Testing a dynamic account of neural processing: Behavioral and electrophsyiological studies of semantic satiation
MetadataShow full item record
In everyday perception, we easily and automatically identify objects. However, there is evidence that this ability results from complicated interactions between levels of perception. An example of hierarchical perception is accessing the meaning of visually presented words through the identification of line segments, letters, lexical entries, and meaning. Studies of word reading demonstrate a dynamic course to identification, producing benefits following brief presentations (excitation) but deficits following longer presentations (habituation). This dissertation investigates hierarchical perception and the role of transient excitatory and habituation dynamics through behavioral and neural studies of word reading. More specifically, the effect of interest is 'semantic satiation', which refers to the gradual loss of meaning when repeating a word. The reported studies test the hypothesis that habituation occurs in the associations between levels. As applied to semantic satiation, this theory supposes that there is not a loss of meaning, but, rather, an inability to access meaning from a repeated word. This application was tested in three behavioral experiments using a speeded matching task, demonstrating that meaning is lost when accessing the meaning of a repeated category label, but is not lost when accessing the category through new exemplars, or when the matching task is changed to simple word matching. To model these results, it is assumed that speeded matching results from detection of novel meaning to the target word after presentation of the cue word. This model was tested by examining neural dynamics with MEG recordings. As predicted by semantic satiation through loss of association, repeated cue words produced smaller M170 responses. M400 responses to the cue also diminished, as expected by a hierarchy in which lower levels drive higher levels. If the M400 corresponds to the post-lexical detection of new meaning, this model predicted that the M400 to targets following repeated cues would increase. This unique prediction was confirmed. These results were tested using a new method of analyzing MEG data that can differentiate between response magnitude versus differences in activity patterns. By considering hierarchical perception and processing dynamics, this work presents a new understanding of transient habituation and a new interpretation of electrophysiological data.