Tool use is primarily, but not exclusively, present in species with otherwise advanced cognitive traits. However, the interaction between such traits and conspecific inter-individual variation in the presence, complexity, or intensity of tool use is far from being established. We addressed this matter among human infants, seeking factors that relate to differences in tool use. We examined, both correlationally and experimentally, whether the propensity to engage in object combinations predicts performance in means–end problem-solving tasks involving or not involving the use of a tool. We tested 71 infants aged 15, 18, 21, and 24 months, dividing them into two subgroups: one exposed to an adult demonstrating object–object combinations (i.e., “prompting” infants to combine objects together) and another with comparable social exposure but where the adult demonstrated single-object manipulations. We found a correlation between the combined level of spontaneous and prompted object combinations and problem-solving performance regardless of the involvement of tools in the problem. However, we did not find differences in tool-use performance between the two demonstration subgroups. The correlational analysis suggests that complexity of play, as measured by the frequency of combining objects, is linked to infants’ problem-solving skills rather than being specifically associated with tool use, as previously suggested in the literature.