The Use of rTMS to Improve Theory of Mind Among Adults With Autism and Asperger's Disorder

Autism and Asperger's disorder are life-long neurodevelopmental disorders that involve significant social, communicative, and behavioural abnormalities. Social impairments are arguably the most debilitating aspect of these disorders, and persist into adulthood, creating difficulties in various domains including employment, education, and interpersonal relationships. This impairment is also frequently associated with additional mental health concerns, including depression and anxiety, with the former affecting as many as 40% of individuals with autism (Stewart, Barnard, Pearson, Hasan, \& O'Brien, 2006). It has been suggested that the core deficit underlying social dysfunction in autism and Asperger's disorder is impairment in "theory of mind" (ToM) (Baron- Cohen et al., 2005).

An important aspect of social cognition, ToM refers to the ability to use environmental cues to infer others' mental states. It includes the recognition that other individuals experience thoughts, feelings, intentions, and desires that may be different to our own. ToM is crucial to social functioning, allowing us to predict others behaviour and gain an appreciation of their perspective, thereby facilitating effective social interactions.

ToM has been assessed using an array of behavioural tasks. Such tasks typically illustrate a social scenario, and the aim of the task is to identify various aspects of a character's mental state. Increasingly, researchers distinguish between cognitive ToM, which encompasses inferences of another individual's beliefs, knowledge or memory, and affective ToM, which is concerned with predicting others' emotional states (Shamay-Tsoory \& Aharon-Peretz, 2007).

Cognitive ToM measures include false-belief tasks, successful completion of which requires an understanding that another individual can hold a mistaken belief (first-order false-belief), and that are therefore different to their own belief, or that another individual can hold a mistaken belief about the mental state of a third person (second-order false-belief; Harrington, Siegert, \& McClure, 2005). Another cognitive ToM task involves inferring intention from a series of short animations that demonstrate interactions between non-human objects (e.g., geometric shapes; Castelli, Frith, Happe, \& Frith, 2002).

By contrast, affective ToM measures include the "eyes task" (Baron- Cohen, Wheelwright, Hill, Raste, \& Plumb, 2001), which requires the inference of another individual's emotional experience through the presentation of their eyes only. The eyes task is considered a ToM measure in that it involves the inference of another individual's mental state from environmental stimuli, but unlike false-belief tasks there is a central affective component.

An additional ToM test that is thought to involve both cognitive and affective aspects of ToM is the faux pas test (Stone, Baron- Cohen, \& Knight, 1998), which assesses the ability to recognize a faux pas within the context of a series of vignettes describing social interaction. This ability requires an understanding of the mental state of two people: the individual who unwittingly committed the faux pas (cognitive ToM), and the individual who might be offended or insulted by the faux pas (affective ToM). Thus, ToM tasks are all concerned with the inference of others mental states, but whereas cognitive ToM tasks require inferences concerning others beliefs, affective ToM tasks require inferences concerning others emotions.

The brain basis of ToM is increasingly well established; recent fMRI studies utilizing cognitive ToM tasks indicate widespread frontotemporal activity. This includes the temporal poles bilaterally and the right superior temporal sulcus (STS), which seem to reflect the role of memory and visual perception respectively (Gallagher \& Frith, 2003; Schultz \& Robins, 2005). It is the activation within the bilateral medial prefrontal cortex, however, that is thought to reflect a unique contribution to ToM. Gallagher and Frith (Gallagher \& Frith, 2003) suggest that a distinct area within the medial prefrontal cortex, the bilateral anterior paracingulate cortex (x = 8, y = 54, z = 1), is the crucial region for ToM ability. While the medial prefrontal cortex is strongly implicated in cognitive ToM, neural networks recruited during affective ToM are somewhat different (despite substantial overlap), and involve the amygdala, orbitofrontal cortex, and ventromedial frontal lobes in addition to temporal regions. A recent lesion study of ToM supports this neuroanatomical dissociation between cognitive and affective ToM (Shamay-Tsoory \& Aharon-Peretz, 2007).

There is substantial evidence for impaired ToM in autism and Asperger's disorder. Individuals with these disorders consistently demonstrate delayed or impaired performance on both cognitive and affective ToM tasks (Baron-Cohen, Jolliffe, Mortimore, \& Roberstson, 1997; Baron-Cohen, Leslie, \& Frith, 1985; Kaland, Smith, \& Mortensen, 2007; Rogers, Dziobek, Hassenstab, Wolf, \& Convit, 2007). Indeed, it has been suggested that ToM impairment is the core deficit of autism, and that it underlies the characteristic social and communicative symptoms (Baron-Cohen et al., 1985; Tager-Flusberg, 2001). Imaging studies among individuals with autism demonstrate reduced activity with the medial prefrontal cortex, particularly during cognitive ToM tasks (e.g., Castelli et al., 2002).

