Conseptual relevance in discourse processing and memory retrieval. N400 ERP modulation

Discourse level and inference sentence-related need a system of predictions about what information is more relevant to process the ongoing sentences. This system should be able to construct specific mental model where information related to present sentence are stored and maintained. Previous model stated that this kind of inference is based on a principle of pragmatic relevance. The present study explored the effect of different degrees of relevance in discourse comprehension by using ERPs analysis. Three levels of relevance of a new information (new sentence) for an old information (target sentence) were manipulated: directly relevant; indirectly relevant; unrelevant. Twenty-one subjects participated to the experiment and they were asked to try to comprehend the set of two paired sentences (old-new paired sentences) based on their conceptual links. A N400 ERP effect was found, peaking at about 420 msec post-stimulus and more anteriorly distributed. Repeated measures ANOVA found that the amplitude of the N400 is modulated by the degree of plausibility of a conceptual dependency and the strength of the underlying associations: indirect relevance resulted in an increased negativity in comparison with direct relevance. Contrarily, unrelevant condition did not produce an increasing in N400 amplitude. The whole independence of this unrelevant information from the knowledge related to the actual mental model may induce a rapid and costless elimination of unrelavant information. The conclusive inference is that a subset of neural processes responding to degree of relevance of information is separable, and that it is directly related to the degree of discourse relevance.