Entropy of eye movement during rapid automatized naming.

Numerous studies have focused on the understanding of rapid automatized naming (RAN), which can be applied to predict reading abilities and developmental dyslexia in children. Eye tracking technique, characterizing the essential ocular activities, might have the feasibility to reveal the visual and cognitive features of RAN. However, traditional measures of eye movements ignore many dynamical details about the visual and cognitive processing of RAN, and are usually associated with the duration of time spent on some particular areas of interest, fixation counts, revisited fixation counts, saccadic velocities, or saccadic amplitudes. To cope with this drawback, we suggested an entropy-based method to measure eye movements for the first time, which first mapped eye movements during RAN in a time-series and then analyzed the time-series by a proper definition of entropy from the perspective of information theory. Our findings showed that the entropy was more sensitive to reflect small perturbation (e.g., rapid movements between focuses in the presence of skipping or omitting some stimulus during RAN) of eye movements, and thus gained better performance than traditional measures. We also verified that the entropy of eye movements significantly deceased with the age and the task complexity of RAN, and significantly correlated with traditional eye-movement measures [e.g., total time of naming (TTN)] and the RAN-related skills [e.g., selective attention (SA), cognitive speed, and visual-motor integration]. Our findings may bring some new insights into the understanding of both RAN and eye tracking technique itself.