Can Individuals with Autism Abstract Prototypes of Faces?

Background: Perceptual explanations for the face processing difficulties seen in individuals with autism are often bottom-up accounts, suggesting that these individuals are biased toward processing local features as opposed to global patterns (Frith & Happé, 1994; Mottron et al., 2006).  However, theories on the development of expertise in face processing suggest it is also critical to consider the impact that top-down processes, such as prototype formation, may have on the ability to process and recognize facial information (e.g., Newell, et al., 2010). 

Objectives: To examine prototype formation ability using subtly varying natural faces and explore individual differences in performance as related to measures of intelligence and behavioral symptoms of autism. Also, to examine eye movement patterns to determine whether differences in attentional distribution can explain difficulties with prototype formation.

Methods: High-functioning adult males with autism (n = 20) and age and IQ matched controls (n = 20) were tested in a prototype formation task utilizing natural faces. Participants were familiarized with sets of faces that were created by manipulating specific features and spatial distances between features.  After each set, participants chose which face was more familiar, the prototype face or a face comprised of features that were previously seen (mode face).  Participants’ eye movements were also recorded to gather information about which areas of the faces participants looked at when viewing the familiarization trials (learning).

Results: The control group (M = 72.5%) selected the prototype faces as familiar more often than the autism group (M = 50.83%) indicating that the individuals with autism had significant difficulty forming prototypes of natural faces (t (38) = -2.96, p < .01).  However, there was a subset of individuals with autism (n = 8) who performed well and appeared to form a prototype.  These individuals had lower Stereotyped Behavior and Restricted Interests Total Scores on the ADOS (M = 1.13) than those who performed poorly (M = 3.00) (t (18) = 3.04, p < .01).  Eye-tracking data indicated that the groups did not differ in the amount of time that they spent looking at the faces or particular facial features.  The general pattern of attention to the facial features was also similar for both groups.

Conclusions: Results are consistent with past research indicating a deficit in prototype formation in individuals with autism.  Prototype formation difficulties cannot solely be accounted for by differential attention to features or different attentional patterns to faces but may be related to differences in the way in which individuals with autism cognitively process information.  This study highlights the need to consider the role that top-down cognitive processes such as prototype formation may play in the syndrome of autism.