Anhedonia and Anxious Arousal Are Associated With Distinct Expectations About the Statistics of a Volatile...

The brain is believed to construct a generative model of the environment, updated via differences between predictions and incoming sensory evidence (i.e., prediction errors). Uncertainty can arise at multiple levels, including when estimating the environment's causal structure (stochasticity; first order), the volatility of this causal structure (second order), and volatility of this volatility (meta-volatility; higher order). Altered estimation of first-order uncertainty is implicated in anxiety and depression, including inflated certainty about the probability of negative events, but the role of second- and higher-order uncertainty estimation in internalizing disorders is less clear.

In a crowdsourced community sample (N = 287), we examined associations between depressive and anxiety symptoms and beliefs about volatility derived with a hierarchical Bayesian model of inference applied to a rewarding and aversive probabilistic reversal-learning task.

Anxious arousal was linked to elevated initial expectations of volatility in an aversive context and greater adjustment of behavior after avoiding a loss. Meanwhile, anhedonia was associated with reduced initial expectations of volatility, relatively less stable volatility beliefs in reward versus loss contexts early in learning (indexed by higher meta-volatility), and reduced exploratory behavior in a loss environment.

Together, the results indicate distinct belief signatures of anxiety and depression in a volatile environment. Elevated expectations of the volatility of aversive contexts could increase hypervigilance, a key anxiety symptom, whereas a belief that the environment is stagnant could increase hopelessness and anhedonia, markers of depression. Distinct alterations in estimating volatility could present candidate mechanisms and treatment targets for anxiety and depression.