Eye-movement markers of mind wandering during reading: A meta-analysis.

Abstract

Mind wandering during reading has been extensively investigated, with multiple studies reporting differences in eye-movement behavior between mind-wandering and on-task reading. More recently, researchers have tried to use eye-movement measures to predict mind wandering during reading; however such indicators of mind wandering during reading have not yet been clearly identified. This article presents results from two studies. Firstly, we carried out a meta-analysis to identify eye-movement indicators of mind wandering during reading. From the initial search, abstracts from 140 articles were reviewed for eligibility, and 39 articles were included for full text reading and data extraction. Finally, we identified 16 individual datasets from 19 articles for which we could compute effect sizes. We calculated effect sizes for nine eye-movement measures: mean fixation duration, fixations count, first-fixation duration, gaze duration, total reading time, saccade length, skipping, blink count, and inter-word regressions. The results indicated that readers skipped more words and made fewer fixations during mind-wandering compared to on-task reading. Following these results, we carried out exploratory analyses with a dataset from our own lab to examine other possible explanations for this eye-movement pattern. Specifically, we investigated readers' sensitivity to word frequency and length and readers' use of corrective regressions. Results indicated that readers show reduced effects of word length and frequency on skipping behavior as well as a reduced use of corrective regressions during mind-wandering compared to on-task reading. Results and implications from both studies are discussed in the context of mind-wandering and reading research.