The ability to mentalize, or ability to attribute mental states to others, has been argued to be at the basis of autism and in most cases of Asperger's Syndrome patients. This "Theory of Mind" hypothesis was proposed and tested by Baron-Cohen, Leslie and Frith in a first-order Theory of Mind task (1985) and in a second-order Theory of Mind task (1989). The first-order test required the subject to recognize a character's false belief in a story about two dolls (Sally and Ann). If Sally hides a marble in her basket and leaves the room, and Ann moves the marble to another box, "Where will Sally look for the marble?" 80% of the autistic subjects failed this test. Very few autistics passed their more advanced second order test which involved understanding "what Mary thinks John thinks", passed by normal children between the ages of 5 and 7. However many gaps and limitations are left in the Theory of Mind Hypothesis of autism (Dahlgren and Trillingsgaard, 1996) and much literature has appeared addressing the problems in the last five to eight years. In an attempt to address the recent studies arguing the validity of the traditional theory of mind tests, which many autistics passed, Baron-Cohen, Jolliffe, Mortimore, and Robertson (1997) designed a new "adult test of theory of mind ability", where information about a person's mental state can be acquired only from photographs of a person's eyes. Other literature has argued that there may be other deficits, more primary to the Theory of Mind deficit, such as inability to disengage from an object (Hughes and Russel, 1993), since recent studies have shown a lack of universality in autistics failing the Theory of Mind tests (Yirmiya, Solomonica-Levi, Shulman, and Pilowsky, 1996). There have been other studies attempting to attribute an anatomical region of the brain to the "theory of mind". I will present one such study by Happe, Ehlers, Fletcher, U. Frith, Johansson, Gillberg, Dolan, Frackowiak, and C. Frith (1996).

    In previous studies Dahlegren and Trillingsgaard demonstrated that between 38 and 60% of autistics passed first-order false belief tests in preliminary studies. In this study they propose to investigate the presence of theory of mind deficits in autistic children within a normal range of intelligence. Their performance on false belief tasks is compared to that of mentally and chronologically age-matched normal (control) children and children with Asperger's syndrome. The first aim of the paper was to determine whether autistic children without mental retardation had the same difficulties on theory of mind tasks as found in earlier studies (Baron-Cohen et al., 1985, 1989). The second aim of the paper was to compare children with Asperger's with autistic children, when both were within a normal range of intelligence.

    Three groups participated in this study. The first was a group of twenty, diagnosed with autism having a mean chronological age of 10.57 years (range 6.25-15.5) and mean IQ of 98.85 (range 77-129). The second group was composed of twenty children with Asperger's syndrome. The chronological age varied from 7.25 to 12.66 with a mean of 10.15. Their IQ varied from 80 to 128, with a mean of 100.35. The third group of twenty children was non-disabled with a mean age of 8.98 and IQ of 116.55. Though no significant differences were found in mental or chronological age, the group of autistic children had a significantly lower verbal IQ than the group of Asperger's children. The subjects were given the same fist-order and second-order theory of mind tests used by Baron-Cohen et al. (1985, 1989). Amazingly, 90% of the autistic children and 85% of the Asperger's children passed the first-order theory of mind test compared to 100% of normal control children. 60% of both autistic and Asperger's children passed the second-order theory of mind test, compared to 90% of normal control. Though the control performed better than autistics, confirming earlier studies in problems with these tasks, the 90% passing is much higher than any previous study. Intellectual capability and verbal intelligence may explain the discrepancy to the Baron-Cohen studies, the subjects of which may have been at a lower level. Addressing the second aim of their study, the groups with autism and Asperger's did not differ significantly in either of the theory of mind tests. This may also be due to the higher than normal verbal performance of the autistic subjects in this study.

    Thus, it appears that verbal competence in combination with intellectual level can be the best predictor for ability to mentalize. "That is, the probability that children with autism and Asperger's will solve theory of mind tasks increases with the level of intelligence, verbal intelligence, and chronological age." This paper presents the limitations of the theory of mind hypothesis in explaining autism, and that other deficits may be more primary in the autistic continuum. The study also suggests that these tests cannot differentiate between high functioning autism and Asperger's syndrome (AS) and that they may lie on the same continuum.

