Improving Theory of Mind Final
Improving Theory of Mind Final
Improving Theory of Mind Final
Annick COMBLAIN*
Coraline SCHMETZ
_________________________________________________________________________________
Abstract
Effective communication requires an understanding of the interlocutor’s perspective. Being able to infer
someone else’s knowledge about a situation is a critical skill in any communication and social
interaction. These abilities are part of Theory of Mind (ToM) skills and are known to be impaired in
Down Syndrome (DS). It therefore makes sense to investigate ToM development in this population. In
our pilot study, we explore the possibility of improving ToM abilities in participants with DS and typically
developing children (TD) matched for nonverbal mental age. Participants were assessed with the
French adaptation of the “ToM Inventory” before and after a 10-week training session. Results show
that trained groups perform significantly better on ToM tasks than untrained groups, whose
performances remain stable between pre- and post-test. These results are encouraging as they suggest
that, with a specific training, children with DS can improve their ToM skills.
Keywords: Down Syndrome, Theory of Mind; Theory of Mind Training, Social cognition
INTRODUCTION
Theory of Mind (ToM) includes the ability to attribute mental states (thoughts, emotions,
beliefs, desires, etc.) to others, as well as the ability to admit that others’ thoughts or feelings may differ
from ours (Losh, Martin, Klusek, Hogan-Brown, & Sideris, 2012). This knowledge is essential to explain
and predict someone else’s behavior as well as to communicate efficiently and more specifically to
acquire the pragmatics of language (Losh et al., 2012). Being able to understand someone else’s
perspective and to infer his/her knowledge about a situation is a critical skill in any communication and
social interaction. The necessary abilities for the emergence of ToM skills develop between the age of 2
and 5-year-old. Generally basic ToM skills come to maturity around the age of 4 or 5-year-old in
typically developing children. ToM skills do not develop in the same way for all children. It is more
common for children with genetic or neurodevelopmental disorders to have difficulties with ToM tasks.
The first studies on pathological populations focused on autism, a neurodevelopmental disorder
characterized by a notable deficit in social interactions (Baron-Cohen, 1989; Baron-Cohen, Leslie, &
Frith, 1985). While these pioneering studies considered the ToM deficit as specific to autism,
subsequent studies found similar deficits in other neurodevelopmental disorders such as intellectual
disability (ID) (see Cobos & Castro, 2010 for a review). In ToM studies, autistic subjects are generally
the target group and children with ID are considered as a control group so that few researches have
been conducted on individuals with ID per se (Giouri, Alevriadou, & Taskiridou, 2010). However, in
addition to a significant limitation in intellectual functioning, individuals with ID are also characterized by
limitations in adaptive behavior including everyday social skills (American Association Intellectual and
Development Disabilities [AAIDD], 2010). ToM skills are at the intersection of these two areas as they
require an adequate level in both intellectual functioning and adaptive behavior. It is therefore
interesting to better understand the impact of intellectual disability, which is the most common cause of
intellectual disability (10% of all cases in the absence of any prenatal screening). Its prevalence is
estimated at 1/750-2000 live births depending on countries’ religious and sociocultural practices
(Diamandopoulos & Green, 2018). Even if DS is the most studied genetic syndrome of intellectual
disability (Touraine, de Freminville, & Sanlaville, 2011), some aspects of cognitive and linguistic
development remain poorly documented (e.g. lexical and semantic development, categorization
strategies, referential communication and more broadly social communication and social cognition) (see
Comblain & Thibaut, in press, for a review). Individuals with DS generally fail to perceive someone
else’s mental states and tend to assign their own thoughts to others. The literature on false beliefs and
emotion recognition in DS points to a deficit in these areas. Moreover, ToM abilities appear to be quite
poor in individuals with DS (Thirion-Marissiaux & Nader-Grosbois, 2008; Nader Grobois, 2011) whose
3
performance in traditional ToM tasks is lower than that of peers with non-specific ID and typically
developing children (TD) (Giaouri et al., 2010). Very few studies focus on ToM in individuals with DS,
and usually highlight a developmental trajectory qualitatively similar to that of TD children. The
difference between children with DS and TD peers tends to rather be in the developmental timing than
in ToM stages per se. In other words, the emergence of ToM would be delayed in children with DS
compared to TD children (Thirion-Marissiaux & Nader-Grosbois, 2007, 2008). According to Giaouri et al.
(2010) ToM deficits in DS can be partially explained by underlying deficits in skills such as language,
executive functions, short-term memory, inhibition, metacognition and attention.
