Mean Size as a Unit of Visual Working Memory

Abstract

Visual environments often contain multiple elements, some of which are similar to one another or spatially grouped together. In the current study we investigated how one can use perceptual groups in representing ensemble features of the groups. In experiment 1 we found that participants' performance improved when items were easily segmented by a grouping cue based on proximity, suggesting that spatial grouping facilitates extracting and remembering ensemble representations from visual arrays consisting of multiple elements. In experiment 2 we found that spatial grouping improved performance only when the grouped subsets were tested for the memory task, whereas it impaired performance when other subsets that were not grouped were tested, suggesting that the benefit from grouping may come from better extraction for storage, rather than later decision processes such as accessibility. Taken together, our results suggest that perceptual grouping of multiple items by proximity facilitates extraction of ensemble statistics from groups of items, enhancing visual memory of the ensembles in a visual array.

Accurate rapid averaging of multihue ensembles is due to a limited capacity subsampling mechanism

“…If the precision of settings around the expected hue is similar when setting to the average of a set of hues or setting to a single hue, it will be a strong indicator that the mean hue is encoded as strongly as individual hues, suggesting that the ensemble is represented by a single average hue. These measurements of precision were also used to address the question of whether the observed hue averaging precision could be the result of a limited sub-sampling mechanism or whether the performance could support the proposal of an efficient holistic mechanism, integrating attention from the whole ensemble and circumventing the limited capacity of visual working memory [16,[37][38][39]. Measurements of internal noise (based on adjustments to single hues) were incorporated into a simulation estimating the precision of mean adjustments based on a random sub-sample of n elements.…”

Ensemble statistics as units of selection

“…Representing an ensemble as a single unit of similar items increases the efficiency of the selection mechanism by providing more information in a compressed form about a larger number of objects than the visual system can process individually. The selected ensembles can then be manipulated and further stored together in the same way as individual objects (Im & Chong, 2014). Selection based on ensembles thus enhances visual processing (Alvarez, 2011;Brady & Tenenbaum, 2010 by increasing the diversity and flexibility of the visual system, and allowing more efficient use of its limited capacity to process items simultaneously.…”

“…visual working memory), in the same manner as individual objects. For example, the visual system can extract and maintain ensemble representations from up to three or four sets of items at once (Halberda, Sires, & Feigenson, 2006;Im & Chong, 2014), but no more than that. This is similar to what has been observed for individual objects, which suggests that representations of ensembles and of individual objects may both be similarly constrained by the capacities of visual attention and working memory (for review, see Feigenson, 2011).…”

Visual statistical learning of temporal structures at different hierarchical levels

“…They suggested that the encoding of hierarchical structures helped observers to base their judgments on global-level aspects of the display (i.e., ensemble statistics) instead of guessing. Im and Chong (2014) also showed that observers could better estimate the mean sizes of groups of items when the items that shared the same color were spatially grouped, thereby allowing extraction of the hierarchical relationship between the items to facilitate the process.…”

“…Another way to cope with the complexity of a scene is to use the hierarchical structure of our environments (Brady, Konkle, & Alvarez, 2011;Im & Chong, 2014). Specifically, the visual information of a scene can be organized into local-or globallevel information (Navon, 1977).…”