Justification for project

Social dysfunction, including impaired ToM, is arguably the most debilitating aspect of autism and associated disorders. There are currently few treatment options for targeting these social deficits in autism and Asperger's disorder. An increasing prevalence of autism spectrum disorders (Chakrabarti \& Fombonne, 2005) ensures that the next decade will see an unprecedented number of adult cases; it is therefore imperative that we now develop effective therapeutic techniques to reduce the impact of core autistic symptoms. Functional imaging data reviewed above suggests that the bilateral medial prefrontal cortex is a promising site for neurobiological intervention in these groups to improve ToM (particularly cognitive ToM) and broader social functioning.

We suggest that repetitive transcranial magnetic stimulation (rTMS) may be an effective therapeutic technique to enhance medial prefrontal cortex activity and improve ToM in individuals with autism, with anticipated benefits to social aspects of this disorder. rTMS is a safe and non-invasive technique that involves the repeated administration of localized magnetic pulses to a distinct cortical region. When applied at high-frequencies, rTMS enhances cortical activity at the site of stimulation. This is particularly useful in a clinical setting, where rTMS is administered to a cortical region that functional imaging suggests is underactive in a particular psychiatric disorder (e.g., left dorsolateral prefrontal cortex [DLPFC] in depression), an approach that is effective in alleviating core symptoms of the disorder. Conversely, low-frequency rTMS decreases cortical activity at the site of stimulation.

Hypothesis/research questions

It is hypothesized that two weeks of high-frequency rTMS to the medial prefrontal cortex, which is expected to increase cortical activity within this region, will significantly improve ToM performance (particularly cognitive ToM) among individuals with autism or Asperger's disorder. Improvements in social function are expected to be seen one month after the last treatment session.

Methodology including project design and sequence of procedures

This project will involve the administration of rTMS via a "deep TMS'" H-system coil (Brainsway Inc., Jerusalem, Israel). This is administered by a modified TMS coil, but uses our existing stimulator equipment (Magstim), which has been used in our previous rTMS studies. This H-system coil will allow deeper penetration of the TMS pulse, which will ensure that we accurately stimulate the part of the medial prefrontal cortex that is most relevant to ToM. Safety and efficacy studies of deep TMS indicate equivalence with the conventional TMS (figure-of-eight) coil (Levkovitz et al., in press; Zangen, Roth, Voller, \& Hallett, 2005).

This study is an rTMS trial to improve ToM and social function among adults with autism or Asperger's disorder. Adults diagnosed with either autism or Asperger's disorder will be randomized (via a computer generated number sequence) to either active rTMS or a sham condition, under strict double-blind conditions. Based on prior imaging data, rTMS will be administered to the bilateral medial prefrontal cortex. This study will involve deep TMS via the use of the H-system coil.

Prior to the rTMS sessions participants will undergo high-resolution (3T) structural and functional MRI. The fMRI component will again include cognitive and affective ToM measures. Using reliable neuronavigational methods, each individual's brain scan will be used to determine the site on the scalp that lies above the cortical region of interest (bilateral medial prefrontal cortex), and which will therefore be targeted for rTMS.

Prior to the first rTMS session, the participant will complete a battery of neurobehavioural tests that assess theory of mind (e.g., false-belief task, eyes test, faux pas test). In order to gauge whether there is any impact on the broader realm of social functioning in autism, participants will also be administered the "Autism Spectrum Quotient" (AQ; Baron-Cohen, Wheelwright, Skinner, Martin, \& Clubley, 2001), a fifty item self-report measure.

At the beginning of the first rTMS session, we will determine resting motor threshold (RMT) by administering TMS to the left motor cortex, and recording (via EMG) the response from the right thumb muscle. RMT is again defined as the lowest intensity at which 3 out of 5 consecutive TMS pulses results in an EMG response greater than 50 µV.

Using the deep TMS coil, participants will receive 15 minutes of either active or sham rTMS (5 Hz) each consecutive weekday for a two-week period. During each session, there will be 10 seconds of 5Hz rTMS, followed by a 20 second rest. Thus, for each session there will be 30 10s trains of 5Hz rTMS administered.

After the final rTMS treatment, the three ToM measures and the AQ will be readministered. These measures will then be administered one-month after the final rTMS treatment. The researcher administering these measures will be blind to the participant's treatment (i.e., active or shame).

After the completion of the one-month follow-up, participants will be unblinded to their study condition. Those who were allocated to the sham condition will be given the opportunity to complete two weeks of active rTMS treatment.

Statistical analyses and sample size calculations

As the effect of rTMS on ToM has not been investigated, it is somewhat difficult to conduct a power analysis. However, based on the sample size of previous rTMS studies that involves modulation of a cognitive ability (e.g., Giovannelli et al., 2006; Kwan et al., 2007; Tunik, Frey, \& Grafton, 2005) or treatment for psychiatric disorders (e.g., Fitzgerald et al., 2006; Horacek et al., 2007), we are confident that our sample size will be sufficient to detect a significant result.

ANOVA will be used to investigate the effect of rTMS to the medial prefrontal cortex on ToM.

Outcome variables will not only include ToM measures, but also an autism-relevant rating scale to determine social functioning and associated aspects. Thus, we will conduct repeated measures ANOVA for each measure (pre, post, and one-month post rTMS), with group (active vs. sham) included as the between-subjects variable.