    In an attempt by the proponents of the theory of mind hypothesis to support their work and ideas, Baron-Cohen et al. designed a new "adult test of theory of mind ability" (1997). They suggest that the original first-order and second-order theory of mind tasks were only probes for four-year-old and six-year-old levels of skill. The paper argues that the tasks cannot reveal much when applied to autistics with a mental age higher than six, and should not be considered complex. In light of this, this study designed and used a new "adult" test of theory of mind in adults with autism and AS. The first part of this task involves an Eye Task, in which subjects look at a photograph of the eye region of the face and make a forced choice between two words describing what the person in the picture might be thinking or feeling (i.e. sad, happy, afraid, angry). In the latter parts of the Eye Task, the subjects have to attribute more 'complex' mental state terms to pictures, such as arrogant, scheming, or planning. It was the aim of this paper to investigate the performance of autistic, AS, and Tourette Syndrome (TS) adults on this adult test of theory of mind. A secondary aim of the paper was to look at sex differences in the normal group. These groups were chosen due to their similarities in the disorders, their stage of development, their similar frontal pathology, and that males are more affected than females.

    Three groups were used as in the previous study. The first included patients 16 subjects (mean age 28.6 ± 9.7 S.D.) with high-functioning autism (n = 4) or AS (n = 12) and a sex ratio of 13:3 (m:f). They were all of normal intelligence (mean IQ 105.31 ± 13 S.D.). Group two consisted of 50 normal age-matched individuals (mean age 30.0 ± 9.12 S.D.) with a 1:1 sex ratio and without IQ information being collected. The third group was one of 10 TS adults, 8 males and 2 females, with a mean age of 27.77 ± 7.8 S.D. and a mean IQ of 103.5 ± 10 S.D. The Eye Task as well as control tasks of gender recognition and basic emotion recognition tasks, which do not require mentalization, were conducted with all subjects.

    The normal group responded correctly to a mean of 20.3 ± 2.6 SD out of 25 pictures correctly. Group one with autism and AS scored a 16.3 ± 2.9 SD mean on this task. And group 3 with Tourette Syndrome scored a mean of 20.4 ± 2.63 SD on the Eye Task. All subjects performed well on the control gender recognition task. It seemed that males of the normal group scored slightly lower (18.8 ± 2.53 SD) than females in the normal group (21.8 ± 1.78 SD). Based on these values and their standard deviations, the significance of the difference between the scores of the groups is not apparent. The paper argues that group one (autistics and AS) differs from the other two control groups, normal and TS. They also claim that females performed significantly better than males in these tasks, as this is related to the ratio of females to males with autism. I do not see how a score of 16.3 means that there is definitely a theory of mind deficit. Yet, the difference as compared to control does indeed indicate a problem with the ability to mentalize. This paper thus provides a more advanced test of the theory of mind, as compared to prior tests of theory of mind which have had mental age ceilings.

    Hughes and Russel (1993) argued that there might be other deficits, more primary to the Theory of Mind deficit, such as inability to disengage from an object, which result in the failure of deception tests and false belief tests by individuals with autism. Russel et al. (1991) demonstrated potential for this "disengagement" hypothesis in a series of tests of deception (window task). The window task required the subject to watch where a piece of chocolate was being placed under one of two boxes with windows facing the subject. An opponent was situated on the other side of the boxes and was unaware of the placement of the reward. The subject has to choose between the boxes, seeing which is empty and which is not. The subject receives the candy if he/she points to the empty box, and the opponent receives it, if the subject points to the box with the chocolate. So the subject must learn to deceive the opponent in order to get the chocolate.

