Poster #11 - Robust Visual Working Memory for object features in 2.5-year olds

• In Event: 2-110 - Poster Session 07
  In Poster Session: PS 07 Section - Attention, Learning, Memory

Integrative Statement

Adults are able to remember, with great detail, more than 1000 images (Brady et al., 2008). Similar results were found in preschoolers (Ferrara et al., 2017). Suggestively, infants can remember more objects when they are heterogeneous, as opposed to when they share a category (Zosh & Feigenson, 2015). However, this interaction between content and capacity has not been well-studied in the early development of visual working memory (VWM). The present study aimed to fill this gap. In two experiments, we used the Delayed Match Retrieval (DMR) paradigm (Kaldy et al., 2016) to test VWM for object-location bindings: will toddlers better remember objects when they are unique (Exp. 1) as opposed to sharing color or shape (Exp. 2)?
In Experiment 1, four (virtual) cards were presented face-down. Then, three, to-be-remembered cards were flipped face-up sequentially, one every 1 s, to reveal a unique abstract object (Figure1a). (We used abstract, non-nameable images to minimize verbal encoding). Then, all three turned face-down. After 1 s, the fourth card, the Sample, flipped face-up. The face of this card was identical to one of the three previously exposed cards (the Match). The reveal of the Sample began a 2 s anticipatory period, after which the Match card was flipped face-up, accompanied by a reward animation at its location (e.g. brief fireworks). During the anticipatory period, toddlers were expected to look back to the (face-down) Match in anticipation of the reward (Figure2). Gaze was recorded by a Tobii T120 eye tracker throughout the trial. Twenty-four 2.5-year-old toddlers (ages: 29;00-37;00, 10 females) were tested. The percent of trials when they looked at the Match (as opposed to either of the two non-matching cards) first during the anticipatory period was significantly above chance (33%) (mean: 46% CI +/- 9%, one-sample t-test: p<0.01, d=0.99). This indicates that toddlers succeed at the DMR task with a memory set of three, when the objects are heterogeneous.
In Experiment 2, we tested how robust the encoded representations are by examining whether toddlers could remember objects that shared some visual features. We manipulated the similarity of the three to-be-remembered cards. The two non-matching cards were designated as either a Mis-Match, which shared shape or color with the Match or a Non-Match which was distinct from the Match (but shared color or shape with the Mis-Match) (see Figure1b). Twenty-six 2.5-year-old toddlers (ages: 29;00 - 37;00, 11 females) were tested. Results showed that toddlers succeeded at this task as well, achieving 41% correct performance (CI +/- 8%) based on the first fixation to the Match (p=0.052,d=0.82). Though performance in Experiment 2 was lower than in Experiment 1, the difference was not significant (two-sample t-test: p=0.40, d=0.22) (see Figure3).
Our findings suggest that 2.5-year-old toddlers can robustly encode multiple features (here, shape and color) of objects in VWM; with these stimuli, the limitation on VWM performance is not the quality of the encoded features.

Authors

• CHEN CHENG, University of Massachusetts Boston
  Presenting Author

• Zsuzsa Kaldy, University of Massachusetts Boston
  Non-Presenting Author

• Erik Blaser, University of Massachusetts Boston
  Non-Presenting Author