Understanding behavioral mechanisms of Pavlovian biases through task reversal paradigms

Abstract

Decision-making processes are often influenced by Pavlovian biases as they subtly guide our actions and reactions unconsciously (Raab & Hartley, 2020). These learning biases are acquired overtime through experiences and result in previously neutral stimuli becoming positively or negatively valanced. Pavlovian biases can elicit automatic yet emotional responses in a variety of contexts ranging from mundane life events to life-threatening scenarios (Chen et al., 2022). Often cognitive disorders associated with corrupted decision-making can stem from unchecked Pavlovian biases. Some of these disorders include Substance Use Disorders (SUDs), depression, addictions, chronic pain as well as eating disorders (Nees et al., 2015). Importantly, Pavlovian biases often work in collaboration with instrumental learning. Instrumental learning is defined as a form of learning where actions are modified based on the consequences that they produce (Balleine & Dickinson, 1998). This study consists of Experiment 1 (N = 90 healthy adults) as well as Experiment 2 (N = 90 healthy adults), designed to explore the interaction between Pavlovian biases and goal-directed instrumental learning by examining how each adapts to task reversal. Both experiments in this study have three key conditions: baseline, reversal of goal-directed instrumental learning and reversal of Pavlovian bias. We hypothesized that Pavlovian biases would be slow to adjust after reversal due to their reliance on inflexible learning mechanisms, whereas the more flexible, goal-directed instrumental learning system would adapt more quickly. In line with our hypothesis, we found that goal-directed instrumental learning adapted rapidly whereas Pavlovian biases initially presented hurdles in participants performance levels however, eventually demonstrated some flexibility which led participants to overcome them. This study presents evidence for flexibility demonstrated by Pavlovian biases in deterministic reversal paradigms and further propels the discourse on how these biases interact with goal-directed instrumental learning.