The Relationship Between Iconicity and Referential Understanding of Pictures in Low-Functioning Children with Autism

Background:  

Low-functioning children with autism (CWA) tend to form associative, rather than referential, relations between words, pictures and objects (Preissler, 2008).  This results in poor generalization of labels to pictured objects and may prevent CWA from understanding that words and pictures can represent categories. This has implications for the effectiveness of interventions such as the Picture Exchange Communication System (PECS). However, it is possible that referential understanding of pictures in CWA is facilitated by high levels of iconicity (the perceptual similarity between picture and referent), as is the case for young neurotypical children (NTC).

Objectives:  

We investigated the relationship between iconicity and referential understanding of pictures in CWA. We also examined whether iconicity influenced the ability of CWA to generalise labels from pictures to novel category members (i.e. differently-coloured versions of depicted entities).

Methods:

Seventeen CWA (M age = 9.6 years) were matched to 14 NTC (M age = 3.9 years) on receptive language (CWA M = 3.9 years; NTC children M = 3.7 years). In each trial, participants were taught a novel word (e.g. “Zepper”) repeatedly paired with a target picture of an unfamiliar object. They were then asked to identify the referent of the newly-learned word from arrays consisting of the target picture paired with the depicted object and then a novel category member. Target pictures were black-and-white line drawings (BWLD), colour line drawings (CLD), greyscale photographs (GP) and colour photographs (CP). Participants received four trials over 2 test sessions, with the iconicity of the target picture varying between trials. In another session, participants completed a preference task to rule out picture/object selection biases.

Results:

CWA most frequently selected the target picture as the referent of the learned label at all levels of iconicity (BWLD = 88%; GP = 76%; CLD = 65%; CP = 50%), while NTC selected the depicted object most often. Chi-square tests revealed significant relations between group and associative responding for BWLD (χ² = 14.07, p < .001) and GP (χ² = 5.24, p < .05), but there was no difference in response types for colour trials (CLD and CP). Regardless of iconicity, CWA were unlikely to generalise a label from a picture to a novel category member (BWLD = 24%; GP = 12%; CLD = 35%; CP = 25%). In contrast, NTC most frequently identified the differently-coloured object as the label’s referent at all levels of iconicity. Both CWA and neurotypical children responded correctly on 98% of preference task trials, eliminating picture/object biases as explanations for between-group differences.

Conclusions:

CWA are significantly more likely to recognise referential relations between pictures and objects that are colour-matched. Although they may extend a label from a highly-iconic colour picture to its specific referent, they will not generalise the label to a novel category member, unlike NTC. These findings suggest that interventionists should use colour pictures when implementing PECS, but it is important that recipients are exposed to pictures of differently-coloured category members in order to promote category formation and facilitate generalisation of labels.