Korinth, S. . 1, 2 & Fiebach, C. 1, 2, 3
1 Department of Psychology, Goethe University Frankfurt am Main, Germany
2 Center for Individual Development and Adaptive Education of Children at Risk (IDeA) Frankfurt am Main, Germany
3 Brain Imaging Center, Goethe University Frankfurt am Main, Germany
There is an ongoing debate about the role that global word-shapes versus the constituent letters play for visual word recognition. This debate is paralleled by the proposal of differential roles of low- versus high-spatial frequency components in visual object recognition. A prominent theoretical framework introduced by Bar (2003), postulates separate neuronal pathways in which the fast low-spatial frequency component provides the basis for prediction generation about an object?s identity while the slower high-spatial frequency component allows a verification or disconfirmation of such predictions. We here explore this model in the domain of visual word recognition. Participants performed a lexical decision task, in which prior to the appearance of the target-word either a black square (zero-prime) or a silhouette-prime (approximating the word?s low-spatial frequency component) were presented. Stimuli consisted of triplets with virtually identical silhouettes that, however, differed in word frequency or lexicality (e.g., glass, gloss, gless), selected to evoke prediction errors of gradually increasing strength. We hypothesized that if silhouette primes generate predictions, frequent words appearing as targets should verify the prediction elicited by the silhouette primes and thus accelerate responses. In contrast, prediction errors of varying strengths should be elicited by infrequent words and pseudo-words, which should reduce or even extinguish prime benefits. Results show significant priming effects (i.e., shorter RTs) for frequent words when preceded by a silhouette prime compared to a zero-prime. Priming effects were not observed for infrequent words and for pseudo-words. This result pattern supports the notion that - as shown for visual object recognition - global shape information facilitates visual word recognition in a prediction-driven manner, which suggests important commonalities between visual object and word recognition.