**6. Summary and conclusions**

ASD is diagnosed in about 1 in every 88 children (CDC, 2012), many of whom will have poor outcomes as adults, requiring some level of assistance throughout their lives (Seltzer, Shattuck, & Greenberg, 2004). In addition to this high prevalence of the disorder, there is a high heterogeneity, high co-morbidity, and the possibility of several subgroups of ASD. Together, these factors make it imperative to better understand the disorder and develop effective interventions. However, as more research accumulates, so do inconsistent findings and unexpected differences in patterns of performance on tasks that were designed to measure the same cognitive process or factor.

114 Current Topics in Children's Learning and Cognition

the marble into the box.

conceptual similarity.

**6. Summary and conclusions** 

locations, a basket and a box and two dolls, Sally and Anne. The story involves Sally placing a marble in a basket, which is then moved into the box by Anne – without Sally being present. The critical task is to determine the location where Sally would search for her marble upon her return. If children understand Sally's mental state, they should pick the basket, because that is the marble's location known to Sally. If, on the other hand, children go by their own beliefs, they should pick the box, because they know that Anne has moved

Many studies have sought to identify general deficits in theory of mind reasoning in individuals with ASD, as a means of better understanding their social deficits (for reviews see Rajendran & Mitchell, 2007). For example, Baron-Cohen et al. (1985) found that 80% of children with ASD failed the Sally-and-Anne test, compared to only 20% of matched controls. This difference cannot be attributed to general difficulties understanding the task instructions, given that children with ASD were able to answer control questions about the various locations of the marbles. Taken alone, these data appear to highlight a pathology

However, highly variable performance between similar theory of mind tasks has been observed (Grant, Grayson and Boucher, 2001; Baron-Cohen, Leslie, & Frith 1985; Yirmiya, Solomonica-Levi, Shulman & Pilowsky, 1996; Zelazo, Burack, Benedetto, & Frye 1996). Consider, for example, findings with the so-called deceptive-box task, a task in which the content of a box does not match with the label on the box (Grant, Grayson & Boucher, 2001). After being shown the content of the box, participants are asked about what another participant would predict about the contents of the box (without having seen inside the box). To answer accurately, the participant must understand that their knowledge of the contents of the box is not accessible to the other participant. Comparison of performance on the deceptive-box and Sally-and-Anne tasks revealed important differences in ASD: Performance was better on the deceptive-box task than the Sally-and-Anne task, despite the

In defense of a reductionist framework for ASD, one could go about dissecting the tasks in order to find the stable factor that could explain ASD. For example, one could argue that the two tasks differ in whether they involve real people (the deceptive box task) or puppets (the Salley-and-Anne task). Variable performance between these tasks may therefore reflect a difference in the perception of behavior associated with living versus inanimate actors. However, this explanation falls apart when the larger body of theory of mind research is considered: The inconsistent findings in theory of mind tasks (for a review, see Rajendran &

ASD is diagnosed in about 1 in every 88 children (CDC, 2012), many of whom will have poor outcomes as adults, requiring some level of assistance throughout their lives (Seltzer,

Mitchell, 2007) to do separate on the fault line of inanimate vs. animate stimuli.

specific impairment to theory of mind reasoning in individuals with ASD.

In the current chapter, we have reviewed some context effects taken from the domains of perception, learning, executive functioning, and social reasoning. On the one hand, while there are robust differences in performance in all of these domains, these differences can disappear under certain task contexts – rendering them less robust than initially thought. Specifically, while ASD is characterized by a focus on local details (vs. on an overall Gestalt), by difficulty with implicit learning, executive functioning, and social reasoning, these differences disappear as the variability in tasks increases. It is plausible that more context effects accumulate as ASD research expands in cognitive development, further exasperated by a focus on individual children.

Context effects are nothing new in the literature of cognitive development (e.g., Kloos & Van Orden, 2009). A plausible reaction is to dismiss them as isolated instances, leaving an existing hypothesis intact, or refining it to incorporate the context effects. The problem, however, is that both these solutions can only address context effects locally. Yet, context effects are not a local phenomenon. And they are not likely to disappear with more participants, more precise methods, or simply more data. In fact, if research with typically developing participants is any guide for predicting the patterns of findings, context effects are a necessary feature of the enterprise. And with more research we are likely to find more context effects (e.g., see Shanks, Rowland & Ranger, 2005, for a discussion on implicit learning in neuro-typical adults). An expanding of a reductionist theory of ASD to incorporate them all is unlikely to retain its usefulness. Instead, context effects undermine a reductionist theory altogether.

Rather than focusing on a binary interpretation of patterns of performance, context effects hint at the possibility of a complex interplay between factors that make a black-and-white approach to understanding performance insufficient. By shifting attention away from searching for a "smoking gun" of ASD, it may be possible to better understand the emergence of how the components are coordinated. It is possible that the coordination among components is compromised in individuals with ASD, reverberating through all areas of functioning, and amplifying itself with development. It gives rise to atypical perception patterns, implicit learning, planning, and social interactions – at least when the immediate task context does not support an adaptive coordination of interdependent components. This approach, while failing to reduce ASD to a single deficit, has important implications for training and teaching strategies. In particular, this approach makes it possible to map out how changes in support of coordination exist in the environment to bring about improved task performance.
