Understanding the social brain in autism

Abstract

Autism is an early onset neurodevelopmental disorder characterized by disruption of early social interaction. Although the social disability of autism remains the central defining feature of the condition, mechanisms that might account for this disability remain poorly understood. This paper briefly reviews some aspects of the social deficit in autism focusing on new approaches to characterizing social information processing problems, potential brain mechanisms, and theoretical models of the disorder. It will touch on aspects of specific social processes that appear to develop in unusual ways in autism including facial perception, joint attention, and social information processing. The importance of adopting more ecologically valid methods and for integrating the various approaches in deriving new models for social deficits in autism will be highlighted. Future research should build on the emerging synergy of different aspects of social neuroscience. © 2011 Wiley Periodicals, Inc. Dev Psychobiol 53:428–434, 2011.

INTRODUCTION

Beginning with my work with William T. Greenough as an undergraduate (Volkmar & Greenough, 1972) I have had an enduring interest in the effects of early experience and the organization of the brain. This paper reviews some of the work from our group on this topic as it related to childhood autism. This highly selective review presents results from our own group but is part of a large, and growing, body of work which is bringing together insights from cognitive psychology, neuroimaging and neurophysiology, and genetics in understanding important aspects of this interesting condition (Volkmar, State, & Klin, 2009).

BACKGROUND

Autism was first described as an inborn disturbance of affective contact (Kanner, 1943). The initial description of autism emphasized the lack of social orientation and social salience for the developing child and the many different areas of disability which followed it. Early work in the area focused, somewhat paradoxically, only on the role of experiential factors. As time went on two different lines of evidence suggested a strong basis for the disorder both in the brain and in genes, e.g., children with autism were at vastly increased risk for seizure disorder (Volkmar & Nelson, 1990) and autism had a very strong familial basis (O'Roak & State, 2008). Since Kanner's original paper, social problems have been repeatedly emphasized as a, if not the, central defining feature of the condition (Volkmar, State, & Klin, 2009). This is true both for categorical and dimensional approaches to diagnosis (Lord & Corsello, 2005). Some gains in social interaction skills, globally defined, do emerge over time; although even for individuals able to achieve personal independence and self sufficiency, marked social oddity typically remains (Shea & Mesibov, 2009). Although social problems change somewhat, over time they remain an area of tremendous vulnerability for persons affected by autism. These social problems appear to be of very early onset. Kanner suggested that the condition was congenital although subsequent work suggests that in a small number of cases infants develop normally or near normally for periods of time before losing skills. Regardless of which pattern is observed, the disruption of basic social processes is profound.

SOCIAL DYSFUNCTION AND THE PHENOTYPE IN AUTISM

Approaches to studying social dysfunction have become increasingly sophisticated in recent years both in terms of methodology and in extending experimental work often initially done with older children and adults, to infants and young children where, presumably, the complicating effects of subsequent development and intervention pose fewer obstacles to clarifying which early processes are disrupted.

Early work, e.g., Ornitz, Guthrie, and Farley (1977), relied on parent report of specific social behaviors rather than on direct observation and characterization. Given the potential for developmental change to complicate studies another line of work focused on analysis of home movies or videos. Osterling and Dawson (1994), used this approach to demonstrate that infants who subsequently had autism exhibited decreased joint attention, fewer typical social behaviors, and more behaviors typical of autism. Differences in attention and a relative lack of social attention but over attention to the inanimate environment were noted in infants with autism (Chawarska, Klin, & Volkmar, 2008). Much interest has centered on the relative lack of attention to the human face and voice—normally the object in the world of greatest relevance to the typically developing child. Before discussing differences in processing faces in autism, a short review of the normal patterns of development is helpful in placing observed differences within a developmental context.

