Comparing humans and nonhuman great apes in the broken cloth problem: Is their knowledge causal or perceptual?

•Chimpanzees, orangutans, and young children solved the broken cloth problem.•Three- and 4-year-old children outperformed 2-year-olds and the nonhuman primates.•All subjects attended to the functionally-relevant features of this causal task.•Only 4-year-old children attended to the equivalent features of a noncausal task. When presented with the broken cloth problem, both human children and nonhuman great apes prefer to pull a continuous cloth over a discontinuous cloth in order to obtain a desired object resting on top. This has been interpreted as evidence that they preferentially attend to the functionally relevant cues of the task (e.g., presence or absence of a gap along the cloth). However, there is controversy regarding whether great apes’ behavior is underpinned by causal knowledge, involving abstract concepts (e.g., support, connection), or by perceptual knowledge, based on percepts (e.g., contact, continuity). We presented chimpanzees, orangutans, and 2-, 3-, and 4-year-old children with two versions of the broken cloth problem. The Real condition, made with paper strips, could be solved based on either perceptual cues or causal knowledge. The Painted condition, which looked very similar, could be solved only by attending to perceptual cues. All groups mastered the Real condition, in line with previous results. Older children (3- and 4-year-olds) performed significantly better in this condition than all other groups, but the performance of apes and children did not differ sharply, with 2-year-olds and apes obtaining similar results. In contrast, only 4-year-olds solved the Painted condition. We propose causal knowledge to explain the general good performance of apes and humans in the Real condition compared with the Painted condition. In addition, we suggest that symbolic knowledge might account for 4-year-olds’ performance in the Painted condition. Our findings add to the growing literature supporting the idea that learning from arbitrary cues is not a good explanation for the performance of apes and humans on some kinds of physical task.