Social and Non-Social Memory In Autism: Delineating the Role of the Hippocampus and Amygdala

Background: Studies of memory abilities in ASD have consistently shown a discrepancy between difficulties in episodic memory for personally experienced events, relative to a preserved semantic memory for facts (Boucher and Bowler, 2008). How may we explain these findings? Are autobiographical memory deficits simply the extreme of a graded scale of information-processing deficits (Happe and Frith, 2006; Minshew and Williams, 2007), or do they present a qualitatively independent impairment in the social domain (Baron-Cohen et al., 2000, Hobson et al., 2006)? Further, on the neurobiological level, can they be traced to structural differences in the hippocampus or amygdala (Salmond et al., 2005), which are associated with memory in typically developing individuals? Previous studies of memory abilities in autism have generally compared simple semantic recall with complex episodic recall (Crane and Goddard, 2008; Losh and Capps, 2003; Klein et al., 1999), thus precluding the possibility of directly testing each alternative hypothesis. The present study aims to disentangle the underlying components of episodic and semantic memory, extending previous studies through carefully controlled behavioral measures, and through voxel-based morphometry of the hippocampus and amygdala.

Objectives: (1) To disentangle the relative importance of information-processing deficits from self- and social-processing deficits in ASD using episodic and semantic memory tasks matched for level of complexity. (2) To determine whether deficits in memory in autism can be traced to abnormalities in the hippocampus or amygdala.

Methods: Participants included 34 8-18 year-old subjects with ASD and 35 age, sex and IQ-matched controls. Autism diagnoses were confirmed using ADOS (Lord et al., 1999) and ADI-R assessments (Lord et al., 2003), and all subjects completed the Social Responsivity Scale (SRS) (Constantino, 2000) and Social Communication Questionnaire (SCQ) (Berument et al., 1999). Sub-tests from the NEPSY-II were used to obtain standardized measures of social and non-social semantic memory and organizational ability. The experimental tasks included a Levels-of-Processing task (based on Toichi et al., 2002) and a narrative recall task (based on Crane and Goddard, 2008) comparing memory for self, mother and favorite fictional character. A subset of 20 ASD subjects and 19 TD controls were scanned using Magnetic Resonance Imaging, and hippocampi and amygdalae region of interest volumes were measured using voxel-based morphometry.

Results: The TD group performed overall higher than the ASD group in all memory conditions (p=0.009); with post-hoc tests revealing a significant difference in semantic memory (p=0.001), memory for self (p=0.013), and memory for mother (p=0.015). Further, memory for self was found to be positively correlated with a test for organizational ability (r=0.353; p=0.003), yet did not correlate with a test for social memory (r=0.208; p=0.087)
Further analyses are underway to determine the relation between these behavioral results and anatomical regions of interest.

Conclusions: These preliminary results suggest that the pattern of memory abilities in autism may be driven by a broad deficit in information-processing, rather than a specific deficit in social information-processing. Further analyses will determine the relation between these behavioral memory patterns and brain structure.