LEARNING AND MOTIVATIONAL PROCESSES OF VALUE-BASED DECISION-MAKING A behavioural, neuroscientific and computational approach to Pavlovian bias

Pavlovian biases influence decision-making by coupling reward seeking with the predictive and motivational signals embedded in environmental cues. Although a wide range of Pavlovian biases has been documented over the past decades, and several theories have been proposed to explain them, the neural and mechanistic pathways through which these biases shape behaviour remain poorly understood. Across a series of behavioural, neurophysiological and computational studies, this thesis characterises how Pavlovian cues bias action selection (i.e., outcome-specific Pavlovian-to-Instrumental Transfer, or PIT) in humans. In the first experiment, electroencephalographic (EEG) data provide direct support for the hypothesis that Pavlovian cues convey sensorimotor information about predicted outcomes, thereby enabling outcome-specific PIT. Subsequent experiments show that such cue-driven influence can distort otherwise optimal decision strategies, revealing a suboptimal component of PIT whose magnitude depends on cue-outcome predictivity and varies according to individuals’ Pavlovian learning style. Eye-tracking, pupillometry and advanced computational modelling further indicate that slower Pavlovian updating in sign-trackers underlies their heightened and persistent maladaptive Pavlovian bias. Finally, additional EEG evidence demonstrates that this suboptimal form of outcome-specific PIT operates through a disruption of premotor activity that ordinarily supports adaptive choice preparation. Taken together, these findings push the boundaries of current models of motivated decision-making and clarify when Pavlovian control supports adaptive choice and when it becomes a source of maladaptive bias.