Pigeons can detect cancer in medical images with 85% accuracy — the same as trained human radiologists
A 2015 study at UC Davis trained pigeons to identify cancerous tissue in digitised slides and mammogram images. With enough reinforcement training, pigeons achieved 85% accuracy — comparable to a trained human radiologist. When flock decisions were pooled (allowing multiple pigeons to 'vote'), accuracy rose to 99%. Pigeons apparently use fine-grained texture and colour discrimination to distinguish malignant from benign tissue. This doesn't mean pigeons understand cancer — they're detecting visual patterns — but the perceptual capacity is real.
Cancer detection requires expert human medical training and is prone to error even among specialists. Finding that a pigeon with several weeks of reinforcement training can match a radiologist's performance makes human pattern-recognition expertise feel less unique and makes pattern recognition feel less cognitive.
“Trained pigeons can detect cancer in medical images with 85% accuracy — the same as radiologists. When multiple pigeons vote together, accuracy hits 99%. They detect visual patterns, not the disease. 🐦 #OddlyHuman”