Examining the Link Between Declarative Memory and Structural Language Ability in Children with Autism Spectrum Disorder (ASD)

Background:

It is well known that ASD is associated with a heterogeneous pattern of structural language ability that ranges from severe impairments in some individuals to superior skills in others. Yet the sources of this heterogeneity are unknown. Here we test the predictions that the language and learning impairments that distinguish low functioning autism (LFA) from high-functioning autism (HFA) most commonly result from an impairment of semantic memory additional to the impairment of episodic memory known to occur in all forms of ASD. More specifically, we hypothesize that whereas episodic memory (critically dependent on recollection) is impaired across the spectrum, semantic memory (critically dependent on familiarity) is impaired only in individuals with LFA. We also predict that impairments in semantic memory will correlate with lexical-semantic ability in children with LFA, when other relevant factors are controlled for.

Objectives:

The objective is to test specific predictions concerning the pattern of memory impairment in HFA and LFA and their relation to structural language ability in these groups (specifically lexical semantic knowledge).

Methods:

Study 1tested familiarity and recollection in 24 children (6-10 years) with HFA (VIQ>90), and 22 adolescents (11-17 years) with LFA (VIQ<75) using a shape recognition task and an action recall source memory task. Two comparison groups were included: a young typically developing (TD) group of 43 children (6-10 years) and a group of 21 intellectually disabled adolescents (11-17 years) without autism (ID), equated with the LFA group for age and ability.

Study 2 tested familiarity and recollection with the same participants as Study 1using forced choice recognition tests developed by Migo, Mayes et al. (2009) that discriminate between the processes of familiarity and recollection.

Participants also completed tests of lexical-semantic knowledge including the Pyramids and Palm Trees test and the Similarities and Vocabulary subtests from the Wechsler Abbreviated Scales of Intelligence. Of interest were not only group differences on the various dependant variables but also the relation between indices of familiarity and recollection and lexical semantic knowledge, whilst controlling for factors such as non-verbal IQ, mentalising ability, weak central coherence, and social economic status.

Results:

Results replicate earlier observations of impaired recollection but preserved familiarity in HFA compared to TD groups; but the LFA participants were surprisingly unimpaired compared to all other groups on the indices of familiarity. Interestingly the LFA group was superior on tests of weak central coherence and performance on this task correlated significantly with measures of familiarity and recollection particularly in this group. Moreover, we observed striking group differences in the pattern of correlations between lexical semantic knowledge and familiarity in both LFA and HFA groups but not in the ID or TD groups.

Conclusions:

LFAs' superior visual perceptual skills may compensate for impaired familiarity in tests of visual-spatial memory, as used here.  Lexical semantic knowledge may emerge through fundamentally different routes in LFA, ID, and HFA, each building on whatever cognitive resources are available. These findings serve to highlight the different pattern of cognitive ability in intellectual disability, with and without autism.