A gorilla image the same size as a box of matches and superimposed into a chest CT scan was missed by over 80% of a group of expert radiologists asked to examine the images, a new study has shown.
Inspired by the now-famous study in which a succession of observers completely missed a gorilla strolling around in the background while they were concentrating on counting the throws of people playing with a ball, Harvard researcher Trafton Drew and his colleagues have done the equivalent experiment on radiologists, individuals who are skilled observers.
A white-outlined gorilla was superimposed into a series of CT scan images so that, were the gorilla really in the person's chest, it would occupy a lung volume equivalent in size to a matchbox. Compared to the majority of chest pathologies, it should be easy to spot.
A group of 24 radiologists were given several minutes to look through and identify white "nodules" (by clicking on them) in a set of 5 CT images; the gorilla-positive scan set was the last one presented to them.
Eighty-three percent of the radiologists failed to spot the gorilla, despite, eye-tracking analysis showed, over half of those who missed it looking directly at it, and scrolling past it an average of 4.3 times!
Amongst a non-medically-trained control group of 25 individuals who were also tested, none spotted the gorilla. But, critically, when it was pointed out, all of the study participants could see it easily.
This is an example, the team say, of inattentional blindness, a phenomenon whereby engagement with a demanding task can act like a set of blinders, making salient stimuli slip unnoticed before the eyes.
It's reassuring, the team observe in their paper in the journal Psychological Science, that the radiologists, being expert observers, did have a lower susceptibility to inattentional blindness than the controls, with 17% finding the gorilla compared with none of the controls.
However, the result is still quite dismal and highlights the fact that even a highly-trained eye can miss the blindingly obvious.
"The message of the present set of results," the team say, "is that even this high level of expertise does not immunise individuals against inherent limitations of human attention and perception. Researchers should seek better understanding of these limits, so that medical and other man-made search tasks could be designed in ways that reduce the consequences of these limitations."
This is surely a Good Thing, and Bad Science.
Nice one, RD!
I can see this result would be poor if the radiologists were actually looking for a tiny gorilla, but they weren't; they would have been looking for clinical signs of lung disease in the scans, which are surely not judged on their resemblance to tiny primates. They almost certainly examined the gorilla-shaped blob, assessed it's clinical significance, and moved on. dlorde, Thu, 22nd Aug 2013
What worries me is that nobody, not even the experimenters themselves, noticed the alien Balinese dancer where the patient's heart should be. I've read the original paper, and she appears on every image.