The eye of the beholder: Can patterns in eye movement reveal aptitudes for spatial reasoning?

Abstract

Mental rotation ability (MRA) is linked to academic success in the spatially complex Science, Technology, Engineering, Medicine, and Mathematics (STEMM) disciplines, and anatomical sciences. Mental rotation literature suggests that MRA may manifest in the movement of the eyes. Quantification of eye movement data may serve to distinguish MRA across individuals, and serve as a consideration when designing visualizations for instruction. It is hypothesized that high-MRA individuals will demonstrate fewer eye fixations, conduct shorter average fixation durations (AFD), and demonstrate shorter response times, than low-MRA individuals. Additionally, individuals with different levels of MRA will attend to different features of the block-figures presented in the electronic mental rotations test (EMRT). All participants (n = 23) completed the EMRT while metrics of eye movement were collected. The test required participants view pairs of three-dimensional (3D) shapes, and identify if the pair is rotated but identical, or two different structures. Temporal analysis revealed no significant correlations between response time, average fixation durations, or number of fixations and mental rotation ability. Further analysis of within-participant variability yielded a significant correlation for response time variability, but no correlation between AFD variability and variability in the number of fixations. Additional analysis of salience revealed that during problem solving, individuals of differing MRA attended to different features of the block images; suggesting that eye movements directed at salient features may contribute to differences in mental rotations ability, and may ultimately serve to predict success in anatomy. Anat Sci Educ 9: 357-366. © 2015 American Association of Anatomists.

Different perspectives: Spatial ability influences where individuals look on a timed spatial test

“…Higher spatial working memory may enable quicker comparisons between images and allow for a more consistently accurate problem solving process (Wilhelm and Schulze, ). In terms of accuracy, average fixation duration was found to be consistent within both MRA groups regardless of answer accuracy, and in line with the previous findings in an untimed environment (Roach et al, ). As a result, average fixation duration may be somewhat constant within individuals regardless of answer accuracy.…”

“…Essentially, what can be inferred from these findings is that both groups complete the task in the same way. Questions that are answered correctly are answered quickly; while those that are not solved are puzzled over; directly supporting the findings observed by Roach et al under the untimed condition (Roach et al, ). Ultimately, the frequency with which the question is solved accurately represents the aptitude for spatial reasoning.…”

“…The areas colored red represent the most attended region of the image, and are indicative of the highest salience across the group. Reprinted with permission from Roach et al ().…”

“…On average, LMRA individuals exhibited longer trial response times, and conducted more fixations per trial. However, the results obtained by Just and Carpenter () were in contrast with the results observed recently by Roach et al (), where a similar test of mental rotation ability (the electronic mental rotation test, EMRT), was employed without a time restriction. Roach and colleagues found little relationship between most chronological (average response time) and numeric (number of fixations per question) measures of eye movement and MRA; but upon conducting an analysis of visual salience, (i.e., where on the images participants devoted visual attention), found wide differences between the HMRA and LMRA groups (Roach et al, ).…”

“…Visual salience can be defined as a quality, or a set of characteristics, held by a region of an image that is perceived by the viewer as being conspicuous relative to its surroundings (Itti et al, ). In the aforementioned EMRT experiment, the authors refer to regions of the images as being “spatially salient.” These “salient domains” of the presented stimuli conveyed depth and positional information, and commanded visual attention, which may have facilitated accurate, and quick completion of the untimed EMRT (Roach et al, ). Despite studies evaluating the performance of HMRA and LMRA individuals in untimed environments, little is known about how individual eye movement behaviors differ in time‐restricted, or “speeded” testing environments.…”