A population response model of ensemble perception.

Abstract

Ensemble representations have been considered as one of the strategies that the visual system adopts to cope with its limited capacity. Thus, they include various statistical summaries such as mean, variance, and distributional properties and are formed over many stages of visual processing. The present study proposes a population-coding model of ensemble perception to provide a theoretical and computational framework for these various facets of ensemble perception. The proposed model consists of a simple feature layer and a pooling layer. We assumed ensemble representations as population responses in the pooling layer and decoded various statistical properties from population responses. Our model successfully predicted averaging performance in orientation, size, color, and motion direction across different tasks. Furthermore, it predicted variance discrimination performance and the priming effects of feature distributions. Finally, it explained the well-known variance and set-size effects and has a potential for explaining the adaptation and clustering effects.

Observers Efficiently Extract the Minimal and Maximal Element in Perceptual Magnitude Sets: Evidence for a Bipartite Format

“…This stimulus structure also changed the standardized difference: The z score of the longest line increased with set size (from approximately 1.75 at a set size of seven to approximately 2.20 at a set size of 11), whereas the shortest line decreased with set size (from approximately −1.00 to approximately −0.85). Models predicting that the longest/shortest line is found through a comparison to the mean-either via a ratio or a standardized-difference metric (Robitaille & Harris, 2011;Utochkin et al, 2023)-would therefore predict a set-size effect. The bipartite format, on the other hand, predicts no effect of set size.…”

“…The bipartite format is also distinct from the similarity-to-the-mean account. Extracting the mean value (e.g., the average length line) improves with set size (Utochkin et al, 2023), predicting higher precision for finding the endpoints at higher set sizes. If the individual items are also equally spaced from each other, then increasing the set size also increases the range, again predicting that the maximal and minimal identification should improve with set size.…”

“…This is analogous to how digital computers order values (e.g., the quicksort algorithm; for extensive review, see Cormen et al, 2022) but necessitates a mathematical limit to its efficiency: The time and memory demands grow with the size of the list because more comparisons must be done, predicting more errors as the set size grows. An alternative algorithm comes from ensemble perception and pooledpopulation normalization: Perception might preattentively code for the typical value in the entire set represented on a linear/log scale, with items that are further from the mean popping out as potential outliers via a similarity metric (Haberman & Whitney, 2012;Utochkin et al, 2023).…”

Ensemble memory of a scene interacts with current perception regardless of attentional requirements.

“…Similar features among memory stimuli may have contributed to the robust processing of ensemble information. Recently, Utochkin et al (2023) demonstrated that population responses are more tightly formed with similar items. From this perspective, similar peripheral details may be aggregated to form a gist with strong visual working memory representation.…”