Abstract
Theory of Mind (ToM) is the capacity to make attributions of mental states such as thoughts, feelings, and beliefs of others. Individuals with autism spectrum disorder are often characterized by delayed or impaired Theory of Mind development and poor social skills. A positive correlation between Theory of Mind skills and social skills exists, but effective interventions that generalize these skills have proven to be a challenge in the field. Furthermore, few effective interventions exist to increase Theory of Mind skills in adolescents with autism spectrum disorder. The Teaching Theory of Mind curriculum is a 12-week intervention aimed at increasing social understanding in children and adolescents with deficits in Theory of Mind. To date, there have not been any empirical studies to test the efficacy of the curriculum. The purpose of this study is to investigate the efficacy of the Teaching Theory of Mind curriculum on adolescents with autism spectrum disorder and explore the impact of the intervention when addressing various domains of social understanding. Results, though tentative, indicated that parents saw improved social understanding following intervention, though direct measures of Theory of Mind with adolescent participants did not change significantly.
Introduction
Theory of Mind (ToM) is a complex, developmental phenomenon that includes the capacity to make attributions of mental states such as thoughts or feelings. Sometimes referred to as “social understanding,” such attributions may be utilized to predict behaviors, rationalize actions, and make sense of the social world (Pavarini, de Hollanda Souza, & Hawk, 2013). Full development of ToM permits one to infer the thoughts, goals, and desires of others, and may also be used to understand lying and deception, and has been characterized as a set of competencies that are necessary for “successful navigation of the social world” (Conway & Bird, 2018). ToM is conceptualized as a developmental outcome emerging around three to five years of age that serves as an important component of social competencies. A growing body of research has documented positive correlations between ToM skills and social skills, though associations between advanced ToM abilities and antisocial behavior have also been observed (e.g., using ToM skills to deliberately deceive others; Lonigro, Fiorenzo, & Baiocco, 2014).
Stages of development
A true understanding of what others are thinking or feeling evolves through multiple developmental achievements (Peterson, Wellman, & Slaughter, 2012; Wellman & Liu, 2004). Pons, Lawson, Harris, and deRosnay (2003) theorize that language provides a mental construct so that emotions are represented and understood, and that language ability is associated with understanding others’ emotions. In addition, ToM is involved in higher order tasks such as understanding sarcasm, irony, and figurative interpretations of language (Peterson et al., 2012). ToM ability is historically measured and attributed to performance on false-belief tasks; however, a single measurement may not capture the complete characteristics of ToM. Ahmed and Stephen Miller (2011) report that distinct cognitive mechanisms are employed during the engagement of different ToM tasks.
The first precursor to ToM skill development found by Wellman and Liu (2004) is evaluating desires. Children understand that two people may have different wishes about the same item (e.g., he wants the cookie but she does not want the cookie). Conceptually, children develop an understanding of desires (i.e., what someone wants) before they understand other’s beliefs (i.e., what someone thinks or believes; Wellman & Liu, 2004). Initial development about other’s beliefs, or first-order ToM task development, is limited to simply understanding another person may have a different belief about the same item, but the child does not know the validity of either belief.
In typically developing children, tangible evidence of ToM emerges during preschool years with most four- to five years olds successfully completing a false-belief task; yet, most three years olds have not developed the necessary skills to complete such a task (Peterson et al., 2012). This suggests rapid development in perspective understanding over the course of one to two years; furthermore, Wellman and Liu (2004) reported that children attain underlying conceptual skills that precede ToM acquisition. In addition, skills for typically developing children are developed in a consistent progression with success on higher scaled tasks indicating mastery on lower scaled tasks (Peterson et al., 2012; Wellman & Liu, 2004).
Sequentially, the child then understands a task about seeing objects in a box and appreciates another person does not know the contents because he or she did not see the inside of the box. Second-order ToM tasks develop following the acquisition of the previous first-order ToM tasks. Second-order ToM tasks include the ability to understand a person’s belief about another person’s belief (Miller, 2009). For example, in the case of a false belief, a child may understand that a person thinks that there is an item in a box because the person did not see it to be removed from the box. Children then develop hidden emotion ToM skills and recognize a person may feel an emotion but display a different emotion (Wellman & Liu, 2004).
