Abstract
Reading requires processing visual stimuli in a hierarchy that spans lower-level features (such as brightness and color) and higher-level features (such as semantic meaning). When multiple words are presented simultaneously, which features can be processed in parallel, and which must be processed serially? While recent literature suggests that colors of multiple stimuli are processed in parallel, the literature contains mixed results concerning parallel or serial processing of word meaning. We used a web-based redundant target paradigm to examine parallel versus serial processing in two closely related tasks: discrimination of colored letters vs. grey letters, and the semantic categorization of living vs. nonliving words. A redundancy gain is when a participant responds more quickly and more accurately when two target stimuli are presented simultaneously compared to when only one target is presented. Such a redundancy gain is predicted by parallel processing models but not by self-terminating serial models. We found large redundancy gains in the color task for both response time and accuracy, consistent with parallel processing of this low-level feature in two letter strings. However, in the semantic task, redundant targets were associated with slower response times, and a small increase in hit rates. This is consistent with serial processing and a speed-accuracy tradeoff. Overall, we conclude that while low-level visual features can be processed in parallel, word meanings cannot be.