Abstract
Background
A wealth of data points to a role for deficient reward-seeking behavior in schizophrenia (SZ), as a mechanism of negative symptoms, in general, and avolition, in particular. Specifically, avolitional patients with SZ appear to experience reinforcers in a normative manner, but do not properly assign prospective value to rewards or actions that frequently lead to rewards. There are many forms of goal-directed behavior, however, with reward-seeking behavior being only one. A second kind of goal-directed behavior is exploration: seeking out information about contingencies in the environment, so as to reduce uncertainty about available options. Successful navigation of the explore/exploit trade-off is critical to adaptive behavior. In several different studies, we have examined information-seeking behavior in individuals with SZ and matched controls, hypothesizing that reductions in goal-directed exploration make substantial contribution to avolition.
Methods
We have used several different behavioral paradigms incorporating an explore/exploit trade-off in samples of SZ patients and matched controls. Two of these paradigms were forms of reversal learning tasks, where individuals learn to choose the most-frequently-rewarded of two or three stimuli, with reward contingencies changing after a certain number of consecutive correct choices. A third paradigm required participants to track the value of three slot machines across stable and volatile runs of trials (when reward contingencies changed suddenly, independent of performance). All of these paradigms were performed in conjunction with fMRI scanning, with single-subject MRI data being analyzed using parametric (amplitude-modulated) regressors constructed based on trial-wise estimates of expected value, certainty about value, and the degree to which choices were reward-seeking (choosing the slot machines with the highest EV) vs. information-seeking (driven by the exploration of unknown contingencies), from computational analyses.
Results
In the context of reversal learning tasks, we have found that SZ patients show: 1) aberrant MRI signals in dorsolateral prefrontal cortex (dlPFC) and frontopolar cortex (FPC) in conjunction with response-set-shifting behavior; 2) a reduced ability to adjust learning rates in accordance with the volatility of the learning environment; and 3) that this behavioral deficit is accompanied by reduced connectivity between dorsomedial prefrontal cortex (dmPFC) and ventral striatum (VS). In the context of the Slot Machines task, individuals with SZ made fewer exploitative choices in the volatile condition (t=2.13, p=0.038), and required more trials to switch to the new-best machine following a reversal (t=2.99, p=0.004). Analyses of correlations among clinical symptoms and fMRI activations in brain regions-of-interest (ROIs) revealed that greater levels of avolition severity were associated with reduced reward tracking-related activity in right orbitofrontal cortex (OFC; r=-.41, p=0.05) and dmPFC (r=.57, p=0.01).
Discussion
These results point to associations among behavioral and brain signals related to goal-directed learning and exploration and clinical measures of avolition in individuals along the psychosis spectrum. Specifically, they support a role for interactions among brain regions involved in reward-seeking behavior (VS and OFC), uncertainty processing (dmPFC), and information-seeking behavior (dlPFC and FPC) in motivated behavior and suggest alternative mechanisms for motivational deficits. Furthermore, even when reward-seeking behavior is intact, SZ patients may exhibit motivational deficits due to impairments in information-seeking behavior.