Training Children’s Theory of Mind in Middle Childhood: The Role of Conversations

• In Event: Tell me, What’s on Your Mind? An Exploration of Mental-State Talk Discussions

Abstract

Longitudinal (Devine & Hughes, 2019) and training (Hoffman et al., 2016) studies show that exposure to stories and conversations containing mental states promotes children’s theory of mind (ToM). While illuminating, these studies were designed in such a way that it was not possible to disentangle the role played by taking part in mental-state conversations from that of being exposed to ToM material. We examined this issue in primary-school children using a conversation-based training program (Lecce et al., 2014).

We recruited 110 children (Mage 9.5 years, SD = 4.70 months) and assigned them to four training conditions: ToM WITH conversations (N = 29, 13 boys), ToM with NO conversations (N = 30, 14 boys), Control WITH conversations (N = 27, 15 boys), Control with NO conversations (N = 24, 13 boys). Each condition was made up of 4 training sessions during which children were given two narratives and were required to answer short questions. In the ToM conditions, the narratives and the questions were about mental states. In the control conditions, the narrative and questions were about physical states. In the ‘WITH conversations’ conditions, children were involved in group conversations. In the ‘NO conversations’ conditions, children were given the correct answers to the questions but were not involved in group conversations. Children’s socio-economic status, verbal ability, executive functions and reading comprehension were evaluated at pre-test as control variables. ToM was tested using the Strange Stories (White et al., 2009) and the Triangle task (Castelli et al., 2000) at pre- and post-test.

A mixed ANCOVA on the Strange Stories (Fig.1) showed a significant time by condition interaction, F(3, 101) = 20.99, p < .001, partial η2 = .38. Pairwise contrasts showed that, at post-test: a) children in the ToM WITH conversations condition significantly outperformed the Control WITH conversations condition, p < .001, d = 1.25, 95% CI [1.51, 4.54] and the Control with NO conversations condition, p < .001, d = 1.45, 95% CI [2.34, 5.45]; b) children in the ToM with NO conversations condition significantly outperformed the two control conditions, p = .042, d = 0.60, 95% CI [0.03, 3.04], and, p < .001, d = 0.88, 95% CI [0.85, 3.96 ]. Crucially, the ToM WITH conversations condition children performed significantly better than the ToM NO conversations condition children at post-test, p = .044, d = 0.65, 95% CI [0.24, 2.95]. A significant time by condition interaction, F(3, 101) = 3.75, p = .013, partial η2 = .10 was found for the Triangle task (Fig. 2). However, the differences between conditions were more nuanced. Whereas at post-test the ToM WITH conversations group outperformed both control groups, p = .043, d = 0.72, 95% CI [0.03, 2.95], and, p = .049, d = 0.70 , 95% CI [0.01, 3.04], the ToM with NO conversations group did not differ either from control conditions, ps = 1.0, or from the ToM WITH conversations condition, p = .741.

Results indicate the causal role of mental-state conversations in children’s ToM and are discussed in the light of the social-constructivist account of ToM (Carpendale & Lewis, 2006).

Authors

• Federica Bianco, University of Bergamo
  Presenting Author

• Serena Lecce, University of Pavia
  Non-Presenting Author