A Functional MRI Meta-Analysis of Reward and Social Motivation Studies of ASD

Background:  The Social Motivation Hypothesis posits that individuals with autism spectrum disorder (ASD) show diminished social motivation and responsiveness to social rewards (Chevallier et al., 2012). Two types of motivation typically distinguished are the “liking” of the reward that occurs during its consumption, and the preceding “wanting” of the reward (Berridge, Robinson, & Aldridge, 2009). Per the Social Motivation Hypothesis, people with ASD may ‘like’ and thus ‘want’ social interactions less, leading to fewer social learning opportunities and diminished social skills. Functional MRI studies can distinguish between ‘wanting’ and ‘liking’ a reward since the two processes are somewhat distinct in timing and neurobiological circuitry. However, it has been difficult to draw strong conclusions as the few studies on this topic have had limited sample sizes. For this reason, a meta-analysis synthesizing the current data may provide needed clarity.

Objectives:  To use fMRI-specific meta-analytic techniques to 1) obtain information about the direction and magnitude of differences in reward processing in individuals with ASD relative to typically developing controls (TDCs), and 2) identify potential moderators of observed differences.

Methods:  We systematically reviewed the fMRI literature on individuals with ASD and identified nine papers meeting inclusion criteria for meta-analysis. We leveraged novel fMRI meta-analytic methodology (Effect Size Signed Differential Mapping; Radua et al, 2012) to account for effect magnitude, effect direction, sample size, and covariates (age and IQ), since study samples showed a large range in mean ages (12-37 years) and IQs (103-127).

Results:  Activation differences were observed in the ventral striatum and other regions (hypoactivation in the left nucleus accumbens (-6,6,-6), posterior cingulate gyrus (4,-6,30), right caudate nucleus (10,6,20), and right frontal pole (28,64,12), and hyperactivation in the hippocampus (26,-14,-34); all p’s<.001) during ‘wanting’ of monetary but not social reward in individuals with ASD relative to TDCs, and no significant differences in ‘liking’ of reward. When we statistically controlled for the age of each sample in exploratory meta-regression, these results were no longer statistically significant, suggesting potentially important age effects in reward processing that have not been adequately accounted for to date. In analyses covarying mean IQ, the left nucleus accumbens and right anterior hippocampus results were preserved and the posterior cingulate gyrus result shifted dorsally (toward anterior cingulate gyrus).

Conclusions:  These results suggest different processing of non-social reward during the ‘wanting’ but not the ‘liking’ phases of reward processing in individuals with ASD. The challenges of providing naturalistic social rewards in the lab, compared to providing actual monetary rewards, may contribute to the observed absence of differences in response to social reward. In addition, the effect for group differences in social reward conditions may be difficult to detect with the sample sizes traditionally used in this field. Exploratory meta-regression results highlight key differences between the ages and IQs of study samples that may explain contradictory results in the literature. Our results suggest that careful examination of current paradigms and systematic investigation of reward processing in ASD at different ages may be necessary to fully understand social motivation in autism.