Exploring the Nature of Neural Correlates of Language, Attention and Memory: Reliability and Validity Studies of Event Related Potentials

Abstract

Comparing data from different subfields of research may help in understanding emerging patterns and refining interpretations. This is especially true in neuroscience because brain functions can be studied at multiple levels of analysis, spatially and temporally, and with a variety of complementary measurement techniques. Within the ERP domain, several subfields of research have evolved over time, typically reflecting the specific time-window of interest and brain function investigated. The current investigation focused on three widely studied ERP effects reflecting a variety of key brain functions: the N400 effect, the P3b effect and the Left Parietal effect. The N400 effect has attracted researchers interested in language processing, the P3b effect researchers interested in attentional processes and the Left Parietal effect researchers focused on episodic recollection. Even though the ERP technology constitutes a common thread across these subfields, there is often a lack of communication across groups of researchers. The literatures on the N400 effect, P3b effect and Left Parietal effect have been written by relatively non-overlapping groups of researchers, and as such the kind of analysis carried out in the current thesis is not a common one, as it compares effects investigated within different subfields. Specifically, the approach taken in the current thesis involves assessment of the comparative reliability of the three effects of interest, and at the same time allowing refining their validity. Results showed that all three effects were found to be reliable at the group level and the N400 effect and the P3b effect were also found to be reliable at the single participant level. A correlational analysis involving all three effects yielded a significant correlation between the P3b and the Left Parietal effect but not between the P3b and the N400, or between the Left Parietal effect and the N400. Following up on the significant correlation, suggesting a convergence between the P3b effect and the Left Parietal effect, a probability manipulation of the Left Parietal effect was carried out to investigate if the old/new effect is sensitive to probability changes similarly to the P3b. The size of the Left Parietal effect was found to be sensitive to the relative probability of old and new items, in a manner consistent with the P3b effect‟s sensitivity to probability manipulations. The results pointing to a relationship between the P3b effect and the Left Parietal effect suggest that attentional processes sensitive to probability may temporally overlap and confound memory processes as indexed by the Left Parietal effect. The N400 effect, in the initial correlational study, was found to be independent from attentional processes as reflected by the P3b, and from episodic recollection as indexed by the Left Parietal effect. The validity of the N400 effect as a measure of semantic processing was then assessed by manipulating associative relationships while keeping constant semantic relationships, with results showing that the effect can be clearly modulated by associative changes when semantic relatedness is kept constant. The same association norms were then used in an old/new recognition experiment to assess if the Bilateral-Frontal old/new effect behaves in reaction to association relationships similarly or differently from the N400, in the attempt of assessing if the N400 is only a measure of associative relationships or also a measure of the process of familiarity. The observed pattern suggests independence between the N400 and the Bilateral Frontal effect. Overall, the N400 effect was found to be independent from memory processes occurring in the same time window, but, contrary to the dominant interpretation of the effect, the effect was modulated by changes in association strength while keeping semantic relatedness constant, suggesting that the N400 effect may be sensitive to a contiguity-based associative learning process not constrained to the linguistic domain.