Distinct Brain Regions Associated with Item and Relational Encoding Impairments in ASD

Background: The human brain is adept at storing and retrieving information from experienced events, and this ‘episodic memory’ system is critical to adaptive functioning in everyday life. Individuals with autism spectrum disorders (ASD) demonstrate a marked impairment in episodic memory, but the neurocognitive bases of this impairment remain unclear. The Relational and Item-Specific Encoding (RiSE) task was developed to differentiate the abilities to remember specific stimuli (item-specific encoding) and associative memory for relationships between stimuli (relational encoding). In a recent study, individuals with ASD showed impaired performance on the RiSE, particularly in the item-specific encoding condition (Solomon, McCauley, Iosif, Carter, & Ragland, 2016).

Objectives: The present study was designed to further investigate item-specific and relational encoding in ASD, and to use functional magnetic resonance imaging (fMRI) to elucidate the neural bases of episodic memory impairments in this population. 

Methods: Data collection is ongoing, but currently 17 young adults and adolescents with ASD and 19 typically-developing participants (TYP) have completed the RiSE in an fMRI environment. The RiSE consists of two phases: i) an encoding phase where items are presented in pairs while participants are instructed to attend to the items themselves (“is one of the items living?”) or to the relationship between items (“can one item fit inside the other?”), and ii) an item recognition (IR) phase where studied items are intermixed with unstudied foils and participants judge whether each item is “old” or “new”. IR accuracy was analyzed using a 2 (encoding condition) x 2 (diagnosis) ANOVA, and preliminary fMRI data were analyzed using a mixed-effects model (FSL’s FLAME1, uncorrected p<.01) and a region-of-interest mask derived from 270 ‘episodic memory’ studies in the Neurosynth data archive.

Results: Behaviorally, we observed significant improvements in accuracy following relational encoding relative to item encoding (p<.001, h²=0.61), and a generalized impairment across encoding conditions in the ASD group (Item: p=.03, d=0.76; Relational: p=.04, d=0.72), with no evidence of an interaction (p=.67). Preliminary fMRI analyses revealed that item and relational encoding impairments were associated with distinct patterns of brain activity. In both the ASD and TYP groups, encoding recruited medial temporal lobe (MTL) and dorsolateral prefrontal cortex (dlPFC), with dlPFC demonstrating greater recruitment during relational vs. item encoding. Between groups, relational encoding was associated with greater recruitment of dlPFC in TYP relative to ASD, but similar levels of MTL recruitment. In contrast, the TYP group demonstrated greater MTL recruitment in the item-specific encoding condition than the ASD group.

Conclusions: The present study provides further evidence for an episodic memory impairment in ASD, which was observed regardless of whether information was encoded in an item-specific or relational fashion. Furthermore, the study provides novel fMRI evidence that distinct MTL and PFC subregions may be associated with item and relational encoding impairments in ASD, respectively, which could lead to a better understanding of the neural mechanisms underlying different types of memory failures in ASD. As data collection is ongoing, rigorous methodological standards for statistical power and multiple comparisons correction will be implemented prior to IMFAR 2017.