Abstract: The short notes of ornithology journals feature new and unusual feeding behaviours, which, when systematically collated, could provide a quantitative estimate of behavioural flexibility in different bird groups. Previous studies suggest that taxonomic variation in the frequency of new behaviours (innovations) is correlated with variation in relative forebrain size. Recent work on primates shows, however, that observer bias can affect innovation frequency. We assess this possibility in birds via three estimates in North America and Australia: the number of full-length papers in academic journals, the frequency of photographs in birding magazines and a questionnaire on reporting bias given to ornithologists at a meeting. We also look at sampling effects due to single journal sources by doing a split-half analysis of our North American database (The Wilson Bulletin vs. six other journals) and adding three new Australian journals to the one we had used previously. In multiple regressions that also included species number per taxon, none of the potential biases could account for the correlation between forebrain size and innovation frequency. Species number was the best predictor of full-length paper frequency, which was the best predictor of photograph numbers. Ornithologists are not preferentially interested in innovative, large-brained taxa, suggesting that the correlation between innovations and neural substrate size is not a spurious effect of the biases examined here.