INCIHUSA   20883
INSTITUTO DE CIENCIAS HUMANAS, SOCIALES Y AMBIENTALES
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
An EEG study of Entropy and Surprisal during artificial grammar processing
Autor/es:
ÁNGEL TABULLO; CHRISTOPHE PALLIER; ALEJANDRO WAINSELBOIM; ASAF BACHRACH
Lugar:
New Orleans
Reunión:
Congreso; 52nd Annual Meeting of the Society for Psychophysiological Research; 2012
Institución organizadora:
Society for Psychophysiological Research
Resumen:
Prediction seems to be a central mechanism in multiple cognitive domains and in particular in language. A wide range of behavioral and electrophysiological paradigms have quantified the correlates of prediction during parsing. The main correlate of prediction that has been studies is prediction error or surprisal, or the degree to which an actually occurring event fits previous predictions. In the ERP literature the P300 and N400 components have been often associated with surprisal. Another correlate of prediction, albeit   far less studied, is the level of uncertainty (or entropy) regarding an upcoming event. In this experiment, we used a probabilistic artificial grammar to create different levels of both surprise and entropy regarding the identity of a specific syllable (bigram token prediction) and regarding the choice of grammatical rule. The subjects performed a repetition detection task  on a corpus instantiating the probabilistic grammar while being monitored using a 19 channel EEG system. Reaction Time analysis of the repetition detection task showed a significant effect of both token and rule entropy but no effect of surprisal. This result demonstrates that subjects have extracted a probabilistic representation of both bigram transitions and more abstract  rules containing variables. The results suggest that uncertainty might represent a more important factor than surprisal. ERP analysis showed a posterior negativity which was larger in high entropy trials. The latency and topography of this component were similar to the N400 found after unexpected lexical items