Benefits and pitfalls of data compression in visual working memory

Abstract

Data compression in memory is a cognitive process allowing participants to cope with complexity to reduce information load. However, previous studies have not yet considered the hypothesis that this process could also lead to over-simplifying information due to haphazard amplification of the compression process itself. For instance, we could expect that the over-regularized features of a visual scene could produce false recognition of patterns, not because of storage capacity limits but because of an errant compression process. To prompt memory compression in our participants, we used multielement visual displays for which the underlying information varied in compressibility. The compressibility of our material could vary depending on the number of common features between the multi-dimensional objects in the displays. We measured both accuracy and response times by probing memory representations with probes that we hypothesized could modify the participants' representations. We confirm that more compressible information facilitates performance, but a more novel finding is that compression can produce both typical memory errors and lengthened response times. Our findings provide clearer evidence of the forms of compression that participants carry out.

Between-item similarity frees up working memory resources through compression: A domain-general property.

“…Previous studies in the visual domain have shown that compression, although enhancing WM precision, can also lead to drawbacks (Haladjian & Mathy, 2015; Nassar et al, 2018). For instance, it has been shown that compression may lead to an oversimplification of information, thereby increasing the proportion of false recognition for more compressible sequences (Lazartigues et al, 2021). Similarly, participant’s responses in visual array tasks appear to be biased toward the mean of the ensemble representation (Brady & Alvarez, 2011; Son et al, 2020), suggesting that some of the original representation is potentially lost during the compression process.…”

Working memory capacity for continuous events: The root of temporal compression in episodic memory?

“…However, it could also be the case that some form of compression mechanism already operates at the level of WM representations (i.e., when constructing the event model of the current situation). Between event boundaries, the continuous unfolding of experience could be summarized through WM processes, enabling the construction of event models covering longer time spans, but in a format further away from reality/perception (Al Roumi et al, 2021;Brady et al, 2009;Lazartigues et al, 2021). Indeed, previous studies have shown that visuospatial information (e.g., spatial sequences, multidimensional objects, or color-changing shapes) that includes regularities tends to be compressed in WM.…”

Statistically based chunking of nonadjacent dependencies.

“…For example, in the visual domain, some studies suggest that individuals utilize such structural dependencies to simplify their mental representation of stimuli to allow their retention in memory. This is thought to take different forms, with individuals encoding both detailed information about specific items and the overall summary of a scene (Brady & Tenenbaum, 2013; Hollingworth & Henderson, 2003; Oliva, 2005), or by regularizing the features of a scene to facilitate compressibility and reduce short-term memory (STM) load—though such compression may also cause memory errors by oversimplifying the data (Lazartigues et al, 2021). Yet other studies have challenged the notion of chunking as a form of data compression that frees up working memory (Norris et al, 2020), and instead suggest that chunking may be achieved by redintegration.…”