Zooming in with attention: Neural and perceptual consequences of attention

Abstract

Although vision is the dominant sense in humans, the brain's capacity to process visual information is severely limited. We use visual attention to dynamically allocate the limited processing capacity to the most relevant information at any given time. Doing so, we prioritize its processing by concentrating neural resources on the attended information Attentional selection of visual information can be based on the spatial location of information (spatial attention) or based on a visual feature, like color, orientation or numerosity, i.e. the amount of a items in a set (feature attention). A recent computational model of attention, the attention field model, predicts that attentional selection acts both like a spotlight by increasing neural responses to the attended information and like a magnifying glass by enlarging the neural representation of the attended information. However, although the spotlight concept has received sufficient empirical support, the empirical evidence supporting the concept of attention as a magnifying glass in humans is currently limited. In this thesis we investigate attentional magnification in humans. In Chapter 2 we use spatial attention to demonstrate that it indeed acts like a magnifying glass on visual information. We show that spatial attention attracts population receptive fields (pRFs; Dumoulin and Wandell, 2008) towards its location across the entire visual field and hierarchy. Doing so, more receptive fields cover the attended information, which enlarges its neural representation. Furthermore, the results from this chapter also raise the possibility that spatial attention acts like a constant influence on visual processing across the entire visual cortex. In Chapter 3 we show that both a feed forward and feedback mechanism underlies this attentional magnification. As feed forward and feedback processes are segregated across cortical depth in human V1, we use high-resolution fMRI to measure attentional magnification across cortical depth in V1. Consistent with a feed forward mechanism, we find attentional magnification at each cortical depth. However, this magnification is strongest near the gray white matter boundary, which can only be explained by a feedback process. In Chapter 4, we relate attentional magnification of the neural representation to changes in human perception. We find that spatial attention produces perceptual magnification. This perceptual effect is captured by the attention field model that also captures attentional magnification of the neural representation. Finally, in Chapter 5 we lay the foundation to apply our approach used in earlier chapters to numerosity processing. Our approach relies on an accurate description of neural populations for a visual dimension, like spatial position. In Chapter 5 we provide an accurate description of the sensitivity of neural populations for numerosity. We demonstrate that numerosity information is processed similarly as spatial position. Therefore, we hypothesize that numerosity processing can be affected by attention in similar ways as well.