I like the parts of the study where they compare the pigeons to human radiologists:
In a study of human radiologists and radiology residents given the same cases to review, accuracy averaged 97% for calc-(negative) images, but only 70% for calc-(positive) images; our birds’ performance did not reliably differ between the categories. In fairness, unlike the pigeons, the radiologists were not shown just small regions of interest, but whole images that had to be searched for abnormalities, a more challenging task.
and
Thus, the pigeons’ errors did not appear to be random; instead, the least-accurately classified images contained features that were relatively unrepresentative of the rest of the corresponding benign or malignant set, emphasizing the importance of ensuring that images used for training fairly represent the gamut likely to be encountered in future specimens. This, of course, is true for pathology residents’ training as well.
and
Overall, our results suggest that pigeons can be used as suitable surrogates for human observers in certain medical image perception studies, thus avoiding the need to recruit, pay, and retain clinicians as subjects for relatively mundane tasks.
Also, here’s a picture of the pigeons’ working conditions!
(The link to the study is in the article, and the contents of this post including the picture were copied from the study.)
EY FATTY BOOM BOOM 