While the children diagnosed with autism achieved ToM task mastery in an atypical sequence, they demonstrated skill development in a consistent manner. Peterson et al. (2012) found children with autism achieve the hidden emotion task before the false-belief task. A total of 80% of the children in this study were found to attain ToM tasks in a dissimilar pattern compared to their typically developing peers. In addition, children with autism outperformed all other groups on the hidden emotion task. A more advanced level of ToM, assessed with second-order false-belief tasks, requires a “recursive understanding of one person’s false belief about another’s belief; for example, the boy thinks (falsely) that the girl thinks an object is at A when it is really at B and that she knows this” (O'Reilly, Peterson, & Wellman, 2014, p. 1865; Table 1).
Typically developing stages of ToM.
Note: ToM task mastery occurs along a continuum with age approximations, especially with second-order tasks. ToM: Theory of Mind.
It is hypothesized that engagement in a structured intervention targeted to increase ToM and social skills will improve overall social understanding. As individuals with autism spectrum disorder (ASD) often show atypical ToM skill development, it is hypothesized that participants will show improvements across both primary foundational ToM skills in addition to more advanced ToM skills. By targeting the improvement of the underlying developmental process of ToM and improving the overall social skills, researchers also anticipate greater generalization as observed by parent report and direct measurement.
Brief review of ToM interventions
Effective interventions to promote and improve ToM capacities must consider and evaluate underlying components of ToM (Westby & Robinson, 2014). Current evidence suggests that the elements of ToM can be effectively taught to children (Allen & Kinsey, 2013; Peters & Thompson, 2018). From a social and pragmatic communication point of view, “the goal is to teach the cognitive, linguistic, and social/emotional skills that underlie ToM so that individuals can use these skills to communicate and interact more effectively” (Westby & Robinson, 2014, p. 365), but this goal can be applied to other children with disabilities such as ASDs.
Thought-bubble training is an intervention technique that has shown to help children develop an alternative to a ToM (Paynter & Peterson, 2013; Wellman et al., 2002; Wellman & Peterson, 2013). Children can learn to understand mental states by using a picture of a person with a thought-bubble. It allows a more concrete visual representation of what a person is thinking (Wellman et al., 2002). Similar to the use of thought-bubbles, researchers such as Swettenham, Baron-Cohen, Gomez, and Walsh (1996) used picture-in-the-head training. In the study, they adapted a mannequin head for training so that a picture could be inserted into a slot inside the head. They found that after the training, children were able to make behavior predictions but were still unable to infer mental states (Swettenham et al., 1996). While picture-in-the-head training has shown improvements in false-belief understanding, Wellman et al. (2002) found that thought-bubble training may show even better results that were more generalizable.
Throughout intervention, there are many tasks that can be used to test for the understanding of concepts such as false belief. Results of interventions designed to teach ToM or social perspective taking have been variable. Southall and Campbell (2015) reviewed the most current literature on systematic instruction, group, and technology-based interventions geared toward teaching ToM. Systematic instruction interventions included tasks such as teaching first- and second-order ToM tasks, role-playing, modeling, or other social perspective-taking interventions. Overall, students with autism demonstrated improved performance on specific tasks following intervention; however, generalization to natural settings did not appear to be evident when rated by others following the intervention. For instance, one study by Chin and Bernard-Opitz (2000) demonstrated an increase in back-and-forth conversational skills following intervention, but ToM scores did not change. Similarly, Southall and Campbell (2015) reviewed multiple studies that reported mastery of concrete ToM rules through group intervention. However, application in the natural setting was either not evaluated or social perspective-taking skills did not improve when measured in other settings.
Technology-based interventions through video modeling have also been utilized to improve ToM with some effectiveness. Video modeling was utilized to help students with autism recognize specific social cues. Like the other two types of interventions, in vivo application was not evaluated or participants demonstrated little overall improvement. However, Southall and Campbell (2015) report that the method is effective in increasing targeted skills.
The gap between success in specific skill and application in real-world settings suggests that future research may need to be geared toward teaching comprehensive social skills across settings with multiple opportunities for practice. In fact, systematic review of training programs for emotional recognition found that many fail to generalize beyond the training to natural social environments (Berggren et al., 2018; Fletcher-Watson, McConnell, Manola, & McConachie, 2014).
Southall and Campbell (2015) suggest completing functional behavioral assessments to identify areas of specific challenges for the student in their typical environments. It is then recommended to progress monitor following intervention within his or her typical environment. In addition, Southall and Campbell suggest that intervention could begin in one typical setting and then be implemented gradually across settings. Studies have generally been limited in monitoring the fidelity of intervention methods (Southall & Campbell, 2015); implementing techniques across settings with repeated practice would then warrant thorough evaluation of the intervention process.
