Adults with ASD who do not also demonstrate language impairment generally obtain average or better vocabulary scores on standardized testing (Stothers & Oram Cardy, 2012). Evident semantic weaknesses are described as pragmatic or social in nature, but this group of adults with ASD also have difficulty with academic, non-social language tasks. The present study examined the possibility that semantic difficulties arise during the integration of unique semantic representations. Semantic integration involves the detection and elaboration of overlap between discrete representations to form a novel, higher-order relationship. As such, it is a form of gestalt perception – a relational process in which meaningful wholes are constructed from stimulus fragments that differ qualitatively from the larger whole. Low scores on nonverbal tests of gestalt perception have been reported in ASD, as has difficulty understanding verbal gestalts such as metaphors.
Objectives were to test the hypotheses that: a) adults with ASD without an accompanying language impairment nonetheless demonstrate weaknesses on tests that require the creation of a linguistic gestalt (e.g., hot + dog = food), b) such weaknesses affect the apprehension of both verbal and nonverbal gestalts, and c) reduced capacity for integrative processing in adults with ASD are evident in both cognitive and EEG data.
Adults with a typical developmental history were compared to adults with a community diagnosis of ASD on baseline measures of vocabulary, as well as semantic integration and nonverbal gestalt perception. Measures of semantic integration included remote associate problems (swiss, cottage, cake = cheese), Similarities (how are two objects or concepts alike?), and metaphor identification. Nonverbal tests included Block Design, a baseline measure of visual-spatial skill, as well as puzzle assembly, gestalt closure, and object identification. EEG data was collected during metaphor and object identification.
1) Adults with ASD had lower scores for language tests that required integration than their peers without ASD. 2) Similar results were obtained for nonverbal integration tests, and verbal and nonverbal gestalt perception scores were positively correlated across the sample. 3) Behavioural results were supported by differences in waveform latency and amplitude in participants with ASD and without.
ASD adults demonstrated low scores for verbal tests that required integration of unique semantic representations, and for nonverbal measures in which a novel whole had to be produced from unlike parts. Results could not be explained by differences in single word knowledge, nor by differences in visual-spatial ability as measured by Block Design; groups were not significantly different on either baseline measure. Time-locked responses to stimuli that required the formation of gestalts during EEG data collection were delayed for participants with ASD. Results supported the hypotheses that verbal semantic integration and nonverbal gestalt perception rely on a common set of processes, and a weakness in gestalt perception distinguishes adults with ASD from their typical peers.