Gizem Tanseli Kaspar¹ (), Sabrina Hansmann-Roth¹; ¹Icelandic Vision Lab, School of Health Sciences, University of Iceland, Reykjavík, Iceland

Ensemble perception enables us to summarize large sets of objects and extract summary statistics efficiently. Subsampling plays a key role in this process, as it has been often reported, that not all objects contribute equally. External and internal factors such as salience or priors influence the average. Here, we investigate whether averaging can be influenced by both, attentional enhancement of targets and distractor suppression. We aim to determine if attention and suppression influence visual sampling in an orientation averaging task. Each trial consisted of two tasks: a visual search and an orientation averaging task. The visual search served as a priming phase where participants searched for an oddly-colored target among distractors. Each subsequent ensemble contained two subsets of lines (15° CCW and 15° CW from the global mean orientation). We created four conditions for the subsets: a target-change condition in which the distractor color remained part of the ensemble but not the target color; a distractor-change condition which contained the target color, a no-change condition in which the two subsets contained the target and the distractor color and a control condition which used two new colors that were neither target nor distractor color. This allowed us to examine the effects of target enhancement and distractor suppression on a subsequent averaging task. We hypothesized that target enhancement and distractor suppression from the visual search task would carry over to the ensemble task and bias orientation judgments towards or away from the corresponding subsets. Orientation judgments were biased towards the target-colored subset and away from the distractor-colored subset, with the no-change condition showing the strongest bias (even larger than the sum of the individual biases). These findings contribute to an ongoing debate on the role of attention in ensemble coding, demonstrating that both, target and distractor templates bias ensemble judgments.

Acknowledgements: Icelandic Research Fund #239774-051