Proceedings of the Annual Meeting of the Cognitive Science Society

Symbolic associations in human children and adults are based on forming equivalence classes which include three mainrelations between the tokens. 1) A = A (Reflexivity), if 2) A –¿ B and B –¿ C then A –¿ C (Transitivity) and 3) if A –¿B then B –¿ A or Symmetry (1). Extensive studies on non-human primates have demonstrated success in Reflexivity andTransitivity in several species but a consistent failure in Symmetry in any given association. Comprehension of symmetryof an association can be a key contribution to linking abstract words to their corresponding tokens and later on in couplingwriting forms of words to their spoken form (2). However to our knowledge it hasnt been investigated whether infants arecapable of spontaneously reversing the direction of an association to any extent. In two EEG studies we investigated if4.5-month-old infants are capable of applying symmetry in the context of word-learning.In the first study we trained 2 groups of 25 infants, to two pairs of word-categories (bird or vehicle). At each trial infantswere presented with a word and an image. The critical consideration was to introduce a 1 s of SOA between the two stimuli.In one group infants were trained on words always preceding the images (Word-Image group) and in the other group infantswere trained on the opposite direction (Image-Word group). In the test blocks 70% of trials were as in the training andthe other 30% were either with the incongruent trials in the original direction or the congruent and incongruent trials inthe reversed direction. We observed significant cluster of electrodes, mainly in the right temporal, in both the trained andreversed directions while contrasting the congruent and incongruent conditions, with the word-image group showing astronger effect.In a 2nd experiment, designed as a comparative study between infants, adult humans and adult macaques, we sought totrain each participant on 4 pairs of word-images, 2 pairs following a word-image direction and the other 2 an image-worddirection, with a 1s SOA between the two stimuli similar to experiment 1. In this experiment the infants attended thetraining phase at home prior to the experiment through three YouTube videos on three consecutive days and on the testday, they were being tested either on the trained or the reversed direction of each single pair in a similar ERP design as instudy 1. The results in a group of 54 4.5-month-old infants follow the pattern of results in study 1 that infants show an earlyas well as a late surprise effect relative to the onset of the second stimulus of the trial, while contrasting the incongruentversus congruent trials in both directions. Furthermore we utilized frequency tagging in both studies as an extra measureto compare the conditions of interest. The overall results suggest that contrary to the consistent failure of non-humananimals, infants can readily learn an association in a bi-directional manner, which can be suggestive of an early access totheir symbolic system.