The Theory of Mind (TofM) hypothesis for autism has been investigated extensively in the last 30 years. This essay will review the literature which examines this theory, considering a wide range of methodological and theoretical approaches. It will be argued that whilst there are several strengths of the theory, the TofM hypothesis cannot account for the non-social symptoms of ASD, and therefore cannot be considered to provide a comprehensive account of the disorder. Cognitive evidence (including a range of TofM tests), theoretical issues, and alternative hypotheses of autism will be considered.

What is the Theory of Mind Hypothesis for Autism?

The TofM hypothesis of autism was first proposed by Baron-Cohen, Leslie and Frith in 1985. It proposed that individuals with autism lack a TofM, leading them to have specific deficits, such as an inability to partake in pretend play and impairment in social skills. The theory has been controversial since its proposal, with several alternative hypotheses being suggested.

Cognitive Evidence

The earliest tests of TofM ability often relied heavily on tests of false belief, in which participants are required to understand that other people do not necessarily know (or believe) the same as them. For example, Baron-Cohen, Leslie and Frith (1985) investigated TofM ability in children with autism and children with Down’s syndrome, and found those with autism were significantly impaired in their understanding of false belief, suggesting a deficit in TofM ability in autism, and thereby supporting the TofM hypothesis for autism. Similar tests involving false belief, such as those carried out by Happe (1995), have supported findings by Baron-Cohen, Leslie and Frith (1985), providing strong support for the TofM hypothesis of autism.

The principle problem in using false belief tests is their reliance on language use. (Bloom and German, 2000). Colle, Baron-Cohen and Hill (2007) investigated concerns that a potential inability to understand language associated with false belief tasks may result in individuals with autism failing these tasks, despite have the TofM ability required to pass tests (e.g. Bloom and German, 2000). They used a non-verbal false belief test to compare the TofM understanding of children with ASD to children with SLI, and found that autistic children have specific deficits in TofM which were not present in the SLI control group. They concluded that previous findings of a TofM deficit by research using standard tests of false belief (e.g. Surian and Leslie, 1999) were unlikely to have been confounded by language ability, thereby revalidating claims made by this research.

However, there are other issues associated with the reliability of false belief tasks. For example, Bloom and German (2000) have suggested that a single test of TofM is unlikely to comprehensively test all elements of TofM ability in an individual. Furthermore, whilst research in the field shows the majority of autistic participants fail tests of false belief, a small minority of high-functioning autistic individuals pass the test (e.g. Baron-Cohen, Leslie and Frith, 1985). This has raised questions regarding the ability of false belief tests to recognise subtle deficits in TofM ability potentially shown in the high-functioning autistic population (Rutherford, Baron-Cohen and Wheelwright, 2002). As a result, new research in the field has often focused on using alternative tests of TofM.

The ‘Reading the Mind in the Eyes’ test (Baron-Cohen et al., 2001) is an alternative to false belief tasks, in which participants are required to recognise emotions based on images of a person’s eyes. To pass this test, participants must rely on subtle cues indicated in the eyes, thereby indicating ability in mental state understanding, or TofM. Baron-Cohen et al. (2001) used this test to show that individuals with ASD are significantly impaired in their ability to recognise emotions using subtle cues, thereby indicating a deficit in TofM ability. A similar test was used by Rutherford, Baron-Cohen and Wheelwright (2002) and Golan et al. (2007) in which participants were required to establish how a person was feeling after listening to a phrase spoken by that person. This test also indicated social impairments caused by a deficit in TofM ability in ASD, as participants with ASD were significantly impaired in their ability to recognise emotions using vocal cues.

Research into TofM ability using the ‘Reading the Mind in the Eyes’ test is particularly beneficial in supporting the TofM hypothesis. Methodologically, the test does not rely on language comprehension, and therefore passing or failing the test is based purely on TofM ability. Both the ‘Reading the Mind in the Eyes’ test and the ‘Voice’ variation are able to detect subtle differences in participants’ social sensitivity, allowing for deficits to be identified in populations that often pass false belief tasks (such as individuals with high-functioning Autism, Baron-Cohen, Leslie and Frith, 1985), and thereby suggesting the TofM hypothesis is generalisable and universal across the highly varied autistic population.

The Faux Pas test was also developed by Baron-Cohen et al. (1999), and assesses recognition of ‘faux pas’ (defined as being when something is said without consideration of whether the information may be something that the listener does not wish to hear or know, Baron-Cohen et al., 1999). In 2009, Zalla et al. investigated faux pas recognition ability in individuals with Asperger’s syndrome using the Faux Pas test, and found these individuals made significantly more errors in detecting faux pas than typically developing individuals, were unable to correctly justify the speaker’s behaviour and were unable to comprehend the impact of this behaviour. Zalla et al. (2009) concluded that this showed a specific deficit in TofM ability in individuals with ASD.

