A neurocomputational investigation of Pavlovian-instrumental interactions in avoidance. (360G-Wellcome-101521_Z_13_Z)

This proposal has three aims. Firstly, I will characterise the neural computations underlying model-based and model-free avoidance using fMRI in healthy subjects. To do so, I will adapt an existing appetitive paradigm to test if the existing framework can capture human avoidance behaviour. I will seek convergent evidence by comparing a standard version to a version designed to shift individuals towards model-free behaviour, e.g. using outcome volatility. Secondly, I will delineate the nature of the interplay between individual differences in Pavlovian fear conditioning and model-based vs. model-free avoidance by using interleaved Pavlovian conditioning trials. I will then use experimental manipulations to move our understanding of this interaction beyond correlation. I will test if increasing the strength of conditioned Pavlovian fear causes subjects to rely more heavily on model-free avoidance and if biasing individuals towards model-free learning reciprocally increases the strength o f Pavlovian conditioned fear and resistance to extinction. Finally, I will use the models developed at NYU to test causal predictions regarding the role of Pavlovian-instrumental interactions in patients with generalised anxiety disorder and obsessive-compulsive disorder diagnoses. In doing so, I hope to formalise dimensions with well-characterised neurocomputational bases that may explain the familiality and co-morbidity across anxiety disorder diagnoses.