Sentence relevance and memory retrieval in discourse comprehension: N400 and LrNERP effects

Abstract

The present study explored the effect of different degrees of relevance in discourse comprehension by using ERPs analysis. A principle of pragmatic relevance is supposed to guide inferential mechanism underlying discourse processing. Discourse level comprehension needs a system of predictions about which information is more relevant in order to process the ongoing sentence meaning. This system should construct a specific mental model, where inferences related to the present sentence are stored and maintained. Three degrees of relevance of a new information (new sentence) with respect to an old information (target sentence) were manipulated: directly relevant; indirectly relevant; not relevant. Twenty-one subjects participated to the experiment and they were asked to try to comprehend a set of two paired sentences (old-new paired sentences) based on their conceptual relevance. Two negative deflections, peaking respectively at about 410 msec post-stimulus (N400), more right anterior-centrally distributed, and at about 550 msec (late negativity, LrN), more right central localized, were found. Repeated measures ANOVA found that the amplitude of both the N400 and LrNis modulated by the degree of relevance and by the strength of the underlying associations between the two sentences. Indirect relevance resulted in increased negativities in comparison with direct relevance. Contrarily, non-relevant condition did not produce an increasing in N400 and LrNamplitude. Unrelevance of the knowledge related to the actual mental model of sentences may induce a rapid and costless discarding of non pertinent information. The conclusive inference is that a subset of neural processes responding to degree of relevance of information is separable and cortically more frontally and centrally localized. Functional differences between N400 and LrN for relevance were discussed.