The growth of comparative cognition has led to the discovery of surprisingly complex behaviour in taxa as different as social insects (e.g. bumblebees), birds (e.g. corvids and parrots), and non-human primates. Accounts based on mental representations, consciousness, or insight are always tempting, but are frequently empirical dead-ends. An alternative research program is to follow experimentally accessible paths combining prior predispositions, reinforcement and concept learning, procedural and episodic memory, and other psychological constructs with the rapidly growing field of synthetic intelligence. I will present findings on complex problem-solving by crows and parrots and our attempts to understand and emulate their performance using virtual and physical synthetic systems.