Esprits atypiques: les sciences cognitives de la différence et des potentialités

Cognitio 2015

Colloque jeunes chercheuses et chercheurs en sciences cognitives

Montréal, 8, 9, et 10 juin 2015

Eye-tracking Events in Autism: Testing the Enhanced Local Processing Hypothesis with Sentences and Dynamic Scenes

Deborah Martin & Roberto G. de Almeida.

Numerous studies have suggested that individuals with Autism (ASD) perceive information in more detail—by hypothesis, a superior local processing bias (Mottron, Dawson, Soulieres, Hubert, & Burack, 2006; Happe & Frith, 2006). These studies have given rise to diverse theories emphasizing a local vs. global processing of visual stimuli (e.g., Dakin & Frith, 2009; Mottron, 2001), with the suggestion that global processing is not necessarily affected in ASD but might be the less preferred processing strategy or a “disinclination” (Koldewyn, Jiang, Weigelt, & Kanwisher, 2013). Thus far few studies investigating visual processing in ASD have employed complex naturalistic dynamic scenes (e.g., Nakano et al., 2010; Norbury et al., 2009; Noris et al., 2012). And, to our knowledge, no studies have combined these types of scenes with linguistic information—a manipulation that might enhance the processing of scene properties such as their constituent objects. We investigated the general “enhanced local processing” hypothesis employing the so-called visual world technique in which spoken and visual stimuli are presented concurrently, while eye movements to objects in scenes are tracked as a function of particular points in the linguistic stream. In the present case, we were interested in understanding whether different types of verbs embedded in sentences and different types of motion contexts in dynamic scenes contributed to enhance attention to particular objects in those scenes. ASD (n = 13; age rage: 9-12) and TD children (n = 9; age range: 9-12) watched 15 video clips of events (e.g. a woman cooking in a kitchen) while listening to related sentences which contained either a causative verb (denoting a change of state in the object of the sentence usually by an agent; e.g., …the women will crack the eggs…) or perception/psychological verb (denotes a perceptual or psychological state; e.g., …the woman will examine the eggs…). In terms of visual context, agents in the scene (e.g., the woman) would either reach towards the target object (the referent of internal argument of the main verb; e.g. the eggs) or would remain neutral. We hypothesized that, due to their well-documented local processing bias, ASD children, in contrast to TD children, would show greater anticipatory eye movements to the target object of causatives as opposed to perception verbs. Anticipatory eye movements—when saccades to an object occur even before the offset of the object’s name—are taken to measure sensitivity to both linguistic (verb-semantic) and visual (object and scene) information as both sentence and scene unfold concomitantly. Although several studies have obtained this anticipatory behavior as a function of verb-semantic restrictions in healthy adults, employing static scenes (e.g., Altmann & Kamide, 1999 [drawings]; Staub, Abbot, & Bogartz, 2012 [pictures]), others have failed to obtain anticipatory behavior when the scenes are dynamic, i.e., involve motion; de Almeida, DiNardo, & von Grunau, 2015). All these studies, however, have shown eye-movement sensitivity to verb semantic restrictions such that the more restrictive verb yields faster saccades to target objects. In the present study, preliminary analyses indicate an interaction between group and verb type such that ASD children looked faster to the target object in the causative condition (as healthy adults do) while TD children looked at the target object faster in the perception condition. These analyses, however, should be seen with caution due to the large number of missing values (trials in which the target was not fixated post-verb). When we analyzed the proportion of trials in which the target was fixated—as a measure of attention taking into account both visual and linguistic information—the pattern of data was different: TD children looked at the target significantly more than did ASD children; this difference appears to be greater in particular in the neutral agent-motion contexts, thus when only linguistic clues are available for enhancing the detection of the object. ASD children, however, perform similarly to TD children when the agent moves towards the object, suggesting that these perhaps “social” cues are taken into account when observing unfolding events. Although our data and analyses are preliminary to draw any strong conclusions, the pattern of results does not appear to be compatible with theories postulating a local-processing superiority in ASD.