Effects of Working Memory Load on Ensemble versus Individual Object Processing

Ensemble perception is the ability of the visual system to summarize object groups by their statistical properties. At a fundamental level, past studies show clearly that ensembles are perceived, and statistical information is sent to working memory such that a subject may report on the averages (Ariely, 2001; Brady and Alvarez, 2011). Studies show that subjects are capable of reporting ensemble statistics for large groups of objects with high accuracy, inferring that this process bypasses the capacity limitations of attention and working memory (Chong and Treisman, 2003; Baijal et al., 2013; Huang, 2015; Epstein and Emmanouil, 2017, 2021). However, few studies have looked at working memory load and how items in working memory may affect storage and/or processing of ensemble information (Bauer, 2017; Epstein and Emmanouil, 2017). Studies so far suggest that there is no decrement in ensemble processing performance under working memory load, supporting the idea that ensemble perception is independent of working memory capacity limitations (Bauer, 2017; Epstein and Emmanouil, 2021). The current study sought to further test this claim by examining whether working memory load influences the speed by which ensemble properties are computed, as measured by event related potentials (Experiment 1) and behavioral measures (Experiment 2). In both experiments, participants performed either an ensemble or an individual object oddball task (Epstein & Emmanouil, 2021) under high or low working memory. To induce a working memory load, a Sternberg memory task was used such that participants memorized a three (low load) or seven (high load) digit string. Experiment 1 measured the P3b component elicited by the oddballs, which is known to reflect stimulus processing time, as well as behavioral reaction time and accuracy. No effects of working memory load were found on performance in both the ensemble and individual conditions. Working memory load effects were also not found in P3b latency or amplitude. However, significant effects in P3b latency were found between ensemble and individual conditions, replicating previous results using a similar oddball task (Epstein & Emmanouil, 2021). Experiment 1 data collection remained incomplete due to COVID-19. Experiment 2 was conducted through Amazon Mechanical Turk to complement Experiment 1 processing speed. Again, we found no main effects of load on either perceptual task but differences in reaction time were observed between the ensemble and individual conditions. These results taken together imply that taxing top-down attentional resources has little influence on a person’s ability to respond quickly and accurately to ensembles and individual objects. Furthermore, differences between ensemble and individual object perception were evident despite cognitive load and varying experimental conditions