Face-like patterns and stimuli attract the attention of newborns (Valenza, Simion, Macchi Cassia, & Umilta, 1996) who have some ability to discriminate between direct and averted gaze (Farroni, Csibra, Simion, & Johnson, 2002). This very early sensitivity appears to have a subcortical basis primarily in the retinotectal system (Johnson & Morton, 1991) and this process may well facilitate the ability of the infant to detect human faces in the environment early in life. By two months infants begin to preferentially focus on the eye region (Hainline, 1978; Haith, Bergman, & Moore, 1977) suggesting an emerging salience of the face as a meaningful source of information and social communication. By 4–5 months, infants become more sophisticated, e.g., in detecting changes in gaze direction or gaze aversion (Vecera & Johnson, 1995). Smiling and vocalization also become more frequently associated with faces that provide direct eye contact (Caron, Caron, Roberts, & Brooks, 1997; Hains & Muir, 1996). Many other changes occur with increasing ability to respond to eye gaze shifts alone with development of a more conceptual understanding of the significance of gaze shift as involving the signal from another person (Moore & Corkum, 1998)—a process that continues to develop for several years (e.g., Anderson & Doherty, 1997). Within the first year, typically developing infants demonstrate the facial inversion effect (difficulty in processing upside down as opposed to right side up faces). By 8 months of age the typically developing infant often responds to its own name—a finding less apparent in autism (see Volkmar & Chawarska).

Studies of infants and, particularly studies of at risk populations (e.g., siblings) have helped to clarify differences in development of gaze and facial information processes in autism. One of the major areas of difficulty for school age children and adults is the ability to effectively use gaze both as a source of information and in modulating ongoing transaction; problems arise with gaze aversion, difficulties in use of mutual gaze, and in perception of faces. What would, to the typically developing person, be less salient aspects of faces become much more relevant in autism (Hobson, 1986a,b). Individuals with autism have difficulties with facial recognition as well as in use of gaze to engage in joint attentional activities (Chawarska et al., 2008). These difficulties with joint attention serve as early warning signs of autism and likely also set the stage for many other difficulties in learning, e.g., in the absence of a shared focus of interest, the child's attention can be drawn to many other aspects of the environment which seem equally as salient (Klin, Jones, Schultz, Volkmar, & Cohen, 2002a,b). Differences in face processing in autism are also observed using electro-physiological methods, e.g., in a study of event-related brain potentials for upright and inverted faces and objects, individuals with autism exhibited longer N170 latencies to faces than typical individuals but were comparable to controls when viewing objects. Conversely typically developing individuals had longer N170 latencies to inverted faces while those with autism did not differ in viewing upright or inverted faces (McPartland, Dawson, Webb, Panagiotides, & Carver, 2004).

Early attempts to characterize social difficulties more globally have some important advantages but do not, of course, focus on particular processes. Similarly, much of the subsequent work relied heavily on interesting tasks that used static images. In real world settings of course, these processes are much more active and dynamic. An important part of our program of work has been the attempt to develop better methods of characterizing social information processing in real time and in a more “ecologically valid” way.

This approach, which utilizes infrared eye tracking technology, allows us to precisely determine the ways in which individuals with autism view social situations. This method (Klin et al., 2002a,b) initially used a small and unobtrusive infrared video cameras to co-register the viewer's point of regard with ongoing social scenes. In Figure 1, for example, we show two viewers a short clip from the move classic “Who's Afraid of Virginia Woolf?” The top line shows the point of regard of the typical individual (moving back and forth between the eye regions) as contrasted to an IQ matched adult with autism (who focuses on the mouth of the speaker). These results are also seen in a larger group of adolescents and adults (Klin et al., 2002b) with highly significant differences in amounts of time spent looking at eye, mouth, body regions, and objects (see Fig. 2). This result, using a more naturalistic and dynamic approach, is consistent with previous work based on static images that suggested a bias for viewing mouths rather than eyes (Happe, 1994; Hobson, 1986a; Langdell, 1978).

Figure 1. Visual focus of an autistic man (lower trace) and a normal comparison subject (upper trace) showing a film clip of a conversation. Reprinted by permission from Klin et al. (2002b).

Figure 2. Percent of viewing time spent focused on mouth, eyes, body, and object regions (Group data). Reprinted, with permission, from The Social Brain in Autism, F. Volkmar, A. Klin, R. Schultz, K. Chawarska, and W. Jones. In The Social Brain Evolution and Pathology. M. Brune, H Ribbert, and W Schiefenhovel, Eds., Chicester, England: John Wiley and Sons, 2003, p. 186.

