Labels refer to object categories in Adults and 9-Month-Old infants

Friday, 2017, March 24 - 12:00
Department of Psychology, Lancaster University

Abstract

Even infants below one year of age display recognition and understanding of some words, but the nature of this skill is debated. We used electrophysiological methods to investigate (1) whether young infants' word knowledge reflects referential understanding, and (2) whether they expect that object labels refer to object categories that exist independent of being labelled. In the first study, mothers of 9-month-old infants introduced objects to their child by pointing gestures and labels. The objects then appeared from behind an occluder, either matching or mismatching the preceding label. We found a clear effect of object-label congruency in terms of an N400 event-related potential component, which is thought to reflect semantic priming. Thus, by setting up a live, ostensive context for referring to objects, we managed to demonstrate that semantic priming occurs in young infants, and that they expect that known words refer to specific object kinds. The second study addressed the question whether labels alone, without perceptual similarities, could make adults and 9-month-olds group objects together. We measured the desynchronization of alpha-band EEG oscillations in a category oddball paradigm. Adults learnt one of two pseudo-words for each of six unfamiliar objects without shared perceptual features. Subsequently, four of the six objects, three sharing the label and one having the other label, were presented without labels on screen, with equal frequency. Participants responded to the oddball category with stronger attenuation of alpha oscillation over the left frontal region. Similar response was found for known categories. Nine-month-olds were engaged in a live familiarisation with an experimenter presenting them the six unfamiliar objects one by one, while uttering the two novel labels. Right after the familiarisation, we presented them with the four objects the same way as above. Stronger alpha attenuation in response to the oddball category suggested that 9-montholds, just like adults, exploited the labels to form two object categories. Moreover, when infants were presented with objects that could be spontaneously grouped by visual similarities they produced disjointed electrophysiological responses for label-based and feature-based grouping. Our data strongly suggest that ostensive communication helps infants to interpret both labels as symbols referring to object categories.