Poster #3 - Visual STM persists across multiple fixations: N-back approach to quantifying capacity in infants and adults

• In Event: Poster Session 05

Abstract

Visual short-term memory (STM) is a foundational component of general cognition that develops rapidly during the first year of life. Despite the fundamental importance of this “working” memory system, adults can remember only about 3-4 objects at a time. Limitations in working memory may be even more critical for preverbal infants, as cognitive development is dominated by visual exploration. Currently, it is unclear if infant tasks tap the same working memory system as adults. Further, it remains unclear if infants can remember multiple sequentially fixated array items. In this poster, we explore these questions using a one-shot passive change detection paradigm modeled after previous work (Ross-Sheehy & Eschman, 2019). However, unlike previous research, the current task incorporated a gaze contingent color change manipulation, with the change location being selected based on the sequence of fixations observed during the sample array (e.g., last item fixated, second-to-last item fixated, or one of the non-fixated items).

Infants (5 and 11mos) and adults were tested in the same task that incorporated a modified N-back manipulation. However, rather than presenting items one-at-a-time, array items were presented all-at-once, and the order of sequential fixations observed during the sample array was used as a proxy for the serial presentation approach. In this way, it was possible to manipulate the location of the color change based on the order in which each circle was fixated during the sample array (i.e., fixation sequence).

Run Count (the number of times the participant visited or re-visited individual circle locations) was analyzed using a repeated measures ANOVA with condition (N-back1, N-back2, change-other, no-change) as a within subjects variable and age (5mos, 11mos) as a between subjects variable. Results revealed a significant main effect of condition, F(3,138)=24.302 p<.001, partial eta squared=.346, with the longest run counts for the N-back1 and N-back2 conditions (Figure 1, top panel). All three change conditions (N-back1, N-back2, and change-other) produced significantly longer run counts than the no-change condition (all ps <.001), providing clear evidence of change detection.

In addition to Run Count, we also looked at change preference. Again, we ran a repeated measures ANOVA, with condition (N-back1, N-back2, change-other) as the within subjects variable and age (5mos, 11mos) as the between subjects variable. Again, results revealed a significant main effect of condition, F(2,92)=31.212, p<.001, partial eta squared=.404, with the highest change preference scores for N-back1, and lowest for change-other conditions (Figure 1, bottom panel). Follow up simple effects tests revealed significantly higher change preference scores for N-back1 relative to both N-back2 (p=.025), and change-other conditions (p<.001). N-back2 change preference scores were also significantly higher than the change-other condition (p<.001).

A permutation analysis examining change preference over time (see Figure 2) suggested differences could not be explained by perseverative looking or location biases.

This finding demonstrates strong evidence of visual working memory for at least two objects that cannot be explained by perseveration biases, or contamination from longer-term memory systems. In addition, this approach could be used to assess individual differences in visual working memory capacity for infants, children, and adults, in a much more naturalistic way.

Authors

• Bret Eschman, University of Tennessee at Chattanooga
  Presenting Author

• Shannon Ross-Sheehy, University of Tennessee - Knoxville