Verbalizable material such as numbers can enhance short-term memory capacity when individual units are grouped into chunks (Miller, 1956). In visual short-term memory (VSTM), perceptual groups can serve a similar chunking function. To date, however, increased capacity has been demonstrated only for grouping by identical colors, but not by identical shapes (Quinlan & Cohen, 2012).

Previous research suggests two psychological mechanisms underlying the influence of perceptual grouping in VSTM. First, grouping reduces overall memory load: all items, including singletons, benefit from the presence of a perceptual group (see Thalmann et al., 2019, for a corresponding result from verbal working memory). Second, attention is automatically biased toward grouped items (Kimchi et al., 2007; Morey et al., 2015), increasing their likelihood of being encoded while reducing that of singletons. Together, these mechanisms explain why grouping effects are typically observed only for grouped items.

Across three experiments using a change detection paradigm, I show that a stochastic model integrating capacity and attentional mechanisms accounts for the observed data with high accuracy. Moreover, I demonstrate a capacity benefit for shape-based grouping—provided that shape discrimination is easier than the corresponding color discrimination.