Prior to SPATHS, the association between spatial skills and mathematical ability was well-established at the level of correlation, and training studies had provided initial evidence of far transfer from spatial to mathematical skills. However, the field lacked a principled, theoretically-grounded account of the mechanisms through which spatial skills contribute to mathematics development. SPATHS addresses this gap directly.
The pre-registered protocol paper provides an overview of the potential mechanisms in a theoretic framework, how the SPATHS project aims to contribute to provide more understanding about the mechanisms but also how future research can and should progress to identify specific learning processes underlying the spatial-maths association.
Substantively, the finding that no single spatial subskill is the primary driver of the spatial-maths association challenges the hypothesis that spatial visualisation is uniquely important due to its role in mentally representing mathematical problems. Instead, results support a more distributed and flexible account: children draw on different spatial skills depending on the mathematical task at hand. This has implications for intervention design, suggesting that broad exposure to a variety of spatial strategies in mathematics education is preferable to targeting one specific spatial skill.
Beyond the initial aims, the project resulted in a theoretical position statement on the concept of "underlying mechanisms", published in the Journal of Numerical Cognition. Through the project, it became clear that a conceptual confusion existed in the literature: the term "underlying mechanism" was being used inconsistently and without a shared framework, making it difficult to design studies that could meaningfully test causal claims. Rather than proceeding without addressing this, a new theoretical framework was developed and published to clarify how mechanisms could be conceptualised and interpreted in cognitive psychology. This contribution encourages researchers to identify what level of mechanism they aim to study and reframes how research exploring underlying mechanisms should include insights from across all different levels.
A further result is the creation of the CHAOS network (Cognitive Heterogeneity Across Outcomes and Studies), which emerged organically from collaborations formed during the fellowship between the host institution at KU Leuven and the secondment at the University of Oxford. What began as an informal meet-up between two research groups has grown into an active international network with 20 members across 12 institutions, holding monthly meetings. This network creates the conditions for future collaborative research across institutions and countries.
For future uptake, the most pressing need is for the completion of the UK dataset and the publication of the full empirical results, including the cross-national comparison. Beyond this, Study 2, which could not be completed due to the early termination of the fellowship, represents a well-prepared next step. The study design and ethical approval are already in place, and conducting this study would allow the learning processes to be tested in an experimental way.
