Affective Salience Influences Looking Preference for Social Elements during Interactive Play Scenes in Young Children with Autism
In typically developing children, social elements within natural scenes substantially bias allocation of attention (Chevallier et al., 2012). Furthermore, affective salience and positive affect act as additional cues to drive attention and engagement within complex scenes (Pool 2016). Reduced preference for social stimuli has been reported in autism and related to social learning impairment (Dawson 2012; Chita-Tegmark 2016).
Eye-tracking studies using dynamic social stimuli, indicated that children with autism specifically fail to attend to social cues in more naturalistic conditions (Chawarska et al., 2012; Chevallier et al., 2015). However, none of these studies have systematically manipulated the level of emotional salience of social cues, and tested its impact on the looking patterns for natural interaction scenes.
To test, using an eye-tracking experiment, whether a) children with and without autism attend equivalently to videos of naturalistic interactive play; b) affective salience and positive affect are able to modulate attention to social-related elements of interactive play scenes equivalently in children with and without autism.
Fifty-eight male children (19 children with autism (ASD) and 29 typically developing children (TD)), aged 35-75 months were enrolled in the study.
The experiment consisted of 6 trials (3 different trials, each repeated in a neutral and an affectively salient condition) of a child and an adult interacting with objects (fish game, blocks and toy train) in a naturalistic play interaction, presented on a computer screen. Each trial lasted 25 seconds and was presented in a random order. Gaze patterns were recorded with a SMI iView XTM RED dark-pupil 120Hz eye-tracking system (Sensomotoric Instruments, 2005) and exported using SMI BeGaze 2.4 software. Statistical analyses were conducted using R (http://www.r-project.org/). A linear mixed effects model (package: lme4) was applied to explore the effect of group and condition (neutral vs salient) to predict looking time (dwell time) in the regions of interest (adult and child face, activity area, background).
A main effect of group (Wald χ2=10.02, p=.002 ; Wald χ2=4.5, p=.03; Wald χ2=3.5, p =.06) and condition (Wald χ2=6.7, p=.01; Wald χ2=4.6, p=.03; Wald χ2=6.04 , p=.01) was observed on the looking time to the adult face in all the trials, with the ASD children looking significantly less than the TD children at the adult face and the salient condition significantly enhancing the looking time to the adult face in both ASD and TD children. No group by condition interaction was found (Wald χ2=0.11, p=.74; Wald χ2=0.4, p=.54; Wald χ2=0.5, p=.5).
We found a significant group difference in the amount of time spent attending to the social elements of the scene, with the ASD children attending less than the TD children to the adult face. Furthermore a significant effect of affective salience was noted, with an increase of time spent attending to the faces in both the ASD and TD children during the salient condition.
This study provides preliminary evidence of a substantial influence of affective salience in visual orientation and exploration of social-related elements in naturalistic play interaction scenes in young children with autism.