Abstract:
This thesis explores the cognitive abilities of the New Caledonian crow, the only corvid species to habitually use and manufacture tools in the wild. I begin by examining whether the crows have sufficiently flexible cognition to produce metatool use (the use of one tool to obtain another). I then examine whether this species has the ability to reason causally by presenting the birds with complex physical problems. Next, I test whether the crows can use abstract, relational knowledge during problem solving. I then investigate whether the crows use imagination to solve string pulling problems. Finally, I examine whether the crows can solve a more complex 3-stage metatool task. Together, my results show that New Caledonian crows can solve problems in a manner that cannot be explained by operant or classical conditioning. I propose that their performances were generated through the use of relational rules and causal reasoning. These cognitive abilities do not fit easily into traditional dichotomous accounts of animal cognition. I therefore outline a system for categorising problem solving based on a modified version of Dennett's tower of generate-and-test (Dennett 1996). This framework should be useful for future studies of animal cognition, particularly when it comes to understanding whether tool manufacture leads to the evolution of complex cognition and whether cognition evolves in a modular fashion.
