Florian Schlagenhauf

Background: Impaired control over alcohol consumption is a core feature of alcohol use disorder (AUD), yet its mechanisms remain unclear. Habit theories posit a shift from goal-directed to habitual control with repeated alcohol use, but human evidence is mixed and often lacks translational alignment with animal models. Recent accounts therefore call for paradigms assessing contingency sensitivity with disorder-relevant reinforcers. Methods: We adapted a free-operant contingency degradation task to compare goal-directed control when responding for a natural (juice) versus disorder-relevant reinforcer (alcohol). Eighty-one participants were tested, comprising 49 non-treatment-seeking individuals with AUD (31.22 ± 8.35 years old; 28 female) and 32 healthy control individuals (27.44 ± 6.71 years old; 22 female). We expected reduced goal-directed control in AUD, indexed by weaker response-rate and causality judgment adaptation to changed contingencies, particularly for alcohol. Results: Contrary to our hypothesis, goal-directed control was preserved in both groups, with behavioral responding and causality judgments remaining contingency-sensitive. Rather than a generalized deficit, a reinforcer-specific attenuation emerged: across groups, behavioral contingency sensitivity was selectively reduced when responding for alcohol. Explicit contingency judgments remained intact and were stronger in individuals with AUD than in healthy control individuals. Conclusion: These findings do not support a generalized deficit in goal-directed control in AUD. Instead, goal-directed control appears selectively attenuated in alcohol-related contexts across groups, potentially reflecting reinforcement-history-dependent reductions in contingency-sensitive responding. Alcohol-specific reductions in goal-directed control may emerge early in alcohol use, underscoring the importance of reinforcer context.

The successor representation (SR) is a reinforcement learning strategy which has been suggested to underlie nuanced forms of habitual behavior, rendering it a candidate mechanism for mental health conditions thought to involve habitual control, such as addiction. However, paradigms used to assess habits in humans cannot detect SR use. Together with the fact that they are usually conducted in disorder-irrelevant contexts, this may explain why the evidence for a hypothesized shift from goal-directed to habitual behavior in addiction is inconsistent. Here, we tested whether individuals with high-risk (n=260) compared to low-risk drinking behavior (n=260) rely more on SR or a rigid goal-based dimension-reduced SR variant, particularly in alcohol-related contexts. An online multi-stage decision-making task performed in either a virtual bar or apartment context allowed us to dissociate (red)SR from model-based and model-free learning strategies by testing how well participants adapted their choices to different types of distal changes in the environment. Findings suggest that a particular type of SR, namely a (reduced) random-policy SR-like strategy, contributes to human behavior, but that high-risk drinkers do not differ from low-risk drinkers in their use of this strategy. Instead, both groups seem to rely less on what is likely a (reduced) random-policy SR strategy in alcohol-related contexts – and instead more on model-free control. Results suggest that the use of an adaptive, resource-efficient decision-making strategy is selectively downregulated in substance-related contexts. This study highlights the importance of considering contextual modulation and nuanced decision strategies in understanding habitual and goal-directed behavior in mental health conditions such as addiction.

Altered neural reward prediction error (RPE)-encoding during learning from drug reinforcers compared to non-drug rewards is at the core of learning theories of addiction. They predict stronger RPE-related fronto-striatal activation when responding for alcohol-related compared to natural rewards, particularly in individuals with alcohol use disorder (AUD). However, research in AUD has largely relied on paradigms that used monetary reinforcers or did not include a learning component. Here, we developed an instrumental learning task (ILT) where correct choices were reinforced by alcohol or juice taste in separate blocks. Twenty-eight individuals with predominantly mild to moderate AUD and 28 healthy control participants (HC) performed the ILT during functional magnetic resonance imaging. Alcohol taste was perceived as less pleasant and elicited less craving than juice taste across groups, and this was driven by HC. Both groups, but particularly individuals with AUD, successfully learned to respond for both reinforcer types, and individuals with AUD, but not HC, became faster over time in alcohol blocks. However, we found no differences in fronto-striatal RPE-encoding between groups and reinforcer types in the expected direction, but instead an opposite effect in dorsolateral prefrontal cortex. While we demonstrate that instrumental learning from alcohol taste elicits RPE-related fronto-striatal activation in AUD and HC, our findings do not support stronger fronto-striatal RPE-encoding when learning from alcohol-related compared to natural reinforcers in individuals with AUD than HC. Future learning research in addiction should take further steps to increase ecological validity by considering the context-dependence of drug-related reinforcers’ rewarding properties.

