Learning Words by Watching: A Comparison of Eye-Tracking and In-Person Measures

Background:

Reduced responsiveness to joint attention (RJA) is often observed among autistic children below a certain developmental level (e.g. Mundy, Sigman, & Kasari, 1994) and is predictive of language development (Sigman & Ruskin, 1999). Difficulty using a speaker’s attention to learn words has been documented primarily among cognitively delayed children with autism (Baron-Cohen, Baldwin, & Crowson, 1997; Preissler & Carey, 2005). Some autistic children are unimpaired at using another’s attention to learn words (Luyster & Lord, 2009). Even autistic children who have attained a fairly advanced cognitive level (one that would typically be associated with few RJA impairments) exhibited reduced RJA when assessed with an eye-tracker during an RJA word learning opportunity (Akechi et al., 2011). Does eye-tracking enhance detection of RJA difficulties that are not apparent with in-person measures?

Objectives:

1. Determine if autistic children exhibit different levels of RJA in an eye-tracker versus in-person.
2. Assess relations between developmental level, RJA and word learning in autism.

Methods:

Fifteen 3 to 7 year old autistic children participated in this study.  Control participants are currently being recruited.  Participants viewed three measures of RJA: an in-person RJA word learning paradigm, an eye-tracking RJA word learning measure, and the RJA component of the ESCS. Two novel words and objects were introduced during each RJA word learning measure in a counterbalanced manner.  Both word learning paradigms consisted of a female model turning toward one of 2 objects 4 times each and labeling them.  Intelligence was assessed with the Mullen Scales of Early Learning or the Differential Ability Scales. Children with an IQ less than 70 were classified as intellectually disabled (ID).

Results:

Similar levels of RJA (a difference score consisting of first looks toward the same object as the model minus first looks toward the other object) and word learning (assessed by object selection) were observed in the eye-tracker and in person (p>.05). Intellectual disability was unrelated to ESCS RJA and to the in-person RJA word learning paradigm (p>.05). However, ID children (M=1.20, SD = 2.77) displayed less RJA in the eye-tracker than non-ID children (M= 4.86, SD= 2.61, p =.042). Both ID and non-ID children exhibited similar levels of word learning in response to the eye-tracking paradigm (p >.05). However, ID children (M=1.00, SD = 2.00) showed less word learning following the in-person RJA word learning paradigm than non-ID children (M=5.22, SD = 1.39, p <.001). Interestingly, word learning following the in-person paradigm was associated with whether or not a child was intellectually disabled (p= .044) but was not associated with RJA during the word learning task (p=.271). 

Conclusions:

Autistic children exhibited similar levels of RJA in-person and in the eye-tracker. However, eye-tracking revealed associations between RJA and developmental level that were not apparent using in-person measures. Developmental level was associated with the ability to learn words in response to social cues while RJA during the task was not. Thus, frequency of RJA may be less important for word learning than the ability to recognize the referential nature of social cues.