Decoding Abstract Picture-Referent Relations: Are Low-Functioning Children with Autism Naïve Realists?

Background:

Low-functioning children with autism (CWA) typically have an impaired understanding of intentionality and a strong tendency to focus on localised perceptual elements when processing visual stimuli. These deficits may impact on their ability to decode non-iconic pictures that do not clearly resemble their intended referents (e.g. abstract art, children’s drawings etc). Neurotypical 3-year-olds spontaneously use intentional cues (e.g. artist’s eye gaze) to determine the referent of an ambiguous pictorial representation, however CWA may instead decode such pictures exclusively in terms of their appearance and thus demonstrate a unique route to picture processing. 

Objectives:

Using a modified version of Bloom and Markson’s (1998) “Size Task”, we investigated whether picture processing in CWA conforms to the theory of “naïve realism”, which contends that pictures represent whatever they look most like to the viewer, irrespective of the artist’s intentions.

Methods:

Participants were 14 CWA (M age = 9.8 years) and 14 neurotypical children (M age = 3.9 years) matched on receptive language ability (CWA M = 3.9 years; neurotypical children M = 3.7 years). In the first session, they were presented with pairs of differently-sized ‘abstract’ pictures (e.g. a small and a large circle) supposedly drawn by a child with a broken arm, and were asked to identify the picture that represented either a small or a large named referent (e.g. elephant or mouse). They were then asked to select the 3-dimensional object that the artist had attempted to depict from an array consisting of the intended referent (e.g. a model elephant), an object that resembled the abstract picture (e.g. a ball) and a distracter. A second test session confirmed whether children could simply match iconic pictures to their referents, to rule out potential task demands. Participants completed 4 trials in each session.

Results:

In the first session CWA used relative size to infer correct picture-referent relations in 75% of trials, a rate significantly greater than chance (t = 3.02, p < .01). Neurotypical children selected the correct picture in 94.6% of trials. When asked to identify an abstract picture’s 3-D referent, CWA selected the object that resembled the picture in 61% of trials and the intended referent in just 30% of trials. Conversely, neurotypical children selected the intended referent and the perceptual referent in 79% and 14% of trials respectively (a significant Group x Response Type interaction, (F = 23.33, p < .001). In the second test session, when the pictures resembled their intended referents, both groups performed at ceiling when asked to identify a target picture and select the intended 3-D referent.

Conclusions:

CWA displayed a surprising ability to infer correct picture-referent relations in the absence of perceptual resemblance. Whilst this could be evidence for intentional reasoning in CWA, we suggest that their success is more likely to be driven by non-intentional problem solving. This conclusion is corroborated by their 3-D referent selections, which indicate that CWA form relations between pictures and objects based on perceptual resemblance rather than referential intent, thus supporting the claim that CWA are naïve realists.