BackgroundImpaired control over alcohol consumption is a core feature of alcohol use disorder (AUD), yet its mechanisms remain unclear. Habit theories posit a shift from goal-directed to habitual control with repeated alcohol use, but human evidence is mixed and often lacks translational alignment with animal models. Recent accounts therefore call for paradigms assessing contingency sensitivity with disorder-relevant reinforcers.MethodsWe adapted a free-operant contingency degradation task to compare goal-directed control when responding for a natural (juice) versus disorder-relevant reinforcer (alcohol). Eighty-one participants were tested, comprising 49 non-treatment-seeking individuals with AUD (31.22 ± 8.35 years; 28 females) and 32 healthy controls (27.44 ± 6.71 years; 22 females). We expected reduced goal-directed control in AUD, indexed by weaker response-rate and causality judgment adaption to changed contingencies, particularly for alcohol.ResultsContrary to our hypothesis, goal-directed control was preserved in both groups, with behavioral responding and causality judgments remaining contingency-sensitive. Rather than a generalized deficit, a reinforcer-specific attenuation emerged: across groups, behavioral sensitivity was selectively reduced when responding for alcohol. Explicit contingency judgments remained intact and were stronger in individuals with AUD than in healthy controls.ConclusionThese findings do not support a generalized deficit in goal-directed control in AUD. Instead, goal-directed control appears selectively attenuated in alcohol-related contexts across diagnostic groups, potentially reflecting reinforcement-history-dependent shifts toward cue-driven responding. Alcohol-specific reductions in goal-directed control may emerge early in alcohol use, underscoring the importance of reinforcer context.

Zusammenfassung: Computational Psychiatry ist ein junges Forschungsfeld, in dem computationale Methoden auf psychiatrische Forschungsfragen angewendet werden. Verhalten und neurobiologische Prozesse können durch generative Modelle beschrieben werden, mit dem Ziel, psychopathologische Symptome besser zu verstehen und über mechanistische Biomarker den Weg zu Präzisionsmedizin und personalisierten Therapien zu eröffnen.

A shift away from goal-directed, model-based behavior is commonly viewed to characterize alcohol use disorder (AUD). Previous research, however, has failed to demonstrate differences between individuals with and without AUD regarding goal-directed control, operationalized as model-based behavior. Instead, findings suggest associations between model-based behavior and alcohol consumption patterns, but mechanistic insights into the link between model-based behavioral and neural signatures and longitudinal, real-life control over alcohol intake remain elusive. Here, we investigated whether experimentally assessed model-based behavior can prospectively predict intentional reduction of alcohol consumption in daily life. Therefore, we related behavioral and neural markers of model-based behavior during a sequential decision-making task in 67 participants with AUD (20 women) to long-term smartphone-based ecological momentary assessments of daily alcohol intake and weekly alcohol consumption intentions over a period of up to one year. Model-based behavior and its neural signatures in bilateral hippocampus and ventral striatum moderated how well individuals succeeded in aligning their alcohol consumption with their drinking intentions during the following year. Specifically, AUD participants with higher model-based behavior and associated stronger hippocampal and weaker ventral striatal learning signals exhibited enhanced capacity to intentionally reduce their alcohol consumption in everyday life. These findings provide evidence for the ecological validity of computational concepts of goal-directed behavior and suggest specific treatment targets for individually tailored interventions to regain control over alcohol use.

