The Relation Between Classical Conditioning and Prototype Learning in Individuals with ASD

Background:  

Studies of implicit learning have indicated a mixed pattern of impaired and intact performance in individuals with ASD (Eigsti & Mayo, 2011).  Several studies of implicit learning in ASD have indicated a relationship between implicit task performance and verbal and non-verbal reasoning abilities (Gastgeb et al, 2012; Klinger et al., 2007).  Taken together, these findings suggest better performance on implicit learning tasks may be directly related to better reasoning in individuals with ASD.

Objectives:  

The primary objective of this study was to examine the relation between implicit learning and verbal reasoning in ASD across two implicit learning tasks. Participants participated in a classical fear conditioning and prototype learning task.

Methods:  

Twenty children with ASD and 17 age- and IQ-matched children with typical development (ages 7 to 14 years) completed two implicit learning tasks.  The conditioning task consisted of three phases (habituation, acquisition, and extinction).  During acquisition, one color (i.e., the conditioned stimulus) was paired with a loud noise (i.e., the unconditioned stimulus).  Skin conductance responses were recorded. The prototype task consisted of a familiarization and test phase. Eight different members from a category were presented during the familiarization phase. The test phase consisted of old, new, and prototype (i.e., mathematical average of category members) test trials and participants made “yes/no” recognition judgments. Prototype learning was measured as false recognition of the prototype animal which, in fact, was not shown during familiarization. Verbal Reasoning was measured by the vocabulary and similarities sub-test of the WASI (Wechsler, 2000).

Results:  

Linear regression analyses were conducted to examine the relation between verbal reasoning and diagnosis to both classical conditioning task performance (CCTP) and prototype learning task performance (PLTP). For each model, verbal reasoning, diagnosis, and a diagnosis by verbal interaction term were hierarchically included.  Diagnosis significantly predicted CCTP, F(1, 34)=11.18, p<.01, however no significant effect was found for verbal reasoning or the interaction with diagnosis (p’s>.42).  In contrast, verbal reasoning significantly predicted PLTP (F(1, 34)=8.32, p<.01), with a small effect of diagnosis, F(1, 34)=3.06, p=.09, and a small interaction between verbal reasoning and diagnosis, F(1, 34)=2.85, p=.10.  A final regression analysis examined the relationship between PFTP and CCTP.  Using PLTP as the dependent variable, CCTP, diagnosis, and diagnosis by CCTP interaction were hierarchically included in the model.  Results indicated a significant diagnosis by CCTP interaction, F(1, 33)=5.00, p=.03, and no significant effects for CCPT or diagnosis (p’s>.20) showing that CCTP and PLTP were highly related for participants with ASD but not for participants with typical development.

Conclusions:  

The present findings observed impairments in two implicit learning tasks in individuals with ASD. A marginally significant interaction between diagnosis and verbal reasoning was also seen in the prototype task (but not the conditioning task) suggesting explicit reasoning may play a role in implicit learning for individuals with ASD but not for individuals with typical development.  Conditioning and prototype task performances were highly related for participants with ASD but not typically developing participants suggesting that these tasks tap common characteristics only for those with ASD.