Note: Most Internet Explorer 8 users encounter issues playing the presentation videos. Please update your browser or use a different one if available.

Assessing the Effect of Reward On Learning: A Novel Eye-Tracking Marker of Treatment Outcome

Friday, 3 May 2013: 18:00
Chamber Hall (Kursaal Centre)
C. McCormick1, G. S. Young1 and S. J. Rogers2, (1)UC Davis MIND Institute, Sacramento, CA, (2)Psychiatry and Behavioral Sciences, MIND Institute; University of California at Davis, Sacramento, CA
Background : Autism symptoms may develop in part due to diminished salience of social stimuli (Dawson et al., 2004; Dawson, Webb, & McPartland, 2005; Mundy, 2003).  Many intervention programs, including the Early Start Denver Model (ESDM), focus on social motivation and social engagement, with the goal of changing the reward value of social stimuli.  

Objectives : The goal of this study was to develop an eye-tracking task sensitive to the effects of social and non-social reward on learning that could discriminate differences in early ASD and typical development.   

Methods : To date, 17 children with typical development (TD; 12 male; mean age 48.71 (4.04) months) and 16 children with ASD (13 male; mean age 49.44 (4.37) months) have completed the protocol.  All children with ASD had received a diagnosis at two years, prior to the study, and intensive behavioural intervention.  In the paradigm, children were rewarded with an interesting video (social or non-social) when they fixated on the same target location located in a circular array of six identical blue dots.  Learning criterion was defined as first fixation to the target for three trials out of six within a block, after which a new block of trials began with a new target location and reward.   Learning was measured by number of trials per block, where fewer trials indicate better performance.  Thus, participants had to learn the overall paradigm, as well as the new location for each block. 

Results : Multilevel models were fit using the MIXED and NLMIXED procedure in SAS (Littell, Miliken, Stout, & Wolfinger, 1996) with maximum likelihood estimation method and number of trials per block as the dependent measure. Four models were tested: no growth, linear, quadratic and exponential growth.  The best fit was the quadratic model.  There was a significant negative linear slope (β= -0.22, t(32)=-2.86, p=.01) and a positive quadratic slope (β= 0.01, t(32)=2.04, p=.05).  Overall, the ASD group had a higher intercept than the TD group (β=0.66, t(280)= 3.16, p<.01;), indicating poorer performance.  There was also a significant negative effect of social reward on the intercept (β=-0.41, t(279)=-3.29, p=.001). There was no interaction between group and reward type or group and linear or quadratic slope; however there was a significant interaction between the linear slope and reward condition (β=-0.11, t(278)=-3.08, p<.01) indicating that on trials with a social reward, both groups demonstrated faster learning.  Final analyses will divide the ASD group into ESDM intervention and community intervention to assess treatment related differences.  We predict the ESDM group will be more sensitive to social rewards than the community intervention group due to the emphasis on social engagement in the ESDM. 

Conclusions : Children with ASD did not perform as well as the TD children; however, they demonstrated the same pattern of learning.  The ASD group showed enhanced social versus non-social reward value. This study demonstrates the feasibility of using gaze contingent eye tracking protocols with young children with and without disability to examine social reward salience as a treatment outcome.

| More