Social interaction has been shown to be a vehicle for cognitive development in young children. It encourages children to communicate and coordinate their actions, leading to the joint construction of knowledge (Isohätälä et al., 2017). However, how interaction supports joint knowledge construction has not yet been fully understood. Research on the impact of children’s social interaction mostly focuses on whether a learning effect occurs, typically following the sequence of pretest-interaction episode-posttest. The performance change between pre- and posttest is known to be influenced mainly by the initial cognitive ability, task type, gender, and sociability or friendship between children composing the interacting dyad. However, findings diverge regarding children’s cognitive ability. Some studies report that equal-ability dyad lead to progress (Cannella, 1992, 1993), while others emphasize the importance of mixed-ability dyad in organizing children’s collaboration for optimal learning outcomes (Garton & Pratt, 2001; Psaltis, 2011; Tudge & Winterhoff, 1993). Furthermore, some studies have questioned whether expert children in the dyad benefit from the interaction, since positive learning effects were mostly found in novice children (Fawcett & Garton, 2005; Tudge, 1989). A related constraint appears in the simplicity of the tasks used. Since the core of learning is the acquisition of skills or knowledge, most studies present tasks that target a specific type of domain-general cognitive function, often grounded in Piagetian perspectives, such as sequencing, rule understanding, flexibility, perspective-taking, or understanding of conservation. While these representative cognitive tasks are meaningful for measuring a child’s level of cognitive development, they may not be entirely suitable for studies on children’s social interaction, as such tasks can lack sufficient challenge for children with more advanced abilities. If collaboration is assumed to involve rich social interactions rather than mere knowledge transfer, the tasks must also be designed adequately to induce these interactions. In this regard, the complex problems defined by Funke can be considered as appropriate for collaborative problem-solving. Funke (2012) characterizes complex problems as novel, opaque, evolving over time, and intricate. In other words, they are new problems that children have not experienced before, whose goals and solutions are not clearly presented, and whose solving process involves multiple stages.

The current study aims to explore the dynamics of social interactions occurring during young children’s mathematical problem-solving by comparing performance and behavioral indicators under both individual and collaborative conditions. Two primary questions guided this study. The first was whether children who worked in pairs showed a better performance than children who worked alone. The second concerned the source of any potential performance difference, and more specifically whether it was rooted in children’s initial cognitive ability or emerged newly during interaction independently from their pre-mathematical ability levels.

Participants A total of 187 children (67 girls, Mage = 67.3 months) were recruited from ten public schools of seven regions of Luxembourg to participate in this study. Procedure The study measured two aspects: basic pre-mathematical abilities and mathematical problem-solving abilities. The first session assessed all children individually, while the second session assessed them individually or in pairs, depending on their assigned condition. All children completed the test of basic pre-mathematical abilities followed by a mathematical problem-solving test. Pre-mathematical abilities included five subtests, two for spatial ability (visual perception, mental transformation) and three for numerical ability (verbal counting, higher counting, non-symbolic comparison). Scores from these subtests were standardized and summed to create a composite pre-mathematical score. Two types of complex mathematical problems (i.e., spatial and numerical task) were developed each with five items. Spatial tasks required replicating a model using 15 colored wooden blocks, as if reflected in a mirror, awarding one point per correct item. Errors were recorded. Numerical tasks involved counting and numerical comparison using snowflake blocks. Children were asked to identify the construction containing most elements among several presented, earning one point for each correct answer. They were free to solve the problems using their preferred approach, such as dismantle and recombine blocks to compare physical size and/or count the number of blocks to compare the number of composing snowflake blocks. The used strategies were recorded. Both problems required mathematical thinking and met Funke's (2012) criteria for complex problem, novelty, unclear goals and solutions, and multi-layered cognitive processes. Data were analyzed using mixed models, binomial regression, and independent t-tests. Given the hierarchical data structure (participants within groups, groups within schools), mixed models with random intercepts for participants and fixed effects for predictors (interaction condition, pre-mathematical score, gender) were used.

