Psychiatry - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemThe Brain-Behavioural Basis of Human Safety Learning: An investigation of Pavlovian conditioned inhibition( 2022)Safety learning allows individuals to associate stimuli with the absence of threat, thus conferring the ability to suppress fear and anxiety in safe situations, and by consequence, maintain psychological and physiological well-being. Disrupted safety learning is thought to be a key component of anxiety-related disorders, but as yet, the basic mechanisms of safety learning remain incompletely understood and lack a formal theoretical definition. Across two studies of healthy adults, this thesis sought to examine the behavioural (Study 1) and neural (Study 2) basis of safety learning in humans. Based on long-standing principles of associative learning theory, a novel iteration of the Pavlovian conditioned inhibition paradigm was developed and implemented in lab-based and 7-Tesla functional magnetic resonance imaging (fMRI) settings. Study 1 (N = 73) was an investigation of the behavioural aspects of safety learning, and moreover, sought to validate the utility of the Pavlovian conditioned inhibition paradigm as an experimental model for safety learning, as well as examining individual differences in trait measures of anxiety. This paradigm trained a robust safety signal (the conditioned inhibitor, X) which was conditioned by delivering threat (loud noise) to a conditioned stimulus on its own (A+), but omitting threat when that stimulus was presented in conjunction with the inhibitor (AX-). As a control cue, two stimuli were similarly unreinforced in compound, but neither was presented alone on other trials (BC-). The paradigm also controlled for several possible confounds, including the use of a safety signal as a control cue, rather than using a novel or neutral cue, among other factors. Both the control safety signal and the conditioned inhibitor were shown to inhibit physiological and cognitive threat responses at test, when paired with aversive conditioned stimuli. However, the inhibitor conferred a significantly greater degree of inhibition for cognitive threat responses, as measured by threat-expectancy ratings during a summation test. Further, trait anxiety was positively correlated with threat expectancy towards the inhibitor during learning, indicative of threat responses to safety signals, which are thought be a feature of maladaptive anxiety. Study 2 (N = 49) investigated the functional neural correlates of safety learning via conditioned inhibition. The same paradigm from Study 1 was adapted for use in neuroimaging, using ultra-high field fMRI. Activations were compared between the safety signals directly (AX vs BC), and learning-specific activation was assessed via contrasts between early and late conditioning trials, and conditioning phase activity versus test phase, under the hypothesis that this should identify regions recruited to form stimulus-safety associations when these contingencies are new and unfamiliar. It was found that conditioned inhibition involved activity across a distinct set of cortico-striatal regions, which aligned with known subcortical circuits of the basal ganglia. Further, though showing similar behavioural responses to the inhibitor, the standard safety signal evoked no subcortical engagement, and instead was associated with an expanse of cortical activity, consistent with regions observed in differential fear-safety processing. In total, these studies indicate that the framework of Pavlovian conditioned inhibition can serve as an experimental model for characterising safety learning in humans, with implications for clinical translational work. It suggests that robust safety learning occurs by way of expectancy violation, or in other words, that a stimulus acquires safety value by predicting the unexpected omission of threat, in line with the principles of formal learning theory. Further, though current human studies often emphasise the safety-learning roles of various higher prefrontal regions, Study 2 demonstrates that safety learn- ing engages several subcortical brain regions that are well-known for their involvement in other domains of reinforcement learning. I discuss the theoretical implications that this research has for defining safety through the lens of associative learning, the neurobiology of safety acquisition, and a basis for separating processes of safety acquisition and safety expression between associative subcortical systems and higher cortical brain regions respectively. Several future directions are proposed for the ongoing study of safety learning, including characterisation of safety prediction errors and testing hypotheses of deficient safety learning in psychiatric disorders.