Attention Capture by and Preference for Faces with Direct Gaze in Toddlers with ASD, DD, and TD

Background:  Faces are prioritized in the attentional system and capture attention at the early stages of visual information processing, e.g., when typical adults are simultaneously presented with a face and an object, their saccadic responses are initiated more rapidly towards the face (Crouzet et al., 2010). Consistent with these behavioral findings are electrophysiological studies suggesting faster brain responses to faces than non-faces in typical infants (De Haan & Nelson, 1999) and adults (Bentin et al., 1996).  Furthermore, faces with direct gaze attract and hold attention to a greater degree than faces with averted gaze in infants (Farroni et al., 2002) and adults (Senju et al., 2006) .  To date, neither of these attentional biases has been studied in toddlers with autism.

Objectives:  We examined: 1) attention capture to faces and objects; and 2) preferential orienting to faces with direct gaze in 18-24 month old toddlers with autism spectrum disorder (ASD), developmental delay (DD), and typical development (TD).

Methods:  Eighty toddlers were tested in a visual preference paradigm: ASD (N=32; M=1.85yrs), DD (N=18; M=1.62yrs), and TD (N=30; M=1.63yrs).  Two conditions were presented: 1) Gaze where a face with direct gaze was paired with the same face with averted gaze (4 trials) and 2) Object, where nondescript objects were presented in an upright and inverted position. Each trial lasted 5s and was preceded by a 1s central fixation point. Saccadic reaction time and direction of the first saccade were coded by three coders with 85% inter-rater reliability.

Results:  Mixed models diagnosis x condition ANOVA indicated that toddlers in each group initiated their first saccade toward one of the two stimuli (target or distracter) faster in Gaze (M=270ms, SD=79) than in Object (M=323ms, SD=104) conditions (p=.001). There were no effects of diagnosis (p=.73) or diagnosis x condition interaction (p=.80).  However, in the Gaze condition, only TD and DD controls directed their first saccade toward the target (direct gaze) (p=.01 and p=.01, respectively).  In the Object condition, the control groups performed at chance level.  Toddlers with ASD performed at chance level in both conditions.

Conclusions:  This study is the first to demonstrate that the presence of faces (relative to objects) accelerates the generation of saccades in toddlers with and without social disability. It is unclear whether saccadic responses were facilitated by the same process in all three groups, e.g., inherent salience versus ubiquitous familiarity of facial stimuli. This finding is consistent with our previous work (Chawarska, Klin, & Volkmar, 2003) suggesting that the very early and elementary stages of face processing in toddlers with ASD might be preserved.  Despite this advantage, toddlers with ASD were less likely to direct their first saccade to the face with greater biological significance (i.e., direct gaze).  This is consistent with other studies suggesting impairments in higher-order face processing in toddlers with ASD including deficits in scanning and recognition (Chawarska & Shic, 2009; Bradshaw et al., 2011).