Abstract
Visual working memory (WM), is a capacity limited system used for the short-term storage and manipulation of visual information (Baddeley & Hitch, 1974). WM capacity is typically assessed using a “change-detection” task, and capacity for each participant is estimated based on behavioral accuracy across different set sizes. Given the relation between capacity and academic achievement (Alloway & Passolunghi, 2011; Bull et al., 2008), WM may be a useful marker of general cognitive development in infants. Although infants cannot make explicit behavioral responses as adults do, previous eyetracking work with adults (Ross-Sheehy and Eschman, under review) suggests that adults make significantly more saccades and have significantly longer fixations for incorrect trials than for correct trials. Thus, in an attempt to further quantify the saccade/fixation correlates of accuracy in adults, we present here results from three adult studies. Seventy-two adult participants (24 in each task) completed a change-detection task while their eye gaze was recorded with an Eyelink 1000 + eyetracking system. Button responses (e.g. same/different) were also collected. Experiment 1 examined visual dynamics during a replication of Luck and Vogel (1997), Experiment 2 was a modified version of Experiment 1 with larger total eccentricity and larger array items, and Experiment 3 replicated Experiment 2 only with increased encoding time (1000ms vs. 100ms in Exps 1–2). Across all three experiments, results were highly consistent: Adults made significantly more saccades for incorrect trials [Exp 1, F(1,12)=9.919, p=.008,h2p=.453, Exp 2, F(1,10)=6.694, p=.027,h2p=.401, and Exp 3, F(1,19)=8.954, p=.007,h2p=.320, and had significantly longer run counts for incorrect trials [Exp 2, F(1,10)=9.087,p=.013, h2p=.476 and Exp3, F(1,19)=4.588, p=.045,h2p=.195]. Taken together these results suggest: 1) Saccades and fixations may be useful new indexes of WM efficiency, and 2) visualdynamics may be an index of WM maintenance, and 3) encoding time, total eccentricity and element size did not influence WM performance, suggesting that adopting infant parameters does not fundamentally change the task.