The complex spatial topography of visual attention: Behavior and physiology

I examined, at a fine spatial scale, perceptual sensitivity (d ') and V4 evoked responses as attention was manipulated. Human and macaque subjects were cued to attend to either a small or large spatial area in order to perform a threshold 2AFC color discrimination task. Stimuli were presented at the location of a peripheral cue on a majority of trials and at uncued iso-eccentric probe locations on the remaining trials. Examined at a coarse spatial scale, results agree with previous work showing that cueing improves performance. However, the finer scale provided by my procedure revealed new features of attentional modulation. When the attended region was small, sensitivity (d') was highest for locations approximately 1 degree from the cued target location, but there was a paradoxical performance deficit at the cued location itself. D' was significantly lower for stimuli that appeared at or within 0.5 degrees of the cued location. Sensitivity decreased for stimuli appearing further than 1 degree from the cue. This pattern of sensitivity was obtained using different cue types and stimulus shapes, and was apparently not due to adaptation, masking, or inhibition of return. Interestingly, neural responses for stimuli at the cued location in the narrow-attention task were inhibited relative to their responses when the same stimuli were presented with a fixation task (no attention). When the attended region was large, V4 firing rates were elevated and animals were more sensitive to stimuli presented further from the center of the cued location than when the attended region was small. In fact, retinotopic areas that elicited no response with a simple fixation paradigm did produce a response with widened attention, as if the RF had expanded beyond its canonical dimensions. To my knowledge this is the first demonstration of increases in V4 RF size as a consequence of widening attention. In general, comparing V4 activity in narrow and wide attention tasks paints a complex picture of attentional modulation in V4. These results show that V4 activity reflects shifts in the locus of attention, but at a finer spatial resolution activity is not directly correlated with the spatial allocation of attention.