Efforts to improve ToM development have been the focus of many ASD treatment programs due to its importance in child development and the challenge it presents to many children with ASDs. There is also a need to determine whether interventions that aim to improve ToM among those with ASD are effective and can generalize results across settings.
For this study, a ToM curriculum (Teaching Theory of Mind: A Curriculum for Teaching Children With High Functioning Autism) was selected because it represented an intervention that had not been previously evaluated, even though the curriculum has been available since 2012. Moreover, the design of the intervention (group setting, delivered over multiple weeks through systematic instruction) suggested that this particular approach may be beneficial. Given the need for effective ToM interventions that demonstrate improvements generalized to daily-life settings (e.g., home, school, and community), we sought to determine whether this particular series of group sessions yielded a positive, measurable impact on ToM.
The curriculum selected for this study has several features that are appealing and separate it from other existing interventions. Teaching Theory of Mind: A Curriculum for Teaching Children With High Functioning Autism was developed as a structured but flexible intervention designed to be delivered in small groups of children aged 7 to 16 years. Group interventions for ToM are rare, as are interventions targeting adolescent populations. The curriculum includes colorful materials and instructions that are easy to follow.
Methods
Participants consisting of 12 adolescents (3 females and 9 males) between the ages of 12 and 17 years (mean (M) = 14.03, standard deviation = 1.80) were recruited as a convenience sample at a university-based autism clinic and a private school inclusive of children with disabilities. All participants had a reported diagnosis of ASD and were located at two sites (Texas and Florida). Participant diagnosis of ASD was not confirmed through independent assessment. Outside diagnosis of ASD was reported by the parent. Investigators also utilized their own professional judgment to confirm that participants indeed displayed the features of ASD. For this study, investigator focus was on administering the Teaching Theory of Mind intervention, a program developed for children aged 7 to 16 years (Ordetx, 2012). Although the materials presented in the Teaching Theory of Mind curriculum demonstrated good face validity, outcomes from the application of the program had not been formally assessed.
The Teaching Theory of Mind curriculum was developed “to incorporate a multi-sensory approach to developing the critical, basic aspects of theory of mind” (Ordetx, 2012, p. 9). The 12-week curriculum was designed for children aged 7 to 16 social cognitive challenges and follows a structured sequence of activities that teach a number of foundational skills, including joint attention, imitation, identification of emotions, reading social cues, cause and effect in social contexts, and several others.
The manual provides detailed instructions for each activity, including materials, targeted skills, and directions for implementation. In addition, each activity comes with a caregiver letter so that skills can be reinforced and generalized in home and community settings. One of the activities, Act it Out, each participant receives with a card with a phrase (e.g., I am so hungry; Let us get out of here; This is so heavy). Without speaking, participants take turns nonverbally acting out the phrase, while other participants try to guess correctly. The group facilitators encourage group members to stay connected and to note the facial expressions and gestures.
Each activity also has a cognitive–behavioral component in which facilitators make explicit the skill outcomes and applications and provide encouragement for the desired outcomes. For this study, there were two group facilitators: a licensed psychologist with 15 years of experience working with children and adolescents, including those with ASDs, and a specialist-level graduate student with training in counseling skills and psychological assessment. The group met weekly for 12 weeks, and each session lasted 30 minutes.
Measures
The Theory of Mind Inventory
The Theory of Mind Inventory (ToMI) was administered to each participant’s parent (N = 12). The ToMI is a 42-item survey that measures caretaker’s perceptions of child ToM and social understanding across the domains of ToM level (early, basic, and advanced) including emotional recognition, pragmatics, and mental state term comprehension. The ToMI was standardized on a sample of parent ratings of 124 typically developing children and parent ratings of 135 children with ASDs. The ToMI developers noted robust sensitivity, with more than 88% of the ASD sample falling at or below the 10th percentile compared to typically developing peers.
