Apes (Gorilla gorilla, Pan paniscus, P. troglodytes, Pong abelii) and corvids (Corvus corax, C. corone) are among the most proficient and flexible tool users in the animal kingdom. Although it has been proposed that this is the result of convergent evolution, little is known about whether this is limited to behavior or also includes the underlying cognitive mechanisms. We compared several species of apes (bonobos, chimpanzees, gorillas, and orangutans) and corvids (carrion crows and common ravens) using exactly the same paradigm: a support task with elements from the classical patterned-string tasks. Corvids proved able to solve at least an easy pattern, whereas apes outperformed corvids with respect to the complexity of the patterns solved, the relative number of subjects solving each problem, and the speed to reach criterion. We addressed the question of whether subjects based their choices purely on perceptual cues or on a more abstract understanding of the problem. This was done by using a perceptually very similar but causally different condition where instead of paper strips there were strip shapes painted on a platform. Corvids' performance did not differ between conditions, whereas apes were able to solve the real but not the painted task. This shows that apes were not basing their choices just on spatial or arbitrary perceptual cues. Instead, and unlike corvids, they must have had some causal knowledge of the task.