A goal of cognitive neuroscience is to provide accounts of brain functions in terms of information processing. Vision has long been used as a beachhead for such approaches: information-processing principles first identified in the visual system, such as receptive fields and divisive normalization, have later been observed in a variety of other sensory and cognitive domains. Here, we formulated a visual-spatial population receptive field model based on divisive normalization, a candidate ‘canonical’ neural computation. We showed that the model unifies and outperforms existing models, and that it does so via local variations in its algorithmic parameters. Next, we hypothesised that neurotransmitter systems might provide the biological-implementational substrate underlying the model’s algorithmic modulations. We showed that model parameter estimates are related to density maps of different serotonin and GABA receptors. Finally, in order to directly probe the role of neurotransmitter systems in human visual computations, we administered psilocybin, a serotonin receptor agonist. We showed that psilocybin systematically alters normalization, and in particular the model’s parameter relevant for suppression (center-surround configurations). Our findings provide direct causal evidence of the involvement of neurotransmitter receptors in visual computations in the living human brain. In sum, this thesis brings together vision science with chemoarchitecture and neuropharmacology, providing new insights into the inner workings of the human visual system.
More information on the thesis