Methodologically, research using tests of faux pas lend strong support for the ToM hypothesis for autism by using an alternative test to false belief tasks. Using faux pas tests helps to revalidate claims made by the theory, which could have been considered questionable if relying purely on false belief tests. As a result, these tests provide strong support for the theory that individuals with autism have deficits in TofM.

The TofM hypothesis for autism is also strongly supported by research into joint attention ability in autism (Baron-Cohen, 2009). Evidence suggests that joint attention and TofM ability are linked, such that early joint attention ability is a precursor to TofM ability (e.g. Charman et al., 2000). Furthermore, research has indicated that joint attention skills are significantly impaired in infants with ASD. For example, Charman (2003) investigated joint attention ability in infants (aged 20 months) with autism, and found significant deficits. Charman then assessed the social and communication symptoms of these infants at 42 months, and found that symptom severity was correlated with joint attention disability indicated at 20 months. Colombi et al. (2009) also found that young autistic children show joint attention deficits. As joint attention ability is considered to be a precursor for TofM ability, this suggests that TofM deficits are present from infancy or birth in individuals with ASD. This thereby provides support for the TofM hypothesis as a causal account of autism, by suggesting that a TofM deficit may be a primary deficit.

This concept is supported by research into symptom severity and TofM ability. In 2003, Tager-Flusberg et al. used a battery of tests to assess TofM ability and symptom severity in autistic participants, and found that severity of social and communication symptoms was partially explained by TofM impairment. These conclusions imply that TofM deficits are causal in producing the social and communication impairments shown in ASD, thereby supporting claims made by the TofM hypothesis that a deficit causes symptoms associated with the condition.

In 2007, Tager-Flusberg presented one of the most comprehensive reviews of the hypothesis to date. She assessed evidence which addressed key issues associated with the theory (for example, why some children pass false belief tasks, and whether a TofM deficit can account for all deficits associated with autism), evaluating a wide range of evidence (including cognitive testing and neurological evidence). She concluded that whilst there is a TofM deficit in autism, the hypothesis cannot account for all deficits in social-affective information processing associated with the condition.

The issue raised by Tager-Flusberg (2007) is perhaps the most important weakness in the TofM hypothesis. Whilst the theory provides an explanation for social and communication deficits typically found in ASD (as successful communication skills require TofM ability, Baron-Cohen, 2009), there is no evidence to suggest that a TofM deficit may account for symptoms such as repetitive behaviour, obsessively pursued interests, and dislike of change (Baron-Cohen, 2009). Joseph and Tager-Flusberg (2004) investigated TofM deficits and repetitive behaviour in ASD, and found no correlation between deficits and severity of repetitive symptoms, thereby supporting the suggestion that the TofM hypothesis cannot account for repetitive behaviour. As a result, the theory cannot be considered comprehensive in terms of explaining all possible symptoms associated with autism.

Another issue associated with the TofM hypothesis is that individuals with other disorders have also shown deficits in TofM. For example, Brune (2005) found that individuals with schizophrenia showed significant impairment in TofM, and associated impairments in social behaviour and emotion recognition, similar to deficits shown by individuals with autism. The TofM hypothesis for autism provides no explanation for how autistic individuals are unique in their symptoms, when other conditions exist which show similar TofM impairment.

A theoretical constraint associated with the TofM hypothesis is the potentially ambiguous nature of the term ‘TofM’. The hypothesis is complicated by debates surrounding the nature of TofM ability in typical individuals. The two principle theories of TofM (‘Theory Theory’ and ‘Simulation Theory’) explain autism in terms of an inability to form thoughts, representations, and hypotheses about the social world (‘Theory Theory’), or alternatively in terms of an inability to simulate the mental states of others (‘Simulation Theory’) (Currie, 1996). Until an absolute theory of TofM can be established, any evidence in support of a TofM hypothesis for autism remains ambiguous in terms of exactly what deficits individuals with autism have (i.e. deficits described by ‘Theory Theory’ or deficits suggested by ‘Simulation Theory’).

