Do Children Cause the Cognitive Stimulation They Receive? Modelling the Direction of Causality

Oginni, Olakunle, and Sophie von Stumm. 2022. “Do Children Cause the Cognitive Stimulation They Receive? Modelling the Direction of Causality.” PsyArXiv. October 8. psyarxiv.com/pqf78

Abstract

We tested the directionality of associations between children’s early-life cognitive development and the cognitive stimulation that they received from their parents. Our sample included up to 15,314 children from the Twins Early Development Study (TEDS), who were born between 1994 and 1996 and assessed at age 3 and 4 years on cognitive development and cognitive stimulation, including singing rhymes, reading books, and playing games.

Across a series of genetically informative models, we found consistent, bidirectional causal influences from cognitive development at age 3 to cognitive stimulation at age 4, and from cognitive stimulation at age 3 to cognitive development at age 4. The prospective causal paths in both directions accounted for a third and up to half of the constructs’ phenotypic correlations. Our findings emphasize the active role that children play in constructing their learning experiences, and challenge the idea that children are passive recipients of environmental inputs.

Discussion

Twin studies can be used to strengthen causal inferences for observed associations between two constructs by controlling for their shared genetic and environmental influences (McAdams et al., 2021). Specifically, twin studies can address whether associations between a putatively environmental exposure and developmental differences in a phenotype remain significant after accounting for the confounding effects of shared etiology (i.e., common causes; McAdams et al., 2021). Recent years have seen an explosion of novel modelling approaches that extend the classical twin design, which enable causal inferences but are yet to be systematically applied in psychological research (Eberli et al., 2019; McAdams et al., 2021). Here, we fitted direction-of-causality (DoC) models (Heath, 1993), Mendelian Randomization (MR) extensions of the DoC model (Minică et al., 2018), and cross-lagged twin models (Burt et al., 2005) to investigate if children’s cognitive stimulation causes their cognitive development or vice versa at age 3 and 4 years.

We found prospective causal associations between children’s cognitive development and their cognitive stimulation that were bidirectional. Thus, contemporaneous mechanisms drive the causal influences between cognitive ability and stimulation: Children benefit from experiencing cognitively stimulating environments, and at the same time their cognitive development evokes the cognitive stimulation that their parents provide. Our findings align with a growing body of evidence documenting that children are not passive recipients of environmental inputs but select, modify, and create experiences that are correlated with their genetic proclivities (Bronfenbrenner & Morris, 2006; Wertz et al., 2019).

The prospective, bidirectional causal influences between cognitive development and cognitive stimulation were significant, accounting for between a third to nearly half of their phenotypic correlations across time. Thus, a substantial proportion of the phenotypic associations between cognitive development and cognitive stimulation can be attributed to bidirectional causal effects, while shared etiology or common causes – in particular, the shared environment – accounted for the remainder of the correlations. Finding that common causes and bidirectional causal effects jointly explain the associations between cognitive development and cognitive stimulation is key for developing realistic expectations for the magnitude of the effects that early-life interventions can possibly achieve (von Stumm, 2022).

We also showed that children’s differences in cognitive development and stimulation are largely due to shared environmental influences (56%-74%), followed by additive genetic influences (24%-40%). A similar pattern emerged for the etiologies of the associations between cognitive development and stimulation, which were again mainly due to shared environmental (76%-83%) and then additive genetic influences (16%-20%). Non-shared environmental influences on cognitive development and cognitive stimulation were negligible. While this study is to our knowledge the first to estimate the heritability of cognitive stimulation and its association with cognitive development, our findings confirm that all traits are heritable, including putatively environmental measures that are genetically influenced (Krapohl et al., 2017; Plomin, 1995; Polderman et al., 2015).

Limitations

Notwithstanding this study’s many strengths, including the analysis of twin and genomic data, large sample sizes, and repeated assessments of the core constructs using state-of-the-art methods, it is not without limitations. First, cognitive stimulation was assessed using seven parent-reported items, but a multi-informant approach (i.e., naturalistic home observations) with a greater number of observed variables would have improved the measures’ validity. Second, data on cognitive stimulation were only collected at the twins ages 3 and 4 years, which made it impossible to test for meaningful changes in the direction of causality over the longer course of childhood. Third, the association between the PGS for years spent in education (Lee et al., 2018) and cognitive development was small, which is indicative of weak instrument biases that can lead to overestimation of causal effects (Burgess & Thompson, 2011). In addition, cognitive development and cognitive stimulation did not have a sufficiently differentiated genetic architecture to meet the assumptions of DoC models (Health, 1993; van Bergen et al., 2018). As a result, our preregistered analysis strategy was not viable, and we fitted cross-lagged twin models instead that require longitudinal data.