Ness, T. & Meltzer-Asscher, A.
Tel Aviv University
Recently, a distinction was suggested between two prediction mechanisms in sentence processing: ?pre-activation?, i.e. activating representations in long-term memory, and ?pre-updating?, i.e. updating the context?s representation in working memory (WM) to include the predicted content. We present evidence for pre-updating, and for its sensitivity to individual differences in WM, from two methodologies. The first experiment employed event-related potentials. Participants with high or low reading span read high- and low- constraint sentences. On a verb leading to a highly predictable noun, high-span participants showed increased P600 amplitude in the high (relative to the low) constraint sentences, followed by decreased P600 amplitude on the predicted noun (these words were separated by the Hebrew accusative marker). The results indicate integration of the noun prior to its occurrence, in participants with high WM capacity. The second experiment used event-based eye-blink rate (ebEBR). Research indicates that spontaneous eye blink rate is an effective indirect measure of dopamine activity in the striatum. Extensive work suggests that phasic dopamine from the dorsal striatum drives WM updating. In a recent ebEBR study, blink rate was shown to increase with increased WM updating requirements in a non-linguistic, reference-back task. However, to our knowledge this technic was never utilized in a language study. The current experiment used word pairs, with the first word either highly predictive of the next word (e.g. global warming) or not (e.g. vegetable soup). Lower ebEBR was elicited by the second word in the high (relative to the low) constraint pairs, indicating that updating was (at least in some trials) not needed, presumably as it was already performed on the first word. This effect showed a trend towards correlation with reading span. By providing corroborating evidence for pre-updating using ebEBR, we also hope to establish this technic as a valuable tool to study other linguistic phenomena.