Do high functioning persons with autism present superior spatial abilities?

Abstract

This series of experiments was aimed at assessing spatial abilities in high functioning individuals with autism (HFA), using a human-size labyrinth. In the context of recent findings that the performance of individuals with HFA was superior to typically developing individuals in several non-social cognitive operations, it was expected that the HFA group would outperform a typically developing comparison group matched on full-scale IQ. Results showed that individuals with autism performed all spatial tasks at a level at least equivalent to the typically developing comparison group. No differences between groups were found in route and survey tasks. Superior performance for individuals with HFA was found in tasks involving maps, in the form of superior accuracy in graphic cued recall of a path, and shorter learning times in a map learning task. We propose that a superior ability to detect [Human Perception and Performance 27 (3) (2001) 719], match [Journal of Child Psychology and Psychiatry 34 (1993) 1351] and reproduce [Journal of Child Psychology and Psychiatry 40 (5) (1999) 743] simple visual elements yields superior performance in tasks relying on the detection and graphic reproduction of the visual elements composing a map. Enhanced discrimination, detection, and memory for visually simple patterns in autism may account for the superior performance of persons with autism on visuo-spatial tasks that heavily involve pattern recognition, either in the form of recognizing and memorizing landmarks or in detecting the similarity between map and landscape features. At a neuro-anatomical level, these findings suggest an intact dorso-lateral pathway, and enhanced performance in non social tasks relying on the infero-temporal pathway.

Introduction

Autism is a neurodevelopmental disability characterized by deficits in several domains, while, in other domains, affected individuals exhibit performance that exceeds that of typically developing individuals. This enhanced performance characterizes individuals with autism as a group, and should therefore be distinguished from the outstanding performances exhibited by “savant” individuals with autism, which are found only in a restricted subgroup of individuals with autism (Miller, 1999).

Superior performance has been demonstrated by individuals with autism in pitch processing and memory (Bonnel et al., 2003; Heaton et al., 1998, Mottron et al., 2000; Mottron & Burack, 2001), pattern discrimination (Plaisted, O’Riordan, & Baron-Cohen, 1998), the block design subtest of the WAIS (Shah & Frith, 1993; Tymchuk, Simmons, & Neafsey, 1977), the graphic reproduction of impossible figures (Mottron, Belleville, & Ménard, 1999) and detecting embedded figures (Jolliffe & Baron-Cohen, 1997, Shah & Frith, 1983). An enlarged surface of activation in the occipital primary visual areas and an enhanced activation in the ventral occipito-temporal regions during the embedded figure task was found in individuals with autism in a study using fMRI (Ring et al., 1999). In addition, an atypical activation of the primary visual cortex during face perception (Pierce, Muller, Ambrose, Allen, Courchesne, 2001) and significantly more dorsal electrophysiological response during visual selective attention task (Hoeksma, Kemner, Verbaten, & van Engeland, 2002) have also been observed. These findings suggest that autistic individuals use different neural structures in the early processing of visuo-spatial stimuli.

The current study represents the first systematic assessment of spatial abilities in individuals with autism. More precisely, it aims to establish if individuals with autism possess a superior ability to learn the spatial layout of the environment built from an explored environment or from a map. Superior spatial ability in high functioning individuals with autism (HFA) may be expected on the basis of empirical evidence for preserved or superior visual spatial abilities (Minshew et al., 1997, Ozonoff et al., 1991; Rumsey & Hamburger, 1988; Lord, Rutter, & Le Couteur, 1994) and from superior recognition memory of topographical landmarks such as buildings, landscapes, and outdoor scenes (Blair et al., 2002, Cipolotti et al., 1999). In addition, anecdotal reports of restricted interest in maps and bus routes, as well as the ability to detect minimal positional changes in the environment (Wing, 1976) suggest that visuo-spatial abilities are enhanced among persons with autism. The visuo-spatial tasks in which persons with autism present enhanced abilities (e.g., block design) consist typically of reproducing a 2D or 3D model by manipulation of its components and require a high degree of spatio-constructive ability. However, this superiority should not be manifested in those spatial tasks drawing heavily on executive functions, because operations that require conscious manipulation of information, such as planning or switching from one mental set to another, are impaired in autism (Bennetto, Pennington, & Rogers, 1996).

