Michele Maslowski^1,2, David H. Hughes^1,2, Adam S. Greenberg^1,2; ¹Marquette University, ²Medical College of Wisconsin

During RSVP experiments (Chun & Potter, 1995), a target image appears among temporally presented distractors. The assumption is that items are processed independently, however evidence of an attentional blink (Raymond, Shapiro, & Arnell, 1992) suggests interactions between target and distractor items. Here, we hypothesize that if target image content is highly correlated with distractor images, target perception and task performance will be affected. Overlapping face and house images were embedded in noise and presented via RSVP. Each frame contained one face and one house image chosen from 32 possible images for each stimulus type; with two specific face and house images as temporal search targets. Subjects attended to either the face or house stream (cued via colored image frame) and identified targets. Frame duration (500 ms, 600 ms, 750 ms, 900 ms, 1000 ms, 1200 ms) was varied across 9 blocks. The distribution of correlations between target and distractors varied, though faces were more strongly correlated with distractors (Face 1: μ = 0.623, σ = 0.0567; Face 2: μ = 0.476, σ = 0.0522; House 1: μ = 0.3168, σ = 0.079; House 2 : μ = 0.2884, σ = 0.0621). We evaluated the correlation between target and subsequent frame (or frame to which subjects responded, for FAs) for the following response outcomes: hits (H), false alarms (FA), and misses (M). We then calculated the difference in correlation between each combination of response outcomes (FA-H, FA-M, H-M). We observed a significant interaction between target face and response outcome (p < 0.01). Significant positive difference scores were observed for FA-H (p < 0.01) and FA-M (p < 0.01) comparisons, suggesting that higher correlation values elicit a FA response. House targets failed to produce significant interactions, suggesting lower correlations limit influence on performance. Thus, RSVP target-distractor correlations affect target perception.