Since the ability to understand someone else’s perspective and knowledge is essential to any
communication and social interaction, the question of the effectiveness of a specific training in
pathological populations is crucial. Few studies address this question and most of these concern TD
children (Melot & Angeard, 2003; Westra & Carruthers, 2017). Regarding interventions in DS we only
noted a case study of a 24-year-old adult (Montoya-Rodriguez, McHugh, Cobos, 2017) focusing on
words expressing deictic relations (“I”, “you”, “he”, “there”, “now” and “then”). Studies on ToM training in
TD children are more frequent. We will briefly discus three of them. In 2003, Melot and Angeard
conducted a study on 111 TD children aged 3;6 to 4;4 years old focusing on false-belief prerequisites.
They showed that training these prerequisites through appearance-reality and false-belief activities
improves performance in false-belief tasks. Moreover, it seems that, on the one hand, performance is
specific to training and, on the other hand, remains stable over time. More recently, Westra and
Carruthers (2017) conducted a study on 75 TD children aged 3 to 5 years old. Various tasks involving
the training of desire, beliefs, hidden emotions, knowledge and false beliefs were presented to children.
The impact of training on belief and false-belief tasks was different: it was greater regarding beliefs than
on false beliefs, the latter requiring pragmatic skills not yet developed in such young children. The
authors suggested that a specific training lasting 6 to 12 months is necessary to observe an
improvement in false-belief tasks. Moreover, mental states understanding, questions to be asked to
elicit a specific situation as well as distinction between relevant and irrelevant cues can also be
improved by training. These improvements probably reflect the child’s new ability to interpret mental
states (Westra & Carruthers, 2017). However, no generalization of the effects of training on other skills
was observed. In 2016, Gombert, Bernat and Roussey studied the progress of 65 TD children aged 5,
trained with a multimodal method using visual, audio-visual, kinesthetic, olfactory material as well as
books in a shared reading activity. Gombert et al.’s (2016) results highlight a significant improvement in
ToM abilities in the trained group compared to the control group. In addition, the authors stress that
using books helps to stimulate imagination and thus to train the ability to infer the knowledge, feelings
4
and emotions of the characters. This procedure promotes an active participation of the child that could
be worth developing among children with Down syndrome.
METHOD
Participants
Twenty participants took part in the study: 10 participants with DS (4 females and 6 males, 8.5
to 18.3-year-old, mean 11.5) and 10 TD children (3 females and 7 males, 3.11 to 4.8-year-old, mean
4.3) matched for nonverbal mental age measured with the Raven’s “Coloured Progressive Matrix”
(Raven, 1998). The French version of the Peabody Picture Vocabulary Test (PPVT) (EVIP, Dunn, Dunn
& Theriaults-Whale, 1993) was also administered to all the participants. The statistical analyses were
performed with Statistical Package for the Social Sciences (SPSS, 2017). Results are significant at p <
0.05. Given our very small sample size, nonparametric statistics are used. DS and TD groups
characteristics are summarized in Table1 (ages are expressed in months).
Table 1.
Characteristics of the DS and TD Groups
Down Syndrome (DS) Typically Developing Children (TD)
N 10 10
Mean NV-MA 55.20 (SD: 5.33) 57.30 (SD: 9.10)
Range: 48 – 63 Range: 48 – 75
Mean LexA 53.00 (SD: 13.77) 52.65 (SD: 9.26)
Range: 30 – 75 Range: 35 – 65
NV-MA = nonverbal mental age, LexA = lexical (age in months)
A Mann-Whitney analysis for independent samples shows that groups were equivalent in nonverbal
mental age (U = 45.50, p = 0.74) and lexical age (U = 46.50, p = 0.80). Half of the DS and TD
5
participants were randomly assigned to an experimental group and the other half to a control group
(Table 2).
Table 2.
Characteristics of the Experimental and the Control groups
Down Syndrome Typically Developing Children
Experimental Group Control Group Experimental Group Control Group
(DS-Experimental) (DS-Control) (TD-Experimental) (TD-Control)
N 5 5 5 5
Mean NV-MA 55.20 (SD: 6.22) 55.20 (SD: 5.02) 60 (SD: 11.02) 54.6 (SD: 6.84)
Range: 48 – 63 Range: 48 – 60 Range: 48 – 75 Range: 48 – 63
Mean LexA 40.40 (SD: 14.12) 50.20 (SD: 13.20) 47.40 (SD: 11.64) 41.40 (SD: 6.30)
Range: 21-61 Range: 29-69 Range: 26-58 Range: 30-46
NV-MA = nonverbal mental age, LexA = lexical (age in months)
To control that nonverbal mental age and lexical age remained equivalent once divided into subgroups,
we used a Kruskal-Wallis analysis [respectively for NV-MA and LexA: H(3) = 0.84, p = 0.84 and H(3) =
3.65, p = 0.30].
