An fMRI Study of Socially Rewarded Learning in High Functioning Autism

Background: Social rewards, such as a smiling face, are often used to guide learning of appropriate behaviors during development. Computational modeling research suggests that an early deficit in reward-related learning may lead to aberrant development of joint attention and gaze following (Triesch et al., 2006), two principal characteristics of autism.

Objectives: To investigate how a social reward, a smiling face, guides implicit learning in high-functioning children with autism (HFA) and typically developing children (TD) using fMRI.

Methods: 16 HFA boys (12.4 + 2.14 years) and 16 age- and IQ-matched TD boys were scanned during a rewarded implicit learning task. Subjects were shown abstract visual stimuli and asked to respond with either a “1” or a “2” button press. Four of the 6 trial types were deterministic with respect to the outcome (1 or 2); the remainder were random. Feedback on half of the trials consisted of a social reward or punishment. The remaining trials were a neutral face with the same text. Subjects were given a post-scan test to determine if they were able to explicitly memorize the associations.

Results: Both groups performed at chance on the post-test, suggesting implicit rather than explicit learning. Task accuracy improved only in the TD group. Overall, TD children demonstrated greater activity than HFA in frontostriatal networks. During positive social rewards for deterministic trials, TDs had greater activity than HFA in dorsal and ventral striatum, anterior cingulate, and inferior frontal gyrus. Equivalent activation in bilateral fusiform gyrus during feedback (t(30)=0.706, p = 0.380) indicate that both groups processed the faces.

Conclusions: These results suggest that positive social rewards engage neural systems involved in reward processing in TD, but not HFA, children. It appears that TD children are able to incorporate this feedback to guide implicit learning whereas HFA children cannot.