Similar patterns arise in other scenes in the movie, e.g., in one clip even when Elizabeth Taylor's character is not speaking, the person with autism focuses on her mouth while the typically developing individual focuses on the eyes. This tendency also suggests new approaches in understanding the interpretation of important social affective cues, many of which are conveyed in the eye region of the face. Similarly, the tendency to focus ONLY on the speaker means that relevant affective responses of others in the scene will be lost to viewers who focus exclusively on the mouth of the person speaking.

In addition to its potential clinical implications, this work raises several interesting questions, e.g., how would very young children with autism or at risk for autism view similar (but developmentally appropriate) social scenes and, secondly, how might we understand the brain basis for these unusual viewing patterns?

Studies of two year olds (including both normally developing and developmentally delayed but not autistic comparison groups) show consistency in these findings with a very striking (and highly significant) preference for young children to focus on mouths (Klin & Jones, 2008; Klin, Lin, Gorrindo, Ramsay, & Jones, 2009). A related body of work shows a more general tendency of such children to orient to nonsocial rather than social components of biological motion (Jones et al., 2008).

BRAIN MECHANISMS

Over the past decade, in particular, advances in methodology and research design have facilitated the study of specific brain mechanisms potentially involved in autism. For example, a substantial body of work now exists on neural correlates of facial perception and it has clarified that, for the typically developing person, a specific brain region, the middle portion of the fusiform gyrus (FG) on the right side is particularly important for face perception (e.g., Allison, McCarthy, Novbre, Puce, & Belger (1994); Kanwisher, McDermott, & Chun, 1997). Kanwisher (2000) has reported, for example that activity in the fusiform face area (FFA) is at least twice as strong to faces as to a wide range of non face stimuli, such as assorted objects, animals without heads, and the backs of human heads.

Our group initially reported (Schultz et al., 2000) decreased activity in the middle portion of the right FFA for persons with autism or Asperger syndrome in comparison to controls during completion of a face identity discrimination task. This finding has been replicated many times (e.g., Critchley et al., 2000) and suggests a major difference in cortical processing of brain information, i.e., faces are perceived as are other objects, perhaps accounting, at least in part, for the relative absence of the facial inversion effect in autism—that is if a face is perceived as an object rather than as a face its orientation would presumably be less salient.

Other brain regions have been implicated in face processing as well. For example, the amygdala appears to have a special role in responding rapidly to emotionally provacative stimuli (LeDoux, 1995); for example the brief (<30 ms) presentation of fearful faces is detected even without apparent fusiform involvement (Whalen et al., 1998).

Other work has focused on the posterior aspects of the superior temporal sulcus (STS) and the amygdala in processing dynamic nonverbal communications, e.g., of emotion or social (Critchley et al., 2000; Pelphrey & Carter, 2008; Puce, Allison, Bentin, Gore, & McCarty, 1998). Some work has suggested that the STS and amygdala are under-active in autism and in related disorders in tasks of various degrees of explicitness (Critchley et al., 2000). Similarly the ability to detect changes in direction of eye gaze direction also appears to be an activity depending on the STS (Allison, Puce, & McCarthy, 2000), for example, for perception of static and dynamic facial movement as well as goal-oriented body movements (Grafton, Arbib, Fadiga, & Rizzolatti, 1996; Hoffman & Haxby, 2000; Puce et al., 1998). The STS appears to work in interaction with the fusiform gyrus (e.g., Kingstone, Friesen, & Gazzaniga, 2000; Puce et al., 1998; Vecera & Johnson, 1995) and the amygdala (Kawashima et al., 1999).

THEORETICAL PERSPECTIVES

Several different overarching theoretical perspectives have been proposed in an attempt to provider a broader context for understanding social disturbances in autism. These approaches attempt to move the discussion from one specific process or feature in order to place observed difficulties in some broader theoretical context. Early work in this area tended to focus on what were presumed to be “core” areas of difficulties, e.g., perceptual, neuropsychological or behavioral problems. Except, perhaps, in the area of language and communication problems this work tended, somewhat surprisingly, to avoid a specific focus on social aspects of autism. More recent approaches have tended to employ broader theoretical models with a goal of understanding the relevance of underlying processes in social dysfunction.