Our study showed that preschool children achieved better outcomes in mathematical problem-solving tasks when working in dyads than when working alone. This pattern appeared in both spatial and numerical tasks, aligning with research on the role of collaboration in joint knowledge construction. Math score was a strong predictor across tasks and conditions, yet its interaction with math score was not significant. The three-way interaction among condition, math score, and partner’s math score was also non-significant, indicating that the partner’s math ability did not significantly moderate the effect of condition. These findings suggest that correct joint construction in pairs likely resulted from social interaction rather than by individual cognitive abilities. While higher baseline math knowledge contributes to performance, collaborative environments offer independent benefits beyond what baseline math knowledge alone can explain. In the spatial task, children in pairs had a lower error rate. Given that interactions with math score and partner’s math score were not significant, this may stem from effective reflection fostered by social interaction. In the numerical task, children in pairs attempted more strategies (1.38 per item versus 0.76 for individuals) and used them more efficiently. Consequently, pairs succeeded using more complex strategies, whereas individuals relied on more intuitive approaches. While the number of strategies used per item was maintained for children in pair over the 5 subsequent items, those who worked individually used significantly fewer strategies on the last item compared to the first. Overall, our study highlights how interaction, often overlooked in preschool settings, supports problem-solving process. Whereas most research focuses on the learning gains from collaboration, our results underscore that collaborative interaction can yield correct joint knowledge constructions beyond baseline math skills, probably mediated by enhanced execution, reflection, and sustained interest.

Cannella, G. S. (1992). Gender Composition and Conflict in Dyadic Sociocognitive Interaction : Effects on Spatial Learning in Young Children. The Journal of Experimental Education, 61(1), 29‑41. Cannella, G. S. (1993). Learning through social interaction : Shared cognitive experience, negotiation strategies, and joint concept construction for young children. Early Childhood Research Quarterly, 8(4), 427‑444. https://doi.org/10.1016/S0885-2006(05)80078-X Fawcett, L. M., & Garton, A. F. (2005). The effect of peer collaboration on children’s problem-solving ability. British Journal of Educational Psychology, 75(2), 157‑169. https://doi.org/10.1348/000709904X23411 Fraysse, J.-C. (1994). Combined effects of friendship and stage of cognitive development on interactive dynamics. Journal of Genetic Psychology, 155(2), 161. https://doi.org/10.1080/00221325.1994.9914769 Funke, J. (2012). Complex problem solving. Encyclopedia of the Sciences of Learning (682-685). Heidelberg: Springer. Garton, A. F., & Pratt, C. (2001). Peer assistance in children’s problem solving. British Journal of Developmental Psychology, 19(2), 307. https://doi.org/10.1348/026151001166092 Isohätälä, J., Järvenoja, H., & Järvelä, S. (2017). Socially shared regulation of learning and participation in social interaction in collaborative learning. International Journal of Educational Research, 81, 11‑24. https://doi.org/10.1016/j.ijer.2016.10.006 Pine, K. J., & Messer, D. J. (s. d.). GROUP COLLABORATION EFFECTS AND THE EXPLICITNESS OF C~~~DREN’S KNOWLEDGE. 18. Psaltis, C. (2011). The constructive role of gender asymmetry in social interaction : Further evidence: Constructive role of gender asymmetries. British Journal of Developmental Psychology, 29(2), 305‑312. https://doi.org/10.1111/j.2044-835X.2011.02029.x Tudge, J. (1989). When collaboration leads to regression : Some negative consequences of socio-cognitive conflict. European Journal of Social Psychology, 19(2), 123‑138. https://doi.org/10.1002/ejsp.2420190204 Tudge, J., & Winterhoff, P. (1993). Can young children benefit from collaborative problem solving? Tracing the effects of partner competence and feedback. Social Development, 2(3), 242‑259. https://doi.org/10.1111/j.1467-9507.1993.tb00016.x