A behavioural model was utilised by researchers to better understand the origins of obsessive-compulsive disorder. They demonstrated that when learning parameters for reinforcement and punishment are significantly out of balance, the cycle of obsession and compulsion can be exacerbated. This study might lead to better mental health treatments.
Obsessive-compulsive disorder (OCD) might be viewed as a result of uneven learning between reward and punishment, according to researchers from Nara Institute of Science and Technology (NAIST), Advanced Telecommunications Research Institute International, and Tamagawa University.
They demonstrated that asymmetries in brain computations that relate present results to previous actions can lead to disordered behaviour using actual testing of their theoretical model.
This can occur when the memory trace signal for previous activities decays differentially for good and poor results. In this situation, “good” indicates that the outcome was better than predicted, while “bad” indicates that it was worse than expected. This research contributes to our understanding of how OCD develops.
OCD as mental condition
OCD is a mental condition characterised by intrusive and repetitive thoughts, known as obsessions, and specific repetitive activities, known as compulsions. Patients with OCD frequently feel helpless to modify their behaviour, even when they are aware that their obsessions or compulsions are unreasonable. In severe circumstances, these may leave the individual unable to lead a regular life.
Compulsive behaviours, such as frequent hand washing or constantly checking if doors are secured before leaving the house, are attempts to ease anxiety produced by obsessions. However, the mechanism by which the cycle of obsessions and compulsions is intensified was previously unknown.
NAIST researchers have now employed reinforcement learning theory to describe the chaotic cycle associated with OCD. In this approach, a better-than-expected event becomes more likely (positive prediction error), while a worse-than-expected outcome is suppressed (negative prediction error). Delays and positive/negative prediction mistakes must also be considered when using reinforcement learning.
This indicates that, when it comes to negative prediction mistakes, the view of previous activities is substantially smaller than when it comes to positive prediction errors. “With uneven trace decay variables (n+ > n-), our model successfully reflects the vicious loop of obsession and compulsion characteristic of OCD,” co-first authors Yuki Sakai and Yutaka Sakai explain.
The researchers put this hypothesis to the test by having 45 OCD sufferers and 168 healthy control volunteers play a computer-based game with monetary incentives and punishments. Patients with OCD had considerably lower n- compared to n+, as expected by OCD computational features. Furthermore, serotonin enhancers, which are first-line drugs for the treatment of OCD, adjusted the unbalanced setting of trace decay factors (n+ > n-).
“Although people believe we always make sensible judgments, our computer model shows that we occasionally unintentionally promote maladaptive behaviours,” explains Saori C. Tanaka, the corresponding author.
Although identifying treatment-resistant patients solely on clinical symptoms is currently problematic, this computational model predicts that individuals with significantly unbalanced trace decay variables may not respond to behavioural therapy alone. These findings might one day be utilised to predict which individuals would be resistant to behavioural therapy before treatment begins.