Better all by myself: Gaining personal experience, not watching others, improves 3-year-olds’ performance in a causal trap task

•Children received different types of initial experience with a causal trap-task.•From age 3, generating one’s own experience led to above-chance performance.•If the same experience was given through demonstration, performance was at chance.•Using a tool was associated with more perseveration than using one’s hands. Children often learn from others’ demonstrations, but in the causal domain evidence acquired from observing others may be more ambiguous than evidence generated for oneself. Prior work involving tool-using tasks suggests that observational learning might not provide sufficient information about the causal relations involved, but it remains unclear whether these limitations can be mitigated by providing demonstrations using familiar manual actions rather than unfamiliar tools. We provided 2.5- to 3.5-year-old children (N = 67) with the opportunity to acquire experience with a causal trap task by hand or by tool actively or from observing others. Initially, children either generated their own experience or watched a yoked demonstration; all children then attempted the trap task with the tool. Children who generated their own experience outperformed those who watched the demonstration. Hand or tool use had no effect on performance with a tool. The implications of these findings for scaffolding self-guided learning and for demonstrations involving errors are discussed.