Previous work has shown a positive correlation between relative forebrain size and feeding innovation frequency, corrected for species number, over different taxonomic groups of birds. Several confounding variables could account for this relationship: ornithologists could notice and report innovations more often in certain taxa because of biased expectations, greater research effort, editorial bias in journals or large population sizes of the taxa. The innovationforebrain correlation could also be spuriously caused by phylogeny or juvenile development mode. We examined these possibilities by entering species number per taxon, population size, number of full length papers, expectations (assessed by a questionnaire), journal source and development mode in multiple regressions that also included relative forebrain size. We did this with and without phylogenetic corrections and tested two behavioural categories, feeding and nesting, where flexibility and learning are clearly thought to differ, but confounds should have similar effects. Through an exhaustive survey covering 30 years in 11 journals, a total of 683 innovations was gathered for the northwestern part of Europe, 507 for feeding and 176 for nesting. Species number per taxon was the only significant confound for both feeding and nesting reports; as predicted, forebrain size was a second significant predictor for feeding innovations, but not for nesting. The frequency of feeding innovations in the short notes of ornithology journals thus appears to be a valid and reliable way to operationalise behavioural flexibility in birds.
Tools are traditionally defined as objects that are used as an extension of the body and held directly in the hand or mouth. By these standards, a vulture breaking an egg by hitting it with a stone uses a tool, but a gull dropping an egg on a rock does not. This distinction between true and borderline (or proto-tool) cases has been criticized for its arbitrariness and anthropocentrism. We show here that relative size of the neostriatum and whole brain distinguish the true and borderline categories in birds using tools to obtain food or water. From two sources, the specialized literature on tools and an innovation data base gathered in the short note sections of 68 journals in 7 areas of the world, we collected 39 true (e.g. use of probes, hammers, sponges, scoops) and 86 borderline (e.g. bait fishing, battering and dropping on anvils, holding with wedges and skewers) cases of tool use in 104 species from 15 parvorders. True tool users have a larger mean residual brain size (regressed against body weight) than do users of borderline tools, confirming the distinction in the literature. In multiple regressions, residual brain size and residual size of the neostriatum (one of the areas in the avian telencephalon thought to be equivalent to the mammalian neocortex) are the best predictors of true tool use reports per taxon. Innovation rate is the best predictor of borderline tool use distribution. Despite the strong concentration of true tool use cases in Corvida and Passerida, independent constrasts suggest that common ancestry is not responsible for the association between tool use and size of the neostriatum and whole brain. Our results demonstrate that birds are more frequent tool users than usually thought and that the complex cognitive processes involved in tool use may have repeatedly co-evolved with large brains in several orders of birds.