    In the more recent paper in 1993, Hughes and Russel put the two hypotheses of "theory of mind" and "disengagement" against each other, offering alternate hypotheses. If the "theory of mind" or metarepresentation hypothesis is correct, subjects should perform better if the element of deception is removed in the windows task. If the disengagement hypothesis is correct, removal of the deceptive element should not change the difficulty of the task for autistics. So in a series of experiments, the subjects had to simply point to the empty box to obtain the chocolate reward with (1) and without (2) the presence of an opponent. To eliminate the possibility that the disengagement were purely at a motor level, the subject had to say the color of the box in a different set of experiments with (3) and without (4) the deceptive element, the opponent.

    Sixty subjects diagnosed with autism (mean age 13.2) and sixty mentally handicapped subjects (mean age 11.4) were used in this study. 15 subjects from each group were assigned to each of the four studies, requiring a non-verbal response with or without an opponent, and requiring a verbal response with or without an opponent. The removal of an opponent, in both the verbal and nonverbal response experiments, caused an decrease in percentage of non-autistics passing the test (form 93.3% to ~ 60%), while no significant increase in percentage was seen in the autistics (from 20% to 6.7% in the nonverbal experiments and constant at 13.3% in the verbal experiments). This means that removal of the opponent did not facilitate the task for the autistics, thus supporting the disengagement hypothesis rather than metarepresentation. Opponent removal also made the task more difficult for the non-autistic MR subjects, which must have used the visualization of the opponent receiving the candy as negative feedback.

    It appears that autistic children fail tests of strategic deception, primarily due to their inability to disengage from focus and attention on objects. The inability to mentalize and deceive seems to come secondary to this attention problem in autistics. Thus, Hughes and Russel argue that "grasp of mental concepts may well depend on the adequate functioning of the executive system." Aside from proving their disengagement hypothesis, the investigators devised a great new device and tool of measuring executive system function alongside with metarepresentation.

    Yirmiya et al., in a study in 1996, showed that the theory of mind deficit, whether primary or secondary to other deficits, exists in individuals with mental retardation (MR) due to Down Syndrome and those with MR of unknown etiology as well. The group set out looking for a correlation between different tasks used to characterize theory of mind and aspects of intelligence. Four groups were used in this study, individuals with autism (n = 25), subjects with MR due to Down Syndrome (n = 19), subjects with MR of unknown etiology (n = 21), and normal children (n = 21). The subjects were first examined for varying measures of intelligence. Measures of chronological age, mental age, verbal ability (VMA) and performance ability (PMA) were taken as determined by multiple volcabulary and performance tests and the Wechsler Intelligence Scale for Children-Revised (WISC-R.F).

    Next, a series of theory of mind tasks were given to the subjects and their performance scored. Two value belief tests involved attribution of a mental state of like or dislike of a food or smell to a person. Two other fact belief tests asked the subject, who is aware of a set of information to attribute an unaware mind to a person to whom the information was unavailable. These tests are rather long and elaborate to be explained here in brevity. The first-order false belief test of Baron-Cohen et al. (1985) was used as well. The investigators used a very complex test of deception tasks, involving procedures adapted from Hala, Chandler, and Fritz (1991) to further experiment on the presence of theory of mind deficits. The deception task described does a great job of differentiating between autistics and normal subjects, 7% and ~ 80% of who respectively passed the test. The study concludes to state that individuals with autism display a more limited understanding of the theory of mind as compared to normal and mentally retarded individuals. In general, autistic individuals displayed more difficulty in theory of mind tasks compared to mentally retarded individuals. The theory of mind abilities seems to be correlated with verbal abilities in individuals with autism. The study revealed that normal children performed significantly better than MR individuals and MR individuals did not perform better than autistics in the tests, other than the value test. These findings support the idea that theory of mind deficits may not be limited to autism, but to other mental disorders as well.