Hutchins, Prelock, and Bonazinga (2011) explained that the ToM levels describe various developmental stages of ToM measured by the ToMI. Early ToM is defined as the ability to read affect and share attention and is typically developed during infancy and toddlerhood (Hutchins et al., 2011; Tomasello, 1995). Reading affect involves the ability to accurately understand facial expressions and their meanings as well as sharing attention is the ability to demonstrate joint attention with a shared experience (Hutchins et al., 2011; Mundy & Newell, 2007). Basic ToM is defined as the ability to employ metarepresentation and developmentally related understandings and emerges around the age of 4 years (Abbeduto & Rosenbberg 1984; Hutchins et al., 2011; Kazak, Collins, & Lewis, 1997; Moore & Furrow, 1991). Metarepresentation is the understanding of mental representations of representations. It can include own representations or the representations of others and is required to understand how others’ behavior is guided by their beliefs and knowledge (Hutchins et al., 2011; Leslie & Frith, 1998). Advanced ToM is defined as the ability to employ complex recursion, understand the mind as active interpreter, and utilize social judgment. This level of ToM typically emerges between six and eight years of age (Hutchins et al., 2011). Complex recursive thinking involves understanding that one person’s belief or thoughts may be perceived in a particular manner by another person’s understanding or perspective (e.g., Ben thinks that John thinks; Hutchins et al., 2011) This skill is integral in the distinction between lies and jokes as it implies the understanding of indirect intention and purpose (Hutchins et al., 2011; Leekam & Prior, 1994; Sullivan, Winner, & Hopfield, 1995). Mind as an interpreter is the understanding that one’s mind and subjective interpretation affects the understanding of information (Carpendale & Lewis, 2006, p. 193). Social judgment is the ability to accurately judge social prompts through subtle social cues within the context of a social setting (Hutchins et al., 2011).
Survey questions were rated on a continuum from strongly disagree to strongly agree but without discrete response points. Participants drew a line to indicate the level of agreement or disagreement and responses were measured, where scores ranged from 0 cm (0% true) to 20.0 cm (100% true).
The Theory of Mind Task Battery
The Theory of Mind Task Battery (ToMB) was administered to each adolescent and is a direct measure of ToM understanding. ToMB is presented in storybook format with illustrations accompanying information and questions. Fifteen questions across nine tasks with embedded memory-control items were presented with total scores ranging from 0 to 15. For example, a storyboard displays a child named Alex who is holding a toy but then sets it on a chair and leaves. Another child, Sara, enters the room and puts the toy inside a cabinet and leaves. Alex returns to the room. Where does he go to get his toy? Options presented are chair, cabinet, table, and shelf.
The ToMB was normed on a sample of 55 typically developing children and 44 children with ASDs. No sensitivity information was provided in the test manual, but the authors noted in their discussion of criterion related validity “a substantial positive relationship (r = .66, p < .05) with variation in scores on the ToMI explaining approximately 44 percent of the variation in children’s scores on the ToM Task Battery” and an even strong correlation with the typically developing sample (67% shared variance; Hutchins & Prelock, 2014).
Intervention
The Teaching Theory of Mind (Ordetx, 2012) curriculum was delivered to participants once per week for 12 weeks. It implemented a cognitive–behavioral therapy approach and incorporated peer models and video self-assessment. Sessions were interactive and included activities to learn and practice session skills. Goals and homework that related to session content were assigned each week. Sessions focused on developing social understanding and foundational skills including joint attention, imitation, identification, reading cues, cause and effect, repair strategies, checking in, social rules, figurative language, and self-assessment. Individual and group management techniques were crucial to facilitate the effective sessions for adolescents with ASD. These included visual schedules, selective peer matching or seating, supplemental assistant, fidget manipulatives, participation and break opportunities, and enforcement of session rules.
Results
Paired t tests and corresponding Wilcoxon signed-rank tests were conducted to compare preintervention ToMB scores and ToMI domain means with postintervention values (see Table 2). The parametric analyses revealed a statistically significant increase on the ToMI from pre- to postintervention in the domains of early and advanced ToM level, emotion recognition, and pragmatics (see Figure 1). Nonparametric analyses indicated statistically significant increases in these domains of the ToMI as well as the remaining ToMI domains of basic ToM level and mental state term comprehension. Inspection of histograms as well as skewness and kurtosis statistics revealed differences in scores from pretest to posttest to be roughly normally distributed for all measures except basic ToM and mental state term comprehension. Consistent with the differences in p values between parametric and nonparametric results, difference scores for mental state term comprehension demonstrated by far the most extreme kurtosis (5.10) and by far the most extreme skewness (–1.88). Difference scores for basic ToM demonstrated the second most extreme kurtosis (1.88). All remaining kurtosis values ranged from –1.04 to 1.16. All remaining skewness values ranged from –0.56 to 0.62.
ToMI and ToMB pre- and postintervention t tests.
