Are we all born synaesthetic? Examining the neonatal synaesthesia hypothesis.

The popular claim that humans are born in a state of 'buzzing confusion' (James, 1890) can be explained either in terms of a lack of connection between sensory outputs which remain isolated, or, on the contrary, as a lack of sensory specificity leading to indistinct perceptual experiences. If both hypotheses seem to make sense of the evidence collected in early infancy, cases of synaesthesia and of arbitrary-looking crossmodal associations observed later in adults have increasingly been taken to support the latter view, known as the 'neonatal synaesthesia' hypothesis (e.g., Maurer, 1993; Maurer et al., 2012; see also Cohen Kadosh et al., 2009a,b; Ludwig et al., 2011). In this paper, we argue that there is neither a good reason to treat the most arbitrary cases of crossmodal associations, such as the correspondence between pitch and visual brightness, as providing evidence for the existence of an initial synaesthetic stage, nor to relate them developmentally to the synaesthesia that has been documented in adults. We stress that the idea that such crossmodal associations prevail at the earliest stages of human development and progressively, but not totally, disappear through exposure and learning remains a speculative or hard-to-pin down thesis, one that rests on a series of individually weak pieces of evidence. The developmental patterns of these correspondences are better explained by their being learned, and the main question they raise concerns the specificity of the learning mechanism(s) that underlies their acquisition. In conclusion, the arbitrariness of a sub-set of crossmodal associations observed in infants and/or adults is certainly only apparent or superficial, which means that it does not provide a sufficient reason to posit an innate synaesthetic confusion, and is instead compatible with most recent models of multisensory development.