Learning about Objects from Referential Gaze Versus Arrows in Children with and without ASD

Background:  Children with autism spectrum disorders (ASD) often struggle with understanding another’s subtle communication cues, such as referential gaze (looking at an object). This is important in different contexts: when a person labels a new object and looks at it (word learning, WL), or when a person looks at an object they will act on next (action understanding, AU). Successful learning from gaze is often attributed to understanding another’s communicative intentions, and poorer learning in ASD has been attributed to a lack of understanding intent. Yet gaze may simply direct attention -- successful learning may not require understanding intent. Prior work has not explored how children with ASD learn from referential gaze across contexts, or versus an attentional control.

Objectives:  We investigate how 6- to 10-year-old children with ASD and typically-developing (TD) children learn in both WL and AU contexts. Specifically, we examine how children attend to and learn from referential gaze versus a moving arrow (attentional control). 

Methods:  Participants were children with ASD (n = 18) or TD (n = 18) matched on age, gender, and nonverbal IQ. Children watched videos that taught the name of a new word (WL) and how to build a tower (AU). In both contexts a cue (referential gaze vs. an arrow) directed attention to one of two objects. An eye-tracker recorded children’s on-line attention during baseline, teaching, and test phases. Children were also asked to point to the learned object during test.

Results:  2 (group) x 2 (cue condition) x 2 (context) mixed ANVOAs were conducted (alpha = .05). During teaching, there was no main effect or interaction with group. All children looked significantly longer at gaze (M = .212, SD = .178) versus the arrow (M = .127, SD = .122). Children also looked significantly longer at the cues in AU (M = .200, SD = .181) versus WL (M = .142, SD= .128). No main effects or interactions were found for looking at the target during teaching. 

During test, results differed by context and group. In WL many children successfully pointed to the target regardless of cue (10 ASD, 14 TD); for successful pointers, there were no main effects or interactions for latency of first look to or proportion of time looking at the target. Remaining children included: inconsistent pointers with either cue (5 ASD, 3 TD), selective pointers with arrow (1 ASD, 1 TD), and unsuccessful pointers (2 ASD). In contrast, few children successfully pointed in AU (1 ASD, 5 TD), and many were consistently unsuccessful (12 ASD, 5 TD). Remaining children included: inconsistent pointers (1 ASD, 3 TD), selective pointers with gaze (2 ASD, 4 TD), and selective pointers with arrow (2 ASD, 1 TD).

Conclusions:  When learning about an object, children with and without ASD treat referential gaze differently from an arrow, both in word learning and action understanding. Despite similar attention to the targets during teaching, TD and ASD groups differed at test. Future analyses will address relationships between teaching and test and the quality of learning over time.