Note: Nonparametric ps are from Wilcoxon signed-rank test. Cohen’s drm is Cohen’s effect size for repeated measures. ToMI: The Theory of Mind Inventory; ToMB: The Theory of Mind Task Battery; SD: standard deviation; CI: confidence interval.
ToMI and ToMB pre- and postintervention outcomes. *p < .05. **p < .01. ToMI: The Theory of Mind Inventory; ToMB: The Theory of Mind Task Battery; ToM: Theory of Mind.
Multiple regression analyses were conducted to investigate the effects of absences on outcomes while controlling for initial scores. Specifically, a separate regression model was estimated for each of the eight measures of the study. Posttest score was regressed on pretest score and number of absences. While absences uniformly had a negative net effect across all domains on the ToMI and for the ToMB, this effect was only statistically significant at the conventional alpha level of .05 for the mental state term comprehension domain of the ToMI. Given the limitations discussed later, these results should be regarded as tentative findings from this exploratory study.
Discussion
Parents of adolescents with ASDs who underwent the Teaching Theory of Mind (Ordetx, 2012) curriculum showed significant increases in their perception of their adolescents’ social understanding. Scores increased from pretest to posttest across all domains of the ToMI. Each change was statistically significant at the conventional alpha level of .05 according to the Wilcoxon signed-rank test. Specifically, adolescents improved in the areas of early (reading affect and sharing attention) and advanced ToM (employing complex recursion, understanding the mind as active interpreter, and utilizing social judgment), emotion recognition, and pragmatics (social language). No significant improvement was perceived in basic ToM development (employ metarepresentation and developmentally related understandings). The improvement in early and advanced ToM development is consistent with the previous literature, suggesting that sequential ToM development is often atypically developed in individuals with ASD and improvements in the early ToM skills may be related to explicit instruction and intervention effects on deficient early ToM skills. Furthermore, as early skills improve in reading affect and sharing attention, latter more complex skills would have the prerequisite foundation to also improve. For example, if the individual has greater mastery of reading affect, he or she may have the ability to more accurately utilize social judgment.
Basic ToM was the only skill parents did not perceive to show improvement across pre- and postintervention. This level of skill consists of the utilization of metarepresentation and developmentally related understanding and may not have improved during intervention at the same rate as the early and advanced skills for various reasons. Possible reasons include a failure of the intervention to comprehensively address this skill, atypical skill acquisition in individuals with ASD (e.g., basic skills were previously mastered before early skills), or due to a differential ability to explicitly teach these abstract skills compared to the early or advanced ToM concepts. It may be the case that metarepresentational thinking is a significantly more difficult concept to teach through explicit instruction than the higher order ToM skills.
While parents perceived the improvement across the areas of ToM development and social skills, improvement in ToM development as directly measured was not statistically significant at the alpha level of .05. The discrepancy between parent perception and direct measurement of ToM change pre- and postintervention may be the result of a failure of the intervention to generalize in an evaluative setting, the result of a placebo effect on parent perception of treatment effect, or a result of a sensitivity discrepancy in ToM measures.
Several limitations were identified including the lack of a control group, convenience sampling, small sample size, and unverified diagnosis of ASD in participants. Formal verification of diagnosis was not required by participants. Participants’ parents verbally confirmed a previous professional diagnosis of ASD and investigators (licensed professionals with years of clinical experience in autism assessment and intervention) used clinical judgment to informally verify the presence of ASD symptoms. In addition, while the examiners sought to adhere to strict intervention guidelines, minor deviations were made to best accommodate participant needs (e.g., additional teaching prompts, allowance of fidgets as needed, etc.). Furthermore, it should be noted that the ToMI and ToMB measures employed differential measurement sensitivity and scaling systems. Therefore, more robust results may be achieved through the development and use of comparably sensitive ToM measurement systems.
Further research should be conducted to investigate the incongruence between parent perception and adolescent performance through methodological improvements such as the inclusion of a control group and additional more sensitive direct ToM measures. Furthermore, investigation of intervention effect across various ToM development levels (early, basic, and advanced) and developmental concepts (e.g., metarepresentation, social judgment, etc.) may allow insight into specific skills that should receive more emphasis or differential teaching strategies in the context of intervention. Historically, it has been a challenge to develop interventions with measurable generalizability effects, and this study was unable to generalize ToM skills across research and home settings. Further research should be conducted to determine what intervention strategies may help to facilitate the generalized efficacy across settings.
