Purchase this article with an account.
J. A. Slemmer, N. Z. Kirkham, S. P. Johnson; Visual statistical learning in infancy. Journal of Vision 2001;1(3):25. doi: https://doi.org/10.1167/1.3.25.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Research on implicit learning has demonstrated that adults discover the statistical structure underlying a continuously presented stream of auditory stimuli (Reber, 1989). Infants, likewise, have been found to use transitional probabilities in an auditory stream to segment speech sounds into constituent groupings, despite the absence of other boundary cues (e.g., Saffran, Aslin, & Newport, 1996). We posit that this type of statistical learning during infancy is not limited to the auditory domain but may also be used to detect statistical regularities inherent in the visual environment. We used an habituation paradigm in which infants viewed a sequence of visual stimuli until looking declined, followed by test displays in which the ordering of individual stimuli either preserved or violated the habituation sequence. We reasoned that longer looking at the random order would reveal discrimination of the ordered and random sequences, given that infants typically exhibit novelty preferences after habituation. Two-, 5-, and 8-month-old infants were habituated to a continuous sequence of three color-shape pairs. There were no cues to boundaries between pairs except transitional probabilities. During test, infants were shown alternating trials of either the same structured sequence or a completely randomized sequence. Significantly longer looking times to the random sequence were seen across all age groups, suggesting that the infants encoded the transitional probability structure seen during habituation. From these results, we suggest that a domain general mechanism may be implicated in some early learning. Future directions include exploring the importance of different stimulus features to learning of visual statistical probabilities.
This PDF is available to Subscribers Only