UC Berkeley

The topics of this dissertation center around the facilitation or hindering of performance on cluttered peripheral vision tasks by spatial attention. I focus primarily on two well-characterized visual phenomena: visual crowding and ensemble perception. Visual crowding is defined as the detrimental effect of clutter on object recognition in peripheral vision. Ensemble perception is defined as the ability to quickly extract a summary representation from a set of similar objects. These two phenomena seem to complement each other. For example, peripheral vision is very limited in identifying individual features and objects due to crowding, but it excels at quickly obtaining a summary-statistical representation due to ensemble perception. The roles of spatial attention in visual crowding and ensemble perception, and the commonalities between them on an individual and a group level, are not as well understood.

In the first study, we investigated the effects of involuntary and voluntary attention on crowding. Visual spatial attention can be allocated in two distinct ways: one that is involuntarily captured by salient external stimuli, and one that is voluntarily directed to behaviorally relevant locations in the world. We used an anti-cueing paradigm to separately measure the effects of involuntary and voluntary spatial attention on the critical spacing of crowding, defined as the the minimum target/flanker spacing at which the target is correctly identified at a specified level of performance. We found that involuntary capture of attention led to faster response times (RTs) and smaller critical spacing, while voluntary allocation of attention led to faster RTs but no significant effect on critical spacing.

In the second study, we expanded on the first study by comparing the effects of different sizes of a peripheral involuntary attention cue on performance of an orientation discrimination crowding task and performance of an ensemble mean orientation discrimination task. The stimuli consisted of a central Gabor surrounded by a ring of uniformly-spaced Gabors. We varied the relationship between the central Gabor’s orientation and the mean orientation of the entire stimulus array to see how the size of the cues interacted with different patterns of ensemble statistics (Gabor orientations) relative to the orientation of a single cued object within the ensemble. We found that only the small cue decreased the effect of crowding, while the large cue that encompassed both the target and flankers did not. We also found moderate differential effects of cue size on ensemble perception, but this interaction was primarily observed if the cued Gabor was more salient than the non-cued Gabors within the ensemble.

In the final study, we expanded on the analysis of the second study by fitting response models of different complexities to the psychophysical data. The goal of the modeling approach was to investigate how observers utilized all of the orientations in the stimulus array to make their responses, to determine the role of task relevance in the response process, and to test if common strategies existed between the two tasks. We found that response patterns were better explained by a combination of the Gabor orientations considered independently, as opposed to a spatial-weighted average of these orientations. We also found that spatial-weighting strategies, inferred based on the values of model parameters, were correlated between the two tasks, even though task performance was not correlated for the two tasks.