Neural mechanisms for Stimulus equivalence and artificial grammar learning: an ERP study

TABULLO, ÁNGEL; YORIO, ALBERTO; SEGURA, ENRIQUE; ZANUTTO, SILVANO; WAINSELBOIM, ALEJANDRO

Punta del Este

Congreso; VI Congreso del Capítulo Latinoamericano de la International Federation of Clinical Neurophysiology; 2011

International Federation of Clinical Neurophysiology

IntroductionRelational frame theory (Hayes et al.,2001) proposes that human language can be defined in terms of derived stimulus relations and relational networks. The simplest derived stimulus relations is the stimulus equivalence (SE), which has been proposed as the basis for semantic meaning in language (Sidman, 1994). The fact that processing of SE relations generates an N400 component (typically associated with semantic processes) supports this claim.Artificial grammar learning (AGL) tasks have been used to study acquistion of human syntax (Petersson et al., 2010). AGL processing has been shown to generate ERPs similar to P600 (typically associated with syntax processing) (Bahlmann et al., 2006).The posibility of overlapping neural mechanisms for SE relations and AG learning has not been addressed yet.Objectives To study and compare the EEG correlates of SE and AGL processing.Methods SE Task) 21 healthy subjects were trained with Matching-to-Sample procedures to form two 3-member equivalence classes, using artificial words as stimuli. In the test stage, subjects were presented with pairs of stimuli which could be related by equivalence (members of the same SE class) or unrelated, and were instructed to classify them. AGL Tasks) The same subjects were trained in an artificial grammar by exposure to grammatical sequences of artificial words. In a test stage, new sequences were presented, half correct and half containing violations to the AG rules. The task was to classify the sequences.Results The unrelated stimulus pairs in the SE test generated a centro-parietally distributed N400-like component (F(1,15) = 4.962; p = 0.042). On the other hand, violations of the artificial grammar generated a posteriorly distributed P600, and an earlier posterior negativity around 400 ms. The negativity observed in both tasks was compared with an ANOVA, which showed that the effect was more negative for the SE task at central and posterior regions (F(2,2) = 20.308; p < 0.001).Discussion violations of SE relation and AG rules elicited an N400 ERP, which indicates a partical overlap in the neural correlates of these processes. However, this effect was more prominent of SE violations. Additionally, a P600 was observed in the AGL task.