The participants assigned to the experimental group received a 10-week ToM training while the control
participants did not receive any specific training. All the participants were presented twice with ToM
tasks: before and after the training sessions (Table 2).
Ethics approval
This study was conducted with the approval of the Ethics Committee of the Faculty of Psychology,
Logopedics and Educational Sciences of the University of Liège (Liège, Belgium). Each participant was
tested by a speech and language pathology student under the supervision of a qualified professional. All
the children who took part in the study following had the written consent of their parents. An adapted
consent form (with pictograms) was also proposed to all the participants. All the data was anonymized
and kept confidential.
Experimental tasks
The experimental protocol was presented twice to all the participants: at the beginning and at
the end of the study (immediately after the training process in the experimental group - see
experimental design below). All the participants were assessed with the “ToM Inventory” designed by
Hutchins, Prelock and Bonazinga-Bouyea (2014) and adapted in French by Nader-Grosbois and
Houssa (2016). We chose the “Theory of Mind Inventory” because of the inadequacy of traditional ToM
6
tasks for children with DS. Indeed, one limitation of the traditional ToM tasks is that children's cognitive
and language levels can influence performance; therefore, it would be difficult to use them with
intellectually impaired individuals such as children with DS (Charman, Campbell & Edwards, 1998). With
the “Theory of Mind inventory”, it is possible to overcome the difficulties identified in the traditional tasks.
It evaluates a wide range of theory of mind competencies and does not suffer from ceiling effects when
administered to individuals with linguistic, cognitive, and motivational deficiencies (Hutchins et al.,
2014). The protocol was divided into 9 subtests for a total of 15 points. The tasks assess the
understanding of emotions, the perception of someone else’s desire or emotion as well as beliefs and
false beliefs (see Appendix 1 for details).
Training process
Training sessions were conducted on an individual basis once a week over a 10-week period
(Figure 1). We focused on identification of emotions and false-belief prerequisites. Five activities were
proposed: (1) emotion assignment, (2) shared reading, (3) symbolic game (acting out), (4) referential
communication tasks (as designed by Glucksberg, Kauss, & Weisberg, 1966) and (5) emotion
assignment (see Appendix 2 for details).
Training process
Week 1 to week 6 Weeks 7 and 8 Weeks 9 and 10
Results
Due to the small sample size, we used nonparametric statistical tests. The signification level is
p < 0.5. Group data are summarized in Table 3 and Figure 2.
7
Table 3.
Performance of Experimental and Control Groups on ToM Pre- and Post-test.
Pretest Post-test
Mean Standard Standard Median Mean Standard Standard Median
Groups N Error Deviation Error Deviation
DS- 5 4.20 1.02 2.28 4.00 9.00 0.55 1.22 9.00
Experimental
DS-Control 5 3.60 0.51 1.14 4.00 3.20 0.58 1.30 3.00
A Kruskal-Wallis analysis was performed to compare the performance of the experimental and
control groups (N = 20) at pre-test and post-test. The results point to a group effect both at pre-test
[median: 5.00; c2(3) = 10.3, p = 0.02] and post-test [median: 13.6; c2(3) = 13.6, p = 0.004]. Additional
Mann-Whitney U tests for independent samples show that the DS control group has significantly lower
ToM scores than TD groups both at pretest and post-test (respectively for the pre-test and the post-test
when compared with the experimental TD group: U = 2.00, p = 0.03 and U = 0.00, p = 0.009; when
compared with the control TD group: U = 0.50, p = 0.01 and U = 1.00, p = 0.14). There is no significant
difference between the DS experimental group and TD groups at the pre-test (respectively when
compared with the TD experimental group and the TD control group: U = 5.0, p = 0.15 and U = 4.00, p =
0.09). This lack of difference between the DS experimental group and the two TD groups in pre-test is
surprising. We expected, as observed in the DS control group, lower performance in the DS
experimental group. This is probably due to the uncontrolled intrinsic characteristic of the experimental
group. However, if we perform a Mann-Whitney analysis to compare the performance of all DS
participants with those of typically developing children, we find a significant difference between the two
groups (U = 11.50, p = 0.02) showing that DS participants’ ToM scores are lower than those of TD
children.
