Impaired Classical Conditioning in Persons with Autism Spectrum Disorders

Background:

Research has indicated that individuals with autism spectrum disorder (ASD) have difficulty with implicit or automatic learning.  The present study utilized a classical fear conditioning paradigm to examine implicit associative learning (i.e., classical conditioning) in individuals with ASD.   Previous studies examining associative learning in ASD have found both intact (Bernier et al., 2004) and impaired learning (Gaigg & Bowler, 2007).  However, both studies examined associative learning across modalities; pairing a visual conditioned stimulus (CS) with an auditory unconditioned stimulus (UCS).  To date, no study has examined associative learning both across modalities (visual CS with auditory UCS) and within a modality (auditory CS with auditory UCS). 

Objectives:

The primary objective of this study was to assess individuals with ASD associative learning across and within modalities.  We predicted that if individuals demonstrated impaired learning across modalities, but intact learning within a modality, this would be consistent with functional connectivity theories of ASD.  However, if individuals with ASD demonstrated impaired learning both across and within modalities, this would suggest more general associative learning impairments

Methods:

Fifteen high-functioning young adults diagnosed with ASD and 16 age- and IQ-matched individuals with typical development were presented with both a visual color (CS visual) and an instrument sound (CS auditory) paired with an aversive sound (UCS).  Three neutral visual stimuli and three neutral auditory stimuli were also presented.  Participants’ skin conductance responses (SCRs) were recorded.  After 40 trials in which the CS and UCS were paired, we examined whether participants showed elevated SCRs when the UCS (aversive sound) did not follow he CS.  This elevation in SCRs provided evidence of learning.  Following the conditioning task, an explicit memory test examined awareness of the learning contingences. 

Results:

Individuals with typical development displayed greater learning than individuals with ASD across both the visual and auditory conditions, F(1,29)=7.90, p <. 01. Participants with typical development demonstrated a large learning effect, F(1,15)=32.13, p < .01, η_p²=.682, whereas participants with ASD did not show significant learning,  F(1,14)=2.31, p =.15, η_p²=.142.   These results suggest individuals with ASD have a general impairment in associative learning. Additionally, a significant interaction between pairing and explicit memory, F(1,13)=8.13, p=.01,was found for individuals with ASD, but not individuals with typical development suggesting that only those individuals with ASD who explicitly learned the contingencies showed reliable associative learning.

Conclusions:

Results demonstrated impaired learning across both visual and auditory modalities, suggesting that individuals with ASD may have a general impairment in associative learning. Additionally, we found that greater explicit awareness was related to greater associative learning among individuals with ASD. These findings are consistent with the claim that individuals with ASD have impairments in implicit, associative learning and may compensate for these impairments by employing explicit learning strategies.  Given the importance of learning by automatic associations and its ubiquitous role in early learning, our findings provide important evidence for basic learning impairments in autism and the importance for targeting associative learning in early intervention strategies.