Individual differences in eye movement patterns are associated with cognitive performance : behavioral and neural evidence in face perception

Abstract

Eye movement has been shown to reflect underlying cognitive processes. Recent research suggested that executive and visuospatial function deficits are likely to be reflected in eye movements. However, it remains unclear whether individual differences in eye movement can reflect one's cognitive performance. Here we examined this issue using the Eye Movement analysis with Hidden Markov Models (EMHMM) approach. This approach summarizes one’s eye movement pattern in a visual task using an HMM with person-specific regions of interest and transition probabilities among them. Common patterns among individuals can be identified through clustering. Similarities between individual patterns and the common patterns can be quantitatively measured using machine-learning methods. In Study 1, two common eye movement patterns, which were also identified in previous studies, were observed among young and older adults: “holistic” (mostly fixating around the face center) and “analytic” (with frequent transitions among the two eyes and the face center). The analytic pattern predicted better face recognition performance. Significantly more older adults adopted the holistic pattern and more young adults adopted the analytic pattern. Importantly, similarity to the holistic pattern was negatively correlated with older adults’ cognitive status (assessed by the Montreal Cognitive Assessment), particularly in executive (Tower of London) and visual attention (Trail Making task) functioning. In Study 2, we conducted a neuroimaging experiment examining the neural differences between the analytic and holistic patterns in young adults. Participants using holistic patterns relied more on the perceptual areas whereas those using analytic patterns engaged more the areas important for executive and visual attention functioning during face recognition, providing evidence for the association of eye movement pattern and cognitive decline revealed in Study 1. The advantage of analytic over holistic patterns in face recognition may be a consequence of more engagement of active gaze planning and top down control of visual attention. Study 3 explored the neural differences between the common patterns of young adults in the temporal dimension with simultaneous recording of EEG and eye movement during a face recognition task. The differences between nose-focused (similar to holistic) and eye-focused (similar to analytic) patterns emerged 100 ms after the participants looked at the face stimuli. Participants using eye-focused patterns engaged more local face processing (larger N170), perceiving faces as more distinct (smaller P200), better perceptual memory representation (larger N250), and engaged in more stimulus evaluation processes (longer P300 latency) than those using nose-focused patterns. Study 4 examined age-related differences in eye movement patterns adopted for identifying "happiness", "sadness", "surprise", "neutral", "fear", "anger", and "disgust". However, the common patterns discovered for each expression do not explain the identification performance very well, especially in young adults. The age-related declines in facial expression identification may be more consistent with the aging-brain explanation. Associations between eye movement pattern and cognitive performance were found in "happiness", "fear", and "disgust" only, suggesting that eye movement patterns in the face recognition task was a better predictor of one's cognitive performance. These studies suggest the potential use of eye tracking as an effective and efficient screening tool for cognitive decline/deficits.