Spatial ability can be decomposed into multiple interrelated functions that will be presented from the more complex to the more elementary. Only the spatio-cognitive aspect of spatial navigation tasks, cognitive mapping, will be investigated in the current study. Cognitive mapping is the process by which an individual acquires, codes, stores, recalls, and decodes information about the relative locations and attributes of the spatial environment (Downs & Stea, 1973). A cognitive map may be grouped under two subheadings: survey map and route map (e.g., Evans & Pezdek, 1980, Hirtle & Hudson, 1991, Thorndyke & Hayes-Roth, 1982). These two types of maps differ in the sources from which they are primarily acquired, the aspects of the environment that they represent, and the tasks in which they are most useful. A survey map is characterized by the knowledge of the global spatial layout from an external perspective, such as a standard road map. This knowledge reflects the individual’s ability to generalize beyond learned routes and to locate the position of objects within a general and fixed frame of reference (Gale, Golledge, Pellegrino, & Doherty, 1990; Hirtle & Hudson, 1991). One easy way to obtain survey knowledge is to look at a plan that provides an overview of a space otherwise too large to be seen in a glance. Frequent travels on a particular route can also lead to survey knowledge (Siegel & White, 1975, Thorndyke & Goldin, 1983). A route map refers to the knowledge of the spatial layout from the ground level observer’s perspective. It refers to the knowledge of sequential locations, or sequence of actions, required to follow a particular route. It includes an explicit representation of decision points (DPs) along the route where turns occur, as well as a representation of the decisions to be taken at each of these points. This knowledge is acquired by navigating through the environment.

FMRI studies during spatial navigation tasks performed in virtual reality environments are informative about the neural networks involved in survey and route map tasks (Shelton & Gabrieli, 2002, Mellet et al., 2000, Ino et al., 2002, Pine et al., 2002). Route tasks are associated with activation of the bilateral medial temporal lobes, including parahippocampal cortex and posterior hippocampus, as well as the bilateral postcentral gyrus (BA 5, 7), the right superior (BA 7) and inferior (BA 40) parietal cortices. Survey tasks are associated with activation of the bilateral fusiform, inferior temporal Gyri (BA 37, 19), bilateral superior parietal cortex (BA 7) and left insula/claustrum (BA 13). Thus, in both route and survey processing, a common network of brain areas is recruited; the survey mapping is associated with a subset of areas also involved by route encoding, though the former operation results in a greater activation in the inferior temporal cortex, postcentral gyrus and posterior superior parietal cortex. These results do not support the hypothesis that route and survey information rely on different neural systems, but suggest a hierarchical relationship between these two spatial functions.

At a lower integration level, spatio-cognitive processes are under the dependence of visuo-perceptual activity. Neuroimaging studies demonstrate two distinct neural pathways that subserve topographical learning. The ventral, occipito-temporal (or “what”) pathways, and more specifically, the inferior temporal cortex in the region of the fusiform gyrus (Moscovitch, Kapur, Kohler, & Houle, 1995) is involved in the recognition of objects. The temporal activation observed in survey knowledge may reflect greater object processing because of the map-like nature of the survey encoding (Tanaka, Saito, Fukada, & Moriya, 1991). Accordingly, maps can be treated as physical objects per se, in addition to providing spatial information. The dorso-lateral, occipito-parietal (or “where”) pathway, and more specifically, the right inferior parietal lobule in the region of the supramarginal gyrus (Moscovitch et al., 1995), is involved in the processing of spatial relations between objects and of spatial locations (Haxby et al., 1991). In addition, the medial temporal lobe is involved in space integration. Specifically, the parahippocampal gyrus is implicated in the encoding of object features and their locations within space (Aggleton & Mishkin, 1983; Maguire et al., 1996, Maguire et al., 1998, McCarthy et al., 1996).

The current research consisted of five spatial tasks. The first set of tasks was performed in a human-size labyrinth and investigated route map learning (Experiment 1; route learning), route map manipulation (Experiment 2; reversing a route) and survey map (Experiment 3; pointing toward unseen location). The level of information manipulation required by these three tasks ranged from minimal (route learning) to intermediate (reversing a route) and high (reorganizing spatial information to produce a survey map). In addition, each of these tasks was performed at several levels of difficulty in order to assess how the level of difficulty interacts with the spatial cognitive functions involved. The level of difficulty—and therefore, the performance level—of the task is dependent on the storage and manipulation components of working memory (Owen, Downes, Sahakian, Polkey, & Robbins, 1990).

The scale of represented space was manipulated in the second set of tasks in order to assess the ability to transfer spatial knowledge across different scales of space. When spatial layouts are not perceived all at once (e.g., a human-size labyrinth), the subject is confronted with a macro-scale space. Spatial information has to be experienced by integration of perceptual experiences over space and time through the use of memory and reasoning (Montello, 1993; Passini, Rainville, & Habib, 2000; Siegel, 1981). By contrast, spatial cognition at a micro-scale refers to situations where a person can perceive a spatial configuration (e.g., a map) from a single point of view (see McDonald & Pellegrino, 1993 for an overview). It allows direct access to survey knowledge. Experiment 4 involved transforming knowledge acquired through macro-scale learning into a micro-scale representation, through memorizing a human-size path and then drawing it on a sheet of paper. Experiment 5, a route execution task, required the transfer of spatial knowledge acquired from a micro-scale space where global relationships were simultaneously perceived and learned (a map) toward a macro-scale space (human-size labyrinth).