Keeping an Eye on the Periphery: How Eccentricity affects Visual Selection

After reading the work of this thesis I hope you are convinced that eccentricity is a major factor that cannot be ignored when it comes to understanding visual selection. More specifically, in Chapter 2 we showed that while the proportion of selecting targets and salient items does not change with eccentricity, the dynamics of saliency- and relevance-driven selection do change. That is, the effect of saliency was protracted at a further eccentricity, while the effect of relevance was delayed. This discrepancy between overall selection and performance over time can be explained by the difference in saccade latencies between conditions. That is, as the saccade latency distribution shifted in time with increasing eccentricity, so did the effects of saliency and relevance. In Chapter 3, consistent with earlier work, we challenged existing models of visual selection, showing that the time at which a saccade is initiated greatly influences whether it will be saliency- or relevance-driven. That is, short latency eye movements are more likely to be saliency driven while later eye movements are more likely to be relevance driven. Importantly, we showed for the first time that this separation in time leads to a brief period in between saliency-driven and relevance-driven selection in which the eyes appear to be in ‘limbo’. That is, selection appears to operate randomly, irrespective of saliency and relevance. By fitting different models on the data, we showed that the dynamics of saliency- and relevance-based selection are best described as two independent processes that do not influence each other. We propose an alternative view on the classic priority map model, in which saliency effects are actually a byproduct of a difference in processing speed between different items. That is, on the priority map, salient items are available for selection earlier than non-salient items as they are processed more quickly and elicit therefore more activation at an earlier point in time. After a while, this difference in activation disappears because then non-salient items are processed as well, resulting in a period of non-selectivity. After this, the influence of behavioral relevance takes effect, and activity for the relevant item increases. In Chapter 4 we showed that subjects are more likely to select items that are presented close to fixation than items presented further away. This central selection bias was larger than would be expected based on low-level sensory differences between eccentricities suggesting an important role for attentional competition. In Chapter 5 we were able to determine, for the first time, the time course that the effect of eccentricity follows. Here we showed that eccentricity mainly influences those saccades that are initiated early. That is, eccentricity operates in a similar time window as saliency. As a consequence, the effects of saliency were diminished as the eccentricity difference between the two items grew, but those of relevance were unaffected. In Chapter 6 we showed that attentional capture by salient distractors is modulated by the bias that is described in Chapter 4. That is, even though we saw no effect of eccentricity on attentional capture in overall manual RTs, using eye movement data we showed that participants are more likely to select an item closer to fixation than an item presented further away. Crucially, on those trials in which an eye movement was made towards the distractor reaction times increased with distractor eccentricity while the likelihood of making an eye movement to the distractor in the first place decreases with increasing distractor eccentricity. As these effects go in opposite directions, overall RT showed no effect of distractor eccentricity when all trials were combined together.