, and those without.

Chronotype and emotion processing: a pilot study testing timing of online cognitive bias modification training.

BACKGROUND: Circadian rhythms influence cognitive performance which peaks in the morning for early chronotypes and evening for late chronotypes. It is unknown whether cognitive interventions are susceptible to such synchrony effects and could be optimised at certain times-of-day. OBJECTIVE: A pilot study testing whether the effectiveness of cognitive bias modification (CBM) for facial emotion processing was improved when delivered at a time-of-day that was synchronised to chronotype. METHODS: 173 healthy young adults (aged 18-25) with an early or late chronotype completed one online session of CBM training in either the morning (06:00 hours to 10:00 hours) or evening (18:00 hours to 22:00 hours). FINDINGS: Moderate evidence that participants learnt better (higher post-training balance point) when they completed CBM training in the synchronous (evening for late chronotypes, morning for early chronotypes) compared with asynchronous (morning for late chronotypes, evening for early chronotypes) condition, controlling for pre-training balance point, sleep quality and negative affect. There was also a group×condition interaction where late chronotypes learnt faster and more effectively in synchronous versus asynchronous conditions. CONCLUSIONS: Preliminary evidence that synchrony effects apply to this psychological intervention. Tailoring the delivery timing of CBM training to chronotype may optimise its effectiveness. This may be particularly important for late chronotypes who were less able to adapt to non-optimal times-of-day, possibly because they experience more social jetlag. CLINICAL IMPLICATIONS: To consider delivery timing of CBM training when administering to early and late chronotypes. This may generalise to other psychological interventions and be relevant for online interventions where the timing can be flexible.

Differentiating neural sensitivity and bias during face-emotion processing in youth: a computational approach.

The ability to interpret face-emotion displays is critical for the development of adaptive social interactions. Using a novel variant of a computational model and fMRI data, we examined behavioral and neural associations between two metrics of face-emotion labeling (sensitivity and bias) and age in youth. Youth and adults (n = 44, M age = 20.02, s.d. = 7.44, range = 8-36) completed an explicit face-emotion labeling fMRI task including happy to angry morphed face emotions. A drift-diffusion model was applied to choice and reaction time distributions to examine sensitivity and bias in interpreting face emotions. Model fit and reliability of parameters were assessed on adult data (n = 42). Linear and quadratic slopes modeled brain activity associated with dimensions of face-emotion valence and ambiguity during interpretation. Behaviorally, age was associated with sensitivity. The bilateral anterior insula exhibited a more pronounced neural response to ambiguity with older age. Associations between sensitivity and bias metrics and activation patterns indicated that systems encoding face-emotion valence and ambiguity both contribute to the ability to discriminate face emotions. The current study provides evidence for age-related improvement in perceptual sensitivity to facial affect across adolescence and young adulthood.

Irritability in Youths: A Critical Integrative Review.

Irritability, defined as proneness to anger that may impair an individual's functioning, is common in youths. There has been a recent upsurge in relevant research. The authors combine systematic and narrative review approaches to integrate the latest clinical and translational findings and provide suggestions for addressing research gaps. Clinicians and researchers should assess irritability routinely, and specific assessment tools are now available. Informant effects are prominent, are stable, and vary by age and gender. The prevalence of irritability is particularly high among individuals with attention deficit hyperactivity disorder, autism spectrum disorder, and mood and anxiety disorders. Irritability is associated with impairment and suicidality risk independent of co-occurring diagnoses. Developmental trajectories of irritability (which may begin early in life) have been identified and are differentially associated with clinical outcomes. Youth irritability is associated with increased risk of anxiety, depression, behavioral problems, and suicidality later in life. Irritability is moderately heritable, and genetic associations differ based on age and comorbid illnesses. Parent management training is effective for treating psychological problems related to irritability, but its efficacy in treating irritability should be tested rigorously, as should novel mechanism-informed interventions (e.g., those targeting exposure to frustration). Associations between irritability and suicidality and the impact of cultural context are important, underresearched topics. Analyses of large, diverse longitudinal samples that extend into adulthood are needed. Data from both animal and human research indicate that aberrant responses to frustration and threat are central to the pathophysiology of irritability, revealing important translational opportunities.

The developmental psychopathology of irritability.

Chronic, severe irritability is common in childhood and is very impairing. Furthermore, childhood irritability predicts suicidality, social impairment, and depressive and anxiety disorders in adulthood. Focusing on both normative and pathologic development, we review the construct of irritability from its origins in aggression and disruptive behavior research to its contemporary relevance for affective psychopathology. We then describe two broad neurocognitive systems that show promise in differentiating irritable from nonirritable youths: aberrant processing of emotional stimuli and impaired context-sensitive regulation. We suggest behavioral, neurocognitive, and physiologic measures that may aid in studying severe irritability and assessing its therapeutics. Finally, we argue for therapeutic trials targeting severe irritability that address emotional aspects of irritability in addition to the associated disruptive behavior.

Editorial: The Neural Systems of Fairness, Retaliation, and Aggression.

"Not fair!" is a call to action regularly heard by parents and mental health professionals. It is well known that a person's perception of fairness might provoke them to anger and aggression.1 Beyond common experience, all sorts of experiments involving people's responses to rigged, interactive games have proved the point. Indeed, it is not just people; de Waal2 charmed the world with a TED talk in which monkeys took umbrage and aggressed in response to unfairness. Knowing this, Mathur et al.3 used unfairness and retaliation to illuminate the complex neural circuity of aggression in adolescents.

A Computational Model Reveals Learning Dynamics During Interpretation Bias Training With Clinical Applications.

BACKGROUND: Some psychopathologies, including anxiety and irritability, are associated with biases when judging ambiguous social stimuli. Interventions targeting these biases, or interpretation bias training (IBT), are amenable to computational modeling to describe their associative learning mechanisms. Here, we translated ALCOVE (attention learning covering map), a model of category learning, to describe learning in youths with affective psychopathology when training on more positive judgments of ambiguous face emotions. METHODS: A predominantly clinical sample comprised 71 youths (age range, 8-22 years) representing broad distributions of irritability and anxiety symptoms. Of these, 63 youths were included in the test sample by completing an IBT task with acceptable performance for computational modeling. We used a separate sample of 28 youths to translate ALCOVE for individual estimates of learning rate and generalization. In the test sample, we assessed associations between model learning estimates and irritability, anxiety, their shared variance (negative affectivity), and age. RESULTS: Age and affective symptoms were associated with category learning during IBT. Lower learning rates were associated with higher negative affectivity common in anxiety and irritability. Lower generalization, or improved discrimination between face emotions, was associated with increasing age. CONCLUSIONS: This work demonstrates a functional consequence of age- and symptom-related learning during interpretation bias. Learning measured by ALCOVE also revealed learning types not accounted for in the prior literature on IBT. This work more broadly demonstrates the utility of measurement models for understanding trial-by-trial processes and identifying individual learning styles.