Debates in the study of animal cognition of often oscillate between over-hyped claims of "human-like" cognitive abilities and deflationary "killjoy" accounts. In this talk I will analyse the inference issues involved in such debates and argue that studies of cognitive evolution should follow studies of morphological complexity and seek incremental Darwinian accounts rather than postulating miraculous cognitive leaps. I will use our work on tool manufacturing New Caledonian crows to illustrate how quite complex behaviour may have incrementally evolved.