    Among the studies present, which attempt to attribute an anatomical region of the brain to the "theory of mind", are several addressing the possible role of the frontal lobe in autism (Baron-Cohen, Ring, Moriarty, Schmitz, Costa, and Ell, 1994). I will present one such study by Happe, Ehlers, Fletcher, U. Frith, Johansson, Gillberg, Dolan, Frackowiak, and C. Frith (1996) which attributes autism to a less studied front part of the brain. This study used positron emission tomography to study activity patterns all over the brain. PET scans of normal subjects reading different types of stories were analyzed, using subtraction techniques, to reveal a brain region specific to metarepresentation from the stories. Previous studies by the lab have implicated an area in the left medial prefrontal cortex to be the center mediating mentalizing abilities in normal adults. Baron-Cohen et al. set out to perform such studies in individuals with high functioning autism and AS.

    Five right-handed males (mean age 24) with AS were used as subjects in this study. A set of three stories was given to the subjects to be read. The stories were physical stories, stories requiring metarepresentation (ToM), and unconnected sentences. PET scans of subjects were performed during these tasks and analyzed using subtraction techniques. When subtracting the PET of the physical story or the unconnected sentences from that of the Tom story in normal individuals, a region of the medial prefrontal cortex appears as active during the Tom. This is not true in individuals with AS, though they did show activation of an adjacent region, possibly showing compensatory activity. Though the frontal lobe dysfunction may be the cause of autism, more task analysis of executive functions and divergence of the components and brain pathways involved need to be done. This study thus provides evidence for a possible brain location, which may be the anatomical correlate of a "theory of mind" system.

    Though all of the papers reviewed above are on different aspects of the role of theory of mind in autism and how much of a role such a deficit has in the disorder, I found that many of the articles were inter-correlated. One paper presented the limitations of the theory of mind hypothesis in explaining autism (Dahlgren and Trillingsgaard, 1996) and that other deficits may be more primary in the autistic continuum. The study also suggests that the present tests, the first and second-order theory of mind tasks, cannot differentiate between high functioning autism and Asperger's syndrome (AS) and that they may lie on the same continuum. Baron-Cohen, Jolliffe, Mortimore, and Robertson (1997) not only pointed out such limitations as having mental age ceilings in the tests, but also provided a more advanced test of the theory of mind, to address the limitations.

    Hughes and Russel (1993) agreed with Dahlgren et al. (1996), in that many high functioning autistics were passing these first and second-order theory of mind tests. They took the perspective that since autistic children fail tests of strategic deception, but pass the false belief tests (first-order and second-order), there may be a deficit more primary to the "theory of mind" deficit. They suggested the subjects' inability to disengage from focus and attention on objects. Hughes and Russel argue that the inability to mentalize and deceive seems to come secondary to this attention problem in autistics. It is this executive function deficit, they and others argue, that lies at the basis of autism and they devised a great new device and tool of measuring such function alongside with metarepresentation. This new deception task they presented, is quite different from the mental state attribution task designed by Baron-Cohen et al. (1997). It seems that they use the same sort of subjects and results to prove different and opposing points. They do so convincingly.

    Yirmiya et al., in a study in 1996 also address the points raised by Dahlgren et al. (1996) about the continuum of disease, but rather apply it to mental retardation. Though autistic individuals displayed more difficulty in theory of mind tasks compared to mentally retarded individuals in their studies, the study revealed that MR individuals did not perform better than autistics in the tests given by the administrators. In a very important and relevant study, Happe et al. (1996) implicated the medial prefrontal cortex as the possible brain location, which may be the anatomical correlate of a "theory of mind" system. This is consistent with other work done in the field, but is a anatomical tangent that I thought was important to present in a good review of the field.

    Though Baron-Cohen and Fritz and Jottliffe have tried to support their 12 year old theory of mind hypothesis over the years, a new group of researchers is trying to address the limitations of such a theory and existing tests for it. This research has taken two stands, one, that another deficit may be co-existent in autism, which results in the limitations of the theory of mind explanations, and the other that another deficit may be more primary to theory of mind in autism. Both groups have as a way of addressing the limitations of prior research addressed that the continuum of autism is broader than imagined in the past. With research in the anatomy of the dysfunction and deficits, an avenue of treatment will hopefully arise from the biomedical field soon.
 

Works Cited