EEG Correlates of Working Memory Predict Gaze Variability during a Real-World Information Foraging Task

• ¹ Wright State University, Psychology, United States
• ² Wright State Research Institute, Wright State University, United States

Electroencephalography (EEG) provides unique advantages to the study of human factors and performance. EEG yields continuous, real-time information with precise temporal indices of changes in alertness and attention. EEG can also be used as an implicit measure of behavior without requiring the awareness or explicit responding of participants. One domain where EEG has proven useful is in working memory and cognitive control. For instance, EEG studies have found that contralateral delay activity (CDA) is a robust neural marker of working memory capacity; its amplitude increases as a function of the number of items to be attended, and its asymptote coincides with behavioral measures of working memory capacity (Vogel & Machizawa, 2004; Luck & Vogel, 2013). Furthermore, CDA has also been used as a measure of participants’ ability to filter out distractors during a working memory task (Vogel, McCollough & Machizawa, 2005), an ability likely related to cognitive control. While CDA is a highly useful index of lab-based measures of working memory and cognitive control processes, its applicability to real-world behaviors is still unknown. Consequently, the present research sought to establish the extent to which CDA encapsulates information pertinent to naturalistic behaviors in a novel real-world information foraging task (IFT). To that end, we correlated neural measures of cognitive filtering and working memory with overt behavioral measures recorded while participants performed the IFT. Participants completed a pair of delayed match to sample tasks (a behavioral working memory capacity assessment, followed by an EEG recording) and the IFT over two separate sessions. We hypothesized that, on average, those higher on CDA and filtering measures should exhibit lower gaze variability—an overt behavioral measure of participant distractibility. In addition, those higher in CDA and filtering should also exhibit higher levels of self-reported cognitive control as a convergent measure. Working Memory Capacity Assessment: Prior to the EEG recording, participants each completed a delayed-match-to-sample task. Every trial, they were shown 2, 4 or 6 differently colored squares, randomly positioned around a fixation dot. Squares appeared for 200ms, followed by a 900ms retention interval, then reappeared for 1000ms. Upon reappearing, one square's color changed on half of trials (randomly). Participants reported changes via button presses. CDA and filtering: During EEG recording, participants completed a modified delayed-match-to-sample task in which either 2 red rectangles, 4 red rectangles or 2 red and 2 blue rectangles could appear on both sides of fixation (see Figure 1 for stimulus schematic). Participants reported orientation changes in any of the rectangles between first and second appearance as quickly and accurately as possible via button press. They were also told to ignore the blue rectangles, as these were only distractors that would never change orientation. Information Foraging Task: In a separate session, participants were given 40 minutes to research an uninteresting topic online before being given a short quiz on that topic. They were informed that during this time their seating position, gaze position, and internet browsing history would be monitored. A list of links to topic-relevant websites was provided, although the participants were told they could freely visit other websites. They were also told that they could spend this time any way they liked (e.g. visiting social media websites if they chose). Following the quiz, participants also completed the Self-Control Scale (Tangney, Baumeister & Boone, 2004), Attentional Control Scale (Ólafsson et al., 2011), Need For Cognition (Cacciopo & Petty, 1982), 12-Item GRIT (Duckworth, Peterson & Matthews, 2007), and Big-Five Personality Inventory (McCrae, 2004). Results & Discussion: We generally find that neural measures of cognitive filtering can be used to predict gaze-related behaviors in a novel ‘real-world’ IFT. Specifically, we show that neural measures of working memory (CDA and cognitive filtering) exhibit robust correlations with the variability of saccade amplitude, as well as other gaze-related measures (e.g. head movements and time spent looking off-screen). We also found moderate correlations between the EEG and self-reported measures related to cognitive control (self-control, attentional-control, and need for cognition). Together, these findings suggest that those with higher working memory capacity and better cognitive filtering capabilities (as assessed by EEG measures) require, on average, fewer fixations and saccades to process similar amounts of information during the IFT. Overall, our results demonstrate that laboratory-based neural correlates of working memory can be used to predict overt real world attention-related behaviors in ecologically valid settings.