Parietal and prefrontal control of distinct components of attention

Abstract

In this thesis, we investigate behavioural mechanisms and neural substrates of distinct components of endogenous spatial attention. Endogenous attention facilitates neural processing of the selected stimulus through one of two component mechanisms: either by influencing perceptual sensitivity, i.e., enhancing the quality of sensory information processing of the selected stimulus, or by altering decisional bias, i.e. by enhancing the weight afforded to selected sensory stimuli in the downstream decision process. It is unclear whether sensitivity and bias components of endogenous attention are under the control of common, shared or dissociated neural mechanisms. Moreover, it is unclear how key regions in the frontoparietal network contribute to sensitivity versus bias control. Here, we characterise how sensitivity and bias are co-modulated during endogenous visuospatial attention using a probabilistically cued, multialternative task (endogenous Posner cueing task), by analysing behaviour with a novel multidimensional signal detection model. We demonstrate for the first time that the model successfully decouples sensitivity and bias from human behavioural data in this attention task. Then, using transcranial magnetic stimulation (TMS), in conjunction with the aforementioned task and model paradigms, we provide novel evidence for the causal contributions of the right hemispheric PPC and FEF towards sensitivity and bias control during endogenous attention. In the first study, we tested whether the effects of endogenous cueing on sensitivity and bias could be decoupled using a probabilistically-cued, five-alternative change detection task tested with n=37 participants. Multi-alternative tasks, with more than two response options, cannot be correctly analysed using a conventional one-dimensional signal detection model. Consequently, we used a novel multidimensional signal detection model the m-ADC model that can decouple and accurately estimate sensitivity and bias parameters in such tasks. After confirmin...