Dimensional psychiatry: mental disorders as dysfunctions of basic learning mechanisms

J Neural Transm (Vienna). 2016 Aug;123(8):809-21. doi: 10.1007/s00702-016-1561-2. Epub 2016 May 4.


It has been questioned that the more than 300 mental disorders currently listed in international disease classification systems all have a distinct neurobiological correlate. Here, we support the idea that basic dimensions of mental dysfunctions, such as alterations in reinforcement learning, can be identified, which interact with individual vulnerability and psychosocial stress factors and, thus, contribute to syndromes of distress across traditional nosological boundaries. We further suggest that computational modeling of learning behavior can help to identify specific alterations in reinforcement-based decision-making and their associated neurobiological correlates. For example, attribution of salience to drug-related cues associated with dopamine dysfunction in addiction can increase habitual decision-making via promotion of Pavlovian-to-instrumental transfer as indicated by computational modeling of the effect of Pavlovian-conditioned stimuli (here affectively positive or alcohol-related cues) on instrumental approach and avoidance behavior. In schizophrenia, reward prediction errors can be modeled computationally and associated with functional brain activation, thus revealing reduced encoding of such learning signals in the ventral striatum and compensatory activation in the frontal cortex. With respect to negative mood states, it has been shown that both reduced functional activation of the ventral striatum elicited by reward-predicting stimuli and stress-associated activation of the hypothalamic-pituitary-adrenal axis in interaction with reduced serotonin transporter availability and increased amygdala activation by aversive cues contribute to clinical depression; altogether these observations support the notion that basic learning mechanisms, such as Pavlovian and instrumental conditioning and Pavlovian-to-instrumental transfer, represent a basic dimension of mental disorders that can be mechanistically characterized using computational modeling and associated with specific clinical syndromes across established nosological boundaries. Instead of pursuing a narrow focus on single disorders defined by clinical tradition, we suggest that neurobiological research should focus on such basic dimensions, which can be studied in and compared among several mental disorders.

Keywords: Alcohol dependence; Computational neuroscience; Depression; Dopamine; PIT; Pavlovian-to-instrumental transfer; Reward learning; Schizophrenia; Sequential learning; Serotonin; Stress; Two-step; Vulnerability.

Publication types

  • Review

MeSH terms

  • Brain / physiology*
  • Conditioning, Psychological
  • Decision Making
  • Humans
  • Learning Disabilities / complications*
  • Mental Disorders / complications*
  • Mental Disorders / psychology*
  • Reinforcement, Psychology*