Prioritization in Visual Attention Does Not Work the Way You Think It Does

• Published in: Journal of experimental psychology. Human perception and performance 2021-02, Vol.47 (2), p.252-268
• Main Authors: Ng, Gavin J. P., Buetti, Simona, Patel, Trisha N., Lleras, Alejandro
• Format: Article
• Language: English
• Subjects: Attention; Candidates; Distraction; Female; Human; Male; Open Data; Peripheral Vision; Reaction Time; Spatial attention; Spatial Perception; Theories; Visual Attention; Visual Displays; Visual perception; Visual Search
• ISSN: 0096-1523; 1939-1277
• DOI: 10.1037/xhp0000887

A common assumption in attention theories is that attention prioritizes search items based on their similarity to the target. Here, we tested this assumption and found it wanting. Observers searched through displays containing candidates (distractors that cannot be confidently differentiated from the target by peripheral vision) and lures (distractors that can be). Candidates had high or low similarity to the target. Search displays were either candidate-homogeneous (all items of same similarity) or candidate-heterogeneous (equal numbers of each similarity). Response times to candidate-heterogeneous displays were equivalent to the average of high- and low-similarity displays, suggesting that attention was allocated randomly, rather than toward the high-similarity candidates first. Lures added a response time cost that was independent of the candidates, suggesting they were rejected prior to candidates being inspected. These results suggest a "reverse" prioritization process: Distributed attention discards least target-similar items first, while focused spatial attention is randomly directed to target-similar items. Public Significance Statement Most theories of attention propose that attention visits locations in a scene in descending order of their priority, with attentional priority reflecting an object's similarity to the target object or feature that the observer is looking for. Here we propose that the inverse is in fact true. Attention starts by evaluating peripheral information in parallel and rejecting unlikely targets as a function of their dissimilarity to the target; that is, attention moves up the similarity scale, not down. Furthermore, we propose that given the processing limitations of peripheral vision, attention cannot be properly guided by visual similarity at the top of the priority scale: Once similarity is sufficiently high, attention simply visits potential target locations at random. This new bottom-to-top conceptualization of attentional processing should be of wide interest to anyone working in an attention-related field, applied or theoretical.