Chapter III: Face Preference Decision-Making And Visual Behavior
3.4 Familiarity versus Novelty Study
3.4.4 Discussion
In the present study, we explored category-specific familiarity and novelty principles in high- functioning autism. We found there were similar patterns of preference segregation between groups, such that people with ASD developed a familiarity preference for faces and a novelty preference for nature scenes, similar to controls’ preferences. In addition, we found that patterns of initial gaze did not differ significantly between groups, such that both groups made similar proportions of their initial fixations to the familiar stimulus, and to the image that was eventually chosen.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Likelihood of Initial Fixation on Chosen Image
Controls Autism
Female Faces
Male Faces
Desert Scenes
Mountain Scenes
Repeated visual exposure to stimuli often increases preference for those stimuli. This familiarity preference has been shown to be specific to certain object categories, such as faces, whereas a novelty preference has been observed for non-social images. Here we found a strong familiarity preference, with no difference between groups, for the face that was presented in all 20 trials. There was also a strong preference in the opposite direction with nature scenes for the novel image also presented in each of the trials. The overall results replicate the segregation of preference bias reported by Park et al.’s (2010) original study.
The familiarity preference for faces and nature scenes was apparent quickly (within the first 5 trials), again with no significant difference between the autism and control group. It is interesting to note, however, that while the male and female faces differed significantly from the mountain scenes, none of the pairwise comparisons involving the desert scenes were significant, indicating a lack of initial bias for the desert images. This may be because desert images generally show less variability in color range and contrast than mountain images, making them less visually appealing, or perhaps because the images that were chosen were not particularly compelling enough to inspire a preference bias in the initial stages.
On average, the preference biases for familiar faces and novel nature scenes were present through the remainder of the trials, with participants showing the strongest familiarity preference for male faces, followed by female faces, and the strongest novelty preference for mountain scenes, followed by a weaker novelty preference for desert scenes. Since these results did not differ between groups, they suggest that the ability to form category-specific preferences, as well as maintain those preferences, is not significantly compromised in high- functioning autism.
Evidence in support of intact preference formation is also seen in the lack of group differences in likelihood of initial gaze to the familiar image. Participants were more likely to make the initial fixation in a trial on the novel nature scene image than the novel face image, suggesting first that the use of top-down attentional strategies can likely be ruled out since these differences were apparent at the first fixation, presumably before participants had the opportunity to foveate on the images and determine consciously which image was the familiar
one. Second, the similarity between groups in likelihood of initial fixation to the familiar image also suggests there is a shared underlying mechanism functioning pre-attentively to direct visual attention, rather than a post-hoc attentional strategy. While the findings reported here are mostly negative (i.e., n.s.) relative to the controls, there are significant implications owing to the baseline positive results in the controls. Once again, it indicates that despite social impairments in ASD, certain aspects of face processing remain intact or can be compensated for by people with ASD to result in similar preference decisions as those made by controls.
While our findings point to normal preference formation in autism, there are several open questions for future studies. Given the significant amount of noise in each trial for both groups, averaging across a larger number of subcategories could reduce the variability and perhaps reveal differences in the time-course of preference development that were obscured in the present study. This could be achieved by running either more repetitions of the existing subcategories or using a greater number of sub-categories.
Future studies could also investigate individual familiarity and novelty preferences within subjects. Given the putative preoccupation with non-social stimuli in autism, familiarity preferences may occur in different object categories for different subjects. A study in which pre-ratings are obtained from each participant for different stimuli categories, and then used to construct image pairs could be informative to understanding if there are subject-specific areas of “expertise” that would elicit a familiarity bias similar to faces. Future studies could also use multiple social stimuli to provide greater ecological validity, since many social situations involve more than one or two others. It could be the case that deficits are only present, or become apparent, when there are more visual and social stimuli competing for attention.
Lastly, it is unclear even in neurotypicals, whether the face familiarity preference is restricted to realistic faces and whether it would extend to face-similar categories, such as schematic faces, paintings of faces, or “man on the moon” type visual effects. Would people with autism see faces in non-face objects (face-similar objects)? In other words, what level of abstraction would be necessary to eliminate the preference bias for faces? Moreover, where is the boundary between face and non-face in autism and in neurotypicals? It would be enlightening for our
understanding of face processing, both in neurotypicals and in individuals with autism, to explore to what degree of abstraction the brain will still respond to by forming a familiarity preference, as it would suggest what some of the necessary elements are for face perception and categorization, as well as for our emotional response to faces.