Probabilistic Learning by Reinforcement and Approach of Reward Versus Avoidance of Punishment in Generalized Anxiety Disorder

Open Access
Lafreniere, Lucas S
Graduate Program:
Master of Science
Document Type:
Master Thesis
Date of Defense:
June 13, 2016
Committee Members:
  • Michelle Newman, Ph.D., Thesis Advisor/Co-Advisor
  • Stephen Wilson, Ph.D., Committee Member
  • Richard Carlson, Ph.D., Committee Member
  • generalized anxiety disorder
  • GAD
  • cognitive bias
  • learning
  • operant conditioning
  • reinforcement
  • punishment
  • reward
  • probabilistic learning
  • implicit
  • explicit
  • Probabilistic Selection Task.
  • Probabilistic Selection Task
Worry research suggests that those with GAD may bear differential learning tendencies compared to controls, such as greater learning by 1) negative reinforcement over positive reinforcement and 2) avoidance of punishment over approach of rewards. The current study used computerized implicit cognitive tasks with probabilistic reinforcement to determine if such differences exist. 157 participants, 59 GAD and 98 non-GAD, took a modified version of the Probabilistic Learning Task (PLT) to study negative vs. positive reinforcement learning and the standard Probabilistic Selection Task (PST) to study approach vs. reward learning. In our specially designed PLT, participants chose between stimuli with specific probabilities of reinforcement, learning over time which of each pair had the highest probability. Correct choices in the negative condition removed an angry face, whereas those in the positive condition made a happy face appear. The unaltered, classic PST design was used. Results showed that compared to non-GADs, those with GAD learned at a significantly slower rate over time and to a lesser degree on both probabilistic tasks both implicitly and explicitly, regardless of condition. Marginally significant explicit learning effects suggested that those with GAD had poorer learning via positive reinforcement than did non-GADs, especially when reward probability was high. No significant findings arose regarding avoidance vs. approach learning. Reaction time differences were minimal to absent. Those with GAD may have deficits in probabilistic learning, facilitating the use of unlikely fearful expectations to buffer against negative emotional shifts. They may also learn worse by positive reinforcement than those without GAD.