| Hebb (1949) first envisioned neural processing in terms of local cell assemblies. Such networks, using reciprocal connections, can reverberate, or oscillate, in response to an input such as a visual stimulus. The presence of such networks may be inferred if two or more brain areas show a high degree of phase synchrony at a consistent time interval after the stimulus. This experiment tests the hypothesis that distinct cell assemblies are transiently formed and dissolved at different times depending on the language processing demands. Specifically, a visual lexical decision task should elicit measurable differences in neural processing between words and non-words, and between open and closed class words. Open class words, such as nouns, are highly imageable, and should show widespread activation and phase synchrony at multiple electrode locations in both hemispheres across visual and motor areas. However, abstract, low-imageable, closed class words such as prepositions should activate a smaller cell assembly in left hemisphere (Pulvermuller, 1999). The time-frequency representation of neural activity may show differences in the beta or gamma band, since these frequencies have been shown to index language processing. Electroencephalograms (EEG) were recorded while subjects performed a lexical decision task. Stimuli were further divided into open class or closed class words, and pronounceable or non-pronounceable non-words. Time-frequency and phase synchrony was calculated using wavelet decomposition. The results do not support a model of lexical decision based on transient cell assemblies, however, there is evidence to suggest that words show differences in activity in the 8--12 Hz range relative to non-words. These results indicate a possible role for alpha band activity in word processing. |
