Abstract
We measured the spatio-temporal distribution of attention during active reading. We examined three variables that are known to affect the processing load of a word: word frequency (high- or low-frequency), whether valid visual information for the word was available before or only after it was fixated (valid or invalid preview, respectively) and the reading task. Using a novel dual-task technique, we probed the focus of attention (6 characters to the left or right of the gaze location) early (temporal offsets of 10 or 40 ms) during the first fixation on target words. For the primary task, participants read either sentences or strings of random words. For reading random words, preview was valid (where a word’s letters were always exposed) for one group of participants and was invalid (where letters in a word were masked and only exposed when the word was fixated) for another group. For reading sentences, preview was valid. For the secondary task, participants made an unspeeded 2-AFC concerning the orientation of the probe, a tilted line briefly superimposed on a character. Discrimination of probe orientation provided the measure of attention. For all task/preview manipulations, fixation durations were longer on low- than high-frequency target words, confirming that reading occurred. The focus of attention was skewed towards the direction of reading (right in English texts) and expanded over time. Attention was more focused on the gaze location for low- than high-frequency words when preview was valid. There was no effect of frequency on the focus of attention when preview was invalid. The results (i) show that the focus of attention in reading is dynamic and is affected by the moment-to-moment processing load of the word, and (ii) are beneficial to models of eye movement control in reading and in general.
Meeting abstract presented at VSS 2013