A Light of Our Own

The sensory world presents an abundant and continuous stream of information. This stream, rushing towards our senses, contains more than can be processed with the finite resources of our brain. We require a means to select and prioritize certain information for enhanced processing; this is the role of attention. In this thesis, we zoom in on one sense in particular: vision. We investigate how attention-driven components and stimulus-driven components, independently and through their interactions, influence human visual processing. We used attention field models to structure this inquiry. Attention field models can implement additive constants for practical purposes (e.g. to avoid division by zero). Through simulations, we probed how seemingly trivial constants can result in profoundly different qualitative predictions, including whether and which amplitudes matter. Beyond simulations, we conducted multiple fMRI experimental studies to investigate components of attention field models. We observed that the precision of attention influences the attraction of population receptive fields (pRFs) towards the attended locus, with narrow attention resulting in stronger attraction. In early visual areas, where receptive fields are small, we further observed local attention effects, observed as attraction of only receptive fields close to the attended locus. This phenomenon was better explained by an attention field model with an offset on the attention field. Next, we investigated the effects of amplitude on attraction of pRFs towards the attended locus. We modulated stimulus-driven amplitude, using stimulus contrast, and observed that attraction towards the attended locus was unaffected by contrast. We further explored whether stimulus contrast altered population receptive field properties, including surround suppression. We used a difference of Gaussians model to probe the effects of stimulus contrast on population receptive field properties. Finally, we explored the interaction between stimulus contrast and attentional precision across different conditions. We found differences in these interactions depending on whether the stimulus was attended, was presented within the spatial extent of attention, and whether the stimulus shared features with an attended stimulus. This body of work sought to clarify how attention dynamically alters population receptive field properties and is influenced by stimulus-driven properties.