One approach has centered around problems in the area of executive functioning (EF). These functions include an ability to engage in forward planning, and to engage an appropriate problem-solving orientation in moving towards a specific objective. Within this model difficulties in flexibility and “set shifting” are highlighted as areas most relevant in autism (Ozonoff, South, & Provencal, 2005). The difficulties that individuals with autism have in coping with “real life” skills and generalization are a nice “fit” with this model and problems in specific brain regions, e.g., dorsolateral prefrontal cortex, have been suggested to underlie the social dysfunction in autism as well (Pennington & Ozonoff, 1996). From a practical point of view there are significant challenges for this as an overarching model for autism. EF difficulties are not unique to autism (Pennington & Ozonoff, 1996) furthermore the relationship of severity of social problems to levels of EF function is not straightforward (Dawson, Meltzoff, Osterling, & Rinaldi, 1998). On the other hand the potential heritability of problems in this area (e.g., Hughes, Leboyer, & Bouvard, 1997) adds to interest in this model.

A rather different theoretical approach has focused on difficulties in capacities for information synthesis, i.e., of integration of information into coherent and meaningful wholes. This approach, referred to as the weak central coherence hypothesis (Frith, 1989), is thought to account for problems in attention and in appreciation of overall context and meaning. While interesting, this hypothesis has received relatively less experimental attention (Mottron, Peretz, & Menard, 2000) and one might just as easily suggest that the difficulties in “central coherence” arise from the underlying social problems, for example, given difficulties in both processing social information and in the “pull” of motivation for social interaction, infants with autism would not be drawn to derive meaning from stimuli such the human face and early processes of social engagement and joint attention.

Undoubtedly the most productive hypothesis in the area, at least in terms of papers, has been the theory of mind (ToM) hypothesis (Baron-Cohen, 1995). In this view, social interaction problems arise as a result of a basic inability to think about mental phenomena in self and others, i.e., in intersubjectivity. Thus individuals with autism are thought be unable to conceptualize other people's beliefs, intentions, desires, and feelings, and therefore cannot use such knowledge in understanding the behavior of other people. This hypothesis accounts for many aspects of social difficulty in autism, such as with social language use. On the other hand there are some fundamental difficulties with this view. Most importantly some of the problems reported in autism, e.g., lack of interest in gaze at parents/care givers, are problems that developmentally arise in the very first days of life—a time well before intersubjective skills of the usual ToM variety are thought to develop (Klin, Volkmar, & Sparrow, 1992). In addition there is a strong relation of ToM skills to overall language and cognitive levels, e.g., more cognitively able persons with autism can solve usual ToM tasks quite readily—but remain severely socially disabled (Dahlgren & Trillingsgaard, 1996).

Other models have been proposed as well. For example, the “enactive mind” hypothesis focuses on early emerging aspects of socialization that precede the kinds of mentalizing abilities in ToM and attempts to encompass other early emerging aspects of autism such as deficits in joint attention and other skills (Klin & Jones, 2008). At the moment each of these perspectives has contributed, to variable degrees, to the research enterprise. A challenge for the next years of work will be to move towards more specifically delineated theories and highly specific hypotheses.

SUMMARY AND DIRECTIONS FOR THE FUTURE

Autism is an early onset disorder marked by a profound social disability which impacts the young child's ability to understand people, their emotions, and to establish social reciprocity. The mechanisms that underlay this social disability remain poorly understood although, as noted above, significant progress has been made in several areas. One major advance has been the attempt to develop better and more precise methods and metrics that rely on more naturalistic and ecological valid social situations. Functional MRI work has focused on differences in brain processing of social stimuli, for example, differences in recognition of faces as compared to objects, as well as to neural correlates of processes such as perception of biological motion and joint attention.

In the coming decade it will be important that this effort build on the emerging synergy of social interaction in neuroscience. As potential genes for the disorder are recognized, likely in the near future, analyses at every level—from genetic substrate to neural mechanism to behavioral phenotype, will need to be conducted in coordination with each other.