8
Results summarized in Figure 2 suggest that both experimental groups benefit more from the
intervention than the control groups whose performance seems to remain stable. To test this hypothesis
and the effect of training in each group, we conducted independent Wilcoxon signed-rank tests for
linked samples. Pre- and post-test results were compared for each group. Analyses confirm that both
experimental groups have significantly higher ToM scores at post-test than at pretest (DS experimental
group: Z = -2.03, p = 0.04 and TD experimental group: Z = -2.06, p = 0.39) while the control groups
have the same pre-test and pot-test ToM scores (DS control group: Z = -0.82, p = 0.41 and TD control
group: Z = -0.82, p = 0.41).
Discussion
Before discussing our results, it is important to remember that this study is a pilot study
conducted on a very small sample. Therefore, while the results are encouraging, they must be
interpreted with caution. The generalization of our conclusions to all the DS population can only be
considered after having replicated our results on a larger sample. Moreover, we cannot exclude that the
progress made in the post-test is the consequence of the Hawthorne effect. In other words, progress
could, at least partially, be attributed to the fact that participants are conscious of participating in an
9
experiment in which they are observed. This awareness of the testing situation can result in greater
motivation for the task itself. So, to further strengthen the results, it would also be interesting to add to
the initial design a group of DS participants and a group of TD children receiving a substitute
intervention.
Our first objective was to compare the overall “Theory of Mind Inventory” scores of
participants with Down syndrome with those of typically developing peers matched for NVMA. According
to the delay hypothesis (Thirion-Marissiaux & Nader-Grosbois, 2008), we expected participants with DS
to perform less well than TD children matched for NVMA. If we consider the DS groups together, our
results confirm this hypothesis. DS participants’ ToM skills are inferior at pre-test to those of typically
developing children matched for nonverbal mental age.
As predicted by our second hypothesis, the two experimental groups (trained DS and TD
participants) improved their ToM performances at the end of the training. Our results are therefore in
line with those obtained by Gombert et al. (2016). These authors showed a significant improvement in
ToM abilities in a group of TD children who had undergone training including shared reading and
activities on mental state recognition and emotion assignment.
A qualitative analysis of DS participants’ performance is also interesting. In accordance with
the developmental trajectory described by Flavell (1999), our subjects seem to understand emotions
better than beliefs and to understand beliefs better than false beliefs. This is not surprising as false
beliefs require the ability to understand someone else’s behavior, which remains a complex task for
participants with DS (Wellman, Cross, & Watson, 2001). Furthermore, false beliefs require pragmatic
skills, which develop after the age of 5 in TD children (Westra & Carruthers, 2017). However, it should
be noted that two of our subjects, the ones with a higher NVMA (63 months at the beginning of the
study), reached the false-belief comprehension level at the end of the training. Once again, further
research with more subjects is necessary to confirm the possibility of high-functioning individuals with
DS (whose mental age is higher than 5 years old) of reaching a false-belief comprehension level once
they have been trained. Lastly, it seems that DS participants who achieve the best result in ToM (hence,
false-belief skills), are also those who perform the best in language tests and who demonstrate good
language skills [respective Spearman’s rho with lexical age, pretest: r = - 0.46 (p = 0.43) and post-test: r
= 0.95 (p = 0.01)]. These results are consistent with those of de Villiers (2005) and Harris (2005). These
authors showed that good language proficiency is an advantage to develop ToM and more specifically
false-belief comprehension. Indeed, our DS participants with limited language skills had greater difficulty
in understanding stories as well as instructions. We observed that their levels of interpretation of the
surrounding world seemed to be quite limited. In summary, good language skills seem to benefit the
most from ToM training.
10
Conclusion
The main goal of this study was to determine the possibility of improving ToM skills in
individuals with Down syndrome. We chose to adapt a multimodal material that showed its effectiveness
among TD children (Gombert et al., 2016). At the end of a 10-week training session, trained participants
with DS performed better on ToM tasks than their non-trained peers. However, we must keep in mind
that this is simply a pilot study. As our experimental sample was quite small and the training quite short
(only 10 weeks), we cannot generalize our results to the whole DS population. It would then be
appropriate to reproduce this design with a larger sample of participants. It would therefore be
appropriate to conduct a longer training and to explore the long-term maintenance of the training effects.
Finally, it would also be interesting to work separately on emotions, beliefs and false beliefs. The
chosen tests would then target a single ToM aspect and give more precise results. Nevertheless,
despite the different suggestions for improving the experimental design, our results seem to highlight
the possibility of some improvement of ToM abilities among DS people.