Psychotic beliefs are typically held with high certainty. Altered computation of uncertainty about a belief and about environmental dynamics may be an underlying mechanism of psychotic symptoms. We set out to shed light on behavioural and neural correlates of uncertainty processing and how it drives belief updating in psychosis. This cross-sectional study included 19 participants with psychosis spectrum disorder (5 female and 14 male) and 40 healthy control participants (21 female and 19 male) between 18 and 65 years of age. Participants performed a predictive inference task that required belief updating of a noisy outcome in a suddenly changing environment during functional magnetic resonance imaging. Behavioural and imaging data were analysed with a computational model that approximates an ideal Bayesian observer. The model expects beliefs to be updated based on the relative belief uncertainty and environmental change point probability. Task performance, model parameters and associated neural activation were compared between groups and associated with self-reported delusional ideation and cognitive functioning. While the belief updating speed overall did not differ between groups, the psychosis group showed lower task performance. Lower performance was associated with higher self-reported delusional ideation, even when controlling for cognitive functioning. Persons with psychosis spectrum disorder tended to persevere on beliefs after large prediction errors that signal environmental changes. They informed belief updates less by the probability of environmental change points, although this capacity seemed to depend on general cognitive functioning. The psychosis group also encoded the change point probability less in the superior occipital and fusiform gyrus, as well as a cluster comprising pre-central to middle frontal gyrus. Activity in these clusters was associated with lower self-reported delusional ideation across the whole sample and lower general and negative symptoms in the clinical sample. Persons with psychosis spectrum disorder did not seem to overestimate environmental volatility in general. Instead, they showed altered processing of information that occurred after environmental change points, whose probability was less well represented in brain regions encoding visual surprise and motor responses. Possibly, persons with psychosis spectrum disorder inadequately integrated visual surprise signals, leading to ineffective transmission to motor regions that eventually guide behaviour. Summarizing, our study suggests that delusions could result from a tendency to stick to old beliefs even in the light of contrary evidence, due to a failure to integrate uncertainty information based on inferred environmental dynamics.

Excessive alcohol consumption remains a global public health crisis, with millions suffering from alcohol use disorder (AUD, or simply “alcoholism”), leading to significantly reduced life expectancy. This review examines the interplay between habitual and goal-directed behaviors and the associated neurobiological changes induced by chronic alcohol exposure. Contrary to a strict habit-goal dichotomy, our meta-analysis of the published animal experiments combined with a review of human studies reveals a nuanced transition between these behavioral control systems, emphasizing the need for refined terminology to capture the probabilistic nature of decision biases in individuals with a history of chronic alcohol exposure. Furthermore, we distinguish habitual responding from compulsivity, viewing them as separate entities with diverse roles throughout the stages of the addiction cycle. By addressing species-specific differences and translational challenges in habit research, we provide insights to enhance future investigations and inform strategies for combatting AUD.

Binge drinking behavior in early adulthood can be predicted from brain structure during early adolescence with an accuracy of above 70%. We investigated whether this accurate prospective prediction of alcohol misuse behavior can be explained by psychometric variables such as personality traits or mental health comorbidities in a data-driven approach. We analyzed a subset of adolescents who did not have any prior binge drinking experience at age 14 (IMAGEN dataset, n = 555, 52.61% female). Participants underwent sMRI at age 14, binge drinking assessments at ages 14 and 22, and psychometric questionnaire assessments at ages 14 and 22. We derived structural brain features from T1-weighted magnetic resonance and diffusion tensor imaging. Using Machine Learning (ML), we predicted binge drinking (age 22) from brain structure (age 14) and used counterbalancing with oversampling to systematically control for 110+ variables from a wide range of social, personality, and other psychometric characteristics potentially associated with binge drinking. We evaluated if controlling for any variable resulted in a significant reduction in ML prediction accuracy.Sensation-seeking (-13.98±1.68%) assessed via the Substance Use Risk Profile Scale at age 14 and uncontrolled eating (-13.98±3.28%) assessed via the Three-Factor-Eating-Questionnaire at age 22 led to significant reductions in ML prediction accuracy upon controlling for them. Thus, sensation-seeking and binge eating could partially explain the prediction of future binge drinking from adolescent brain structure.Our findings suggest that binge drinking and binge eating at age 22 share common neurobiological precursors discovered by the ML model. These neurobiological precursors seem to be associated with sensation-seeking at age 14. Our results facilitate early detection of increased risk for binge drinking and inform future clinical research in trans-diagnostic prevention approaches for adolescent alcohol misuse.