Earlier theory of mind predicts later lie-telling in preschool children

Prior findings on the relation between children's theory of mind (ToM) and their lie-telling behaviors were mixed. While some studies reported that ToM ability is positively associated with children's lying, others found a null association between ToM and children's lying (Wang et al., 2017). Recent meta-analysis reports revealed statistically significant but small effects between children's ToM and lying behavior (Lee et al., 2021; Sai et al., 2021). However, the findings were mostly based on cross-sectional studies, with very little longitudinal evidence on the causal relationship between the two constructs. Two training studies (Ding et al., 2015; Ding et al., 2018) revealed that training ToM caused honest children to lie; meanwhile, training children to lie also improved their ToM, suggesting a bidirectional causal relationship. The current study provides much-needed cross-lagged longitudinal evidence on the association between ToM and lying in young children.

Adopting a three-time-point short-term longitudinal design, we tested 104 normally developing children's (64 boys, M = 54.0 months) false belief understanding and lie-telling behavior three times at roughly three months intervals. At Time 1, children participated in a hide-and-seek deception task (Ding et al., 2015), 1st order false belief understanding tasks (content false belief and location false belief), an inhibitory control task ("Simons says"), and a verbal ability task (the Bus Story). Parents also reported family socioeconomic status and only child status at Time 1. At Time 2 and 3, children again participated in a hide-and-seek deception task and two 1st order false belief understanding tasks that were slightly different from those in Time 1 in format. In addition, we added a 2nd order false belief task at Time 3. Correlation analysis (Table 1) showed that children's Time 1 verbal ability was correlated with Time 1 ToM (r = .259, p = .010). Time 1 inhibitory control was correlated with Time 1 ToM (r = .196, p = .048) and Time 3 ToM (r = .232, p = .027). Age was correlated with Time 3 ToM (r = .287, p = .006). Structural equation modelling was used to examine the longitudinal association between ToM and lying behaviors (Figure 1). Results showed the cross-lagged model fit the data well, χ2(24) = 19.692, p = .714, CFI = 1.00, TLI = 1.00, and RMSEA = 0. Lying behaviors exhibited moderate stability across the three time points (β = .513, p < .001 from T1 to T2; β = .344, p = .001 from T2 to T3). ToM exhibited moderate stability between the first two time points (β = .335, p <.001) but not between T2 and T3 (β = .111, p =.335). There was a significant cross-lagged association between Time 1 ToM and Time 2 lying (β = .311, p < .001) and between Time 2 ToM and Time 3 lying (β = .232, p =.020). There was no significant cross-lagged association between earlier lying and later ToM (ps > .05). We conclude that earlier ToM predicts later lie-telling in young children, but not vice versa. Copyright © 2024 All Academic, Inc.