Acknowledgements
We would like to thank APEM-T21(Down Syndrome parents’ association) for their kind participation in
this study. We would also like to thank David Magis and Jessica Simon (Laboratory of Statistics, Faculty
of Psychology - University of Liege) for their help and their statistical advice.
11
References
Abbeduto, L., Warren, S. &, Conners, F. (2007). Language development in down syndrome: from the
prelinguistic period to the acquisition of literacy. Mental Retardation and Developmental Disabilities,
13(1), 247-26. https://doi.org/10.1002/mrdd.20158
American Association Intellectual and Development Disabilities. (2010). Intellectual disability: Definition,
classification, and systems of supports (11h Edition). Silver Spring: AAIDD.
Baron-Cohen, S. (1989). The autistic child’s theory of mind. A case of specific developmental delay.
Journal of child Psychology and Psychiatry, 30, 285-297. https://doi.org/10.1111/j.1469-
7610.1989.tb00241.x
Baron-Cohen, S. &, Leslie, A.M., Frith, U. (1985). Does the autistic child have a theory of mind?
Cognition, 21, 37-46. Retrieved from
https://www.academia.edu/31528693/Does_the_autistic_child_have_a_theory_of_mind_
Charman, T., Campbell, A., & Edwards, L. (1998). Theory of mind performance in children, adolescents,
and adults with a mental handicap. Cognitive Development, 13, 307-322.
https://doi.org/10.1016/S0885-2014(98)90013-2
Cobos, F.J., & Castro, M.C. (2010). Theory of Mind in young people with Down’s syndrome.
International Journal of Psychology and Psychological Therapy, 10(3), 363-385. Retrieved from
http://www.redalyc.org/articulo.oa?id=56017068002
Comblain, A. & Thibaut, J.P. (in press). Approche neuropsychologique du syndrome de Down. In, S.
Majerus, M. Van der Linden & M. Poncelet (Eds.), Traité de neurospychologie de l'enfant. Bruxelles :
de Boeck
de Villiers, J.G. (2005). Can language acquisition give children a point of view? In J. Astington and J.
Baird (Eds). Why Language Matters for Theory of Mind (pp.186-219). New York: Oxford University
Press.
Diamandopoulos, K., & Green, J. (2018). Down syndrome: An integrative review. Journal of Neonatal
Nursing, 24(5), 235-241. https://doi.org/10.1016/j.jnn.2018.01.001
Dunn, L., Dunn, L., & Theriaults-Whalen, C. (1993). Echelle de vocabulaire en images Peabody.
Adaptation française du Peabody Picture Vocabulary Test [Questionnaire & Manual]. Toronto:
Psycan.
Flavell, J. (1999). Children’s knowledge about the mind. Annual Review of Psychology, 50, 21-45.
https://doi.org/10.1146/annurev.psych.50.1.21
Gioaouri, S., Alevriadou, A., & Tsakiridou, E. (2010). Theory of mind abilities in children with Down
syndrome and non-specific intellectual disabilities: An empirical study with some educational
12
SPSS (2017). Statistical Package for the Social Sciences (version 25). France: IBM Corporation.
Thirion-Marissiaux, A.F, & Nader-Grosbois, N. (2007). Theory of mind “beliefs”, developmental
characteristics and social understanding in children and adolescents with intellectual disabilities.
Research in Developmental Disabilities, 29, 547-566. https://doi.org/10.1016/j.ridd.2007.09.004
Thirion-Marissiaux, A.F., & Nader-Grosbois, N. (2008). Theory of mind and socioaffective abilities in
disabled children and adolescents. European Journal of Disability, 2, 133-155.
https://doi.org/10.1016/j.alter.2008.02.003
Touraine, R., de Fréminville, B., & Sanlaville, D. (2011). La Trisomie 21. Collège National des
Enseignants et Praticiens de Génétique Médicale. Université Médicale Virtuelle Francophone.
Retrieved from https://docplayer.fr/14750541-Support-de-cours-version-pdf-la-trisomie-21-college-
national-des-enseignants-et-praticiens-de-genetique-medicale.html
Wellman, H., Cross, D., & Watson, J. (2001). Meta-analysis of theory-of-mind development: The truth
about false belief. Child development, 72, 665-684. https://doi.org/10.1111/1467-8624.00304
Westra, E., & Carruthers, P. (2017). Pragmatic development explains the theory-of-mind scale.
Cognition, 158, 165-176. https://doi.org/10.1016/j.cognition.2016.10.021
14