When considering the literature investigating the TofM hypothesis, it is worthwhile acknowledging the wide body of neurological research. Research in this field provides strong support for a TofM deficit in ASD, with studies such as those by Kana et al. (2009), Mason et al. (2008), and Dapretto et al. (2006) suggesting abnormal brain activation in autistic participants in brain regions associated with TofM. However, research into a neurological basis for autism has also emphasised the issues associated with the TofM hypothesis. For example, Hadjikhani (2007) state that it is unclear whether dysfunction in brain regions associated with TofM could explain behavioural symptoms shown in autism (such as repetitive behaviour and concrete thought), or how structural and functional brain abnormalities originate in autistic individuals. This research highlights the key strengths and weaknesses associated with the TofM hypothesis as a cognitive account of autism, and reinforces cognitive research suggesting the hypothesis cannot be considered as a comprehensive account of the condition.

Summary

The preceding review provides strong evidence to support a deficit in TofM in autism, which is likely to be present from birth. The review also suggests that symptom severity can be at least partially accounted for by degree of TofM deficit. However, there are significant issues when considering the TofM hypothesis for autism. These issues include an inability of the hypothesis to explain how non-social symptoms can be accounted for by a TofM deficit, a lack of explanation of the significant symptom variability shown in the autistic population, as well as theoretical constraints. Until these issues are addressed, the TofM hypothesis cannot be considered as a comprehensive account of autism.

Alternative Theories of Autism

The issues associated with the TofM hypothesis led to a range of alternative theories of autism being developed, which must be considered when reviewing literature investigating the TofM hypothesis. One of the first theories to be proposed since the TofM hypothesis was the Executive Dysfunction theory (Ozonoff, Pennington and Rogers, 1991). This theory contradicted the TofM hypothesis, in that it proposed that individuals with autism have deficits in executive function ability (Yerys et al., 2006). This theory became popular in autism research, as the theory appeared to provide more comprehensive accounts of the autistic traits (both social and non-social) presented in these individuals daily (Hill, 2004). Furthermore, a large number of investigations found executive dysfunction in autistic individuals. For example, Ozonoff and McEvoy (2008) examined cognitive development in autistic individuals over a three-year period, and found significant deficits in executive functioning. McEvoy, Rogers and Pennington (2006) also found a deficit in executive functioning in autism.

However, recent research has suggested that executive dysfunction may be secondary to the autism. Yerys et al. (2006) investigated executive function in young children with ASD and found no evidence of dysfunction, suggesting that executive dysfunction in adults with ASD has developed gradually and is not a causal factor of the condition. Dawson et al. (2002) also found evidence for normal executive function ability in young autistic children, and concluded that dysfunction found at a later age was a consequence of living with the disorder. As a result, whilst this theory initially appeared to provide a valid and reliable alternative to the TofM hypothesis for autism, there is now strong evidence to refute executive dysfunction as a causal account of autism, and this research cannot be considered to provide evidence to invalidate the TofM hypothesis.

A more successful theory was proposed by Baron-Cohen (2008) as a development of the TofM hypothesis of autism. The Empathizing-Systemizing theory proposes that individuals with autism have deficits in understanding other’s mental states and responding appropriately (empathy), but have an average or above average systemizing ability (i.e. using rules and regularities to analyze and construct systems). As a result, autistic individuals show deficits in social ability (caused by empathy deficits) and non-social symptoms such as repetitive behaviours, obsessively pursued interests, and dislike of change (caused by average or above systemizing ability) (Baron-Cohen, 2009).

This theory is particularly successful for several reasons. From a methodological perspective, the theory provides an explanation for both social and non-social symptoms of autism, a key issue in the TofM hypothesis. The theory also explains the uniqueness of autism, such that whilst other conditions may show deficits in empathy (e.g. schizophrenia, Montag et al., 2007), no other condition also shows average or above average systemizing ability (Baron-Cohen, 2009).

The theory is also strongly supported by research. For example, Lawson et al. (2004) investigated empathizing and systemizing in participants with ASD and typically developing controls, and found those with ASD were significantly worse at empathizing than controls, but were not significantly different in their ability to systemize. Wakabayashi et al. (2007) found individuals with ASD were impaired in their ability to empathize, but significantly better than typical controls in their ability to systemize. This theory has therefore become a valid and reliable alternative to the TofM hypothesis for autism.

Conclusions

This essay has reviewed the literature which critically examines the TofM hypothesis for autism. It is clear that whilst there is adequate evidence to suggest a TofM deficit in autistic individuals, the hypothesis cannot account for non-social elements of ASD. New theories of autism must incorporate the concept of TofM deficits, whilst aiming to account for all social and non-social symptoms, as well as symptom variability and severity. A greater emphasis on the understanding how cognitive deficits associated with autism generate the symptoms shown in the condition would have important implications for the development of interventions aimed at improving social skills and symptom management in individuals with autism.