The ‘Light from Above' Prior Is Intact in Children on the Autism Spectrum

Thursday, May 12, 2016: 5:30 PM-7:00 PM
Hall A (Baltimore Convention Center)
A. E. Croydon1, T. Karaminis1, L. E. Neil1, D. C. Burr2 and E. Pellicano1, (1)Centre for Research in Autism and Education (CRAE), UCL Institute of Education, University College London, London, United Kingdom, (2)CNR Institute of Neuroscience, University of Florence, Pisa, Italy
Background: Sensory information entering the retina is inherently ambiguous. The brain makes sense of it by anticipating or predicting the sensory environment based on prior knowledge. Some authors have proposed that this predictive process may be atypical in autism, in that internal assumptions, or priors, may be under-weighted or less utilized than in typical individuals (Pellicano & Burr, 2012).

A robust internal assumption used by adults is the light-from-above prior (Sun & Perona, 1998), in which shading patterns on an object are interpreted as if the light source is located above (and slightly to the left) of an object, even when shading information is consistent with alternative light source locations.


We investigated whether children with autism use prior information to estimate the shape of an object, that is, whether they show the so-called ‘light-from-above prior’ to the same degree as typical children.

Methods:  A group of autistic children (n=17) and a group of typical children (n=27), all aged 7 to 13 years, and matched in terms of age and intellectual ability, took part. Following Andrews et al. (2012), children were asked to judge the shape of a 7 hexagon stimulus – whether it appeared concave or convex. Within the context of a developmentally-appropriate game, they decided whether a bee should fill the cell with honey (if the stimulus was perceived to be concave) or not (if perceived to be convex). Twelve orientations of the hexagon stimulus were presented in a randomised order. Children completed 120 trials across 3 blocks. The testing room was lit only by the computer monitor to minimise environmental lighting cues.

Results:  The relation between the proportion of convex judgements (‘no’ answers in our game) and stimulus orientation was estimated for each child using a multivariate logistic regression. The light source location most consistent with those judgements was then calculated for each child following the procedure in Andrews et al. (2012). Children’s assumed light source direction ranged from -30.37˚ to -43.12˚. There were no significant group differences, t(42) = -1.04, p= 0.30. 

Conclusions:  Contrary to our expectations (cf. Pellicano & Burr, 2012), we found no significant differences in assumed light source position between autistic and typical children. Children on the autism spectrum seem to use prior assumptions to make sense of 3-dimensional information to the same degree as typically developing children. Future research should examine whether this prior is as adaptable (i.e., modifiable with training) in autistic children, as it is in typical adults (Adams, et al., 2004, 2010).