February 28, 2004

Gorillas and umbrella women

Michael Shermer's Skeptic column in the March 2004 Scientific American discusses research by Daniel Simons and others on "inattentional blindness." These studies show that "when observers were actively engaged in an unrelated task, they sometimes failed to see ... an unexpected event, or UE". For example, the "unrelated task" might be viewing a video and counting rapid basketball passes made by a group of people wearing white t-shirts, while ignoring the passes made by people wearing black t-shirts; the "unexpected event" might be the appearance in the video of a woman carrying an umbrella or a person wearing a gorilla suit.

Some demos of the videos used in these tasks are here. The gorilla video and the umbrella video are certainly striking. A key paper is Daniel Simons and Christopher Chabris, "Gorillas in our midst: sustained inattentional blindness for dynamic events." Perception (1999), v. 28 pp. 1059-1074. From their abstract:

...we are surprisingly unaware of the details of our environment from one view to the next: we often do not detect large changes to objects and scenes (`change blindness'). Furthermore, without attention, we may not even perceive objects (`inattentional blindness'). Taken together, these findings suggest that we perceive and remember only those objects and details that receive focused attention. ... Our results suggest that the likelihood of noticing an unexpected object depends on the similarity of that object to other objects in the display and on how difficult the priming monitoring task is.

Here's part of the data from their experiment, showing the percentage of subjects noticing the unexpected event in each condition.

Easy task
Hard task
White t-shirts
Black t-shirts
White t-shirts
Black t-shirts
Umbrella woman

The videos (as you can see for yourself above) show two groups of students weaving in and out while passing basketballs around. One group is wearing white t-shirts, while the other group is wearing black t-shirts. A subject may be asked to attend to either the "team" wearing white or the team wearing black. The "easy task" is just to keep a silent mental count of the number of passes made by the monitored team; the "hard task" is to count bounce passes and aerial passes separately.

When subjects are performing no monitoring task, they always notice the umbrella woman and the gorilla (the "unexpected event" or UE). They notice the UE more often when performing the easy task than when performing the hard task. They notice the gorilla more often when they're monitoring the black team, and the umbrella woman (who is wearing pale colors) more often when they're monitoring the white team.

It's important to recognize what these results don't show -- they don't show that we ignore things unless we're looking for them. Instead, they show that when we're performing a cognitively difficult monitoring task, we may fail to notice (very salient) things that are not relevant to the task.

One obvious practical application of these results is in designing jobs and work procedures, so that (for instance) pilots monitoring instruments don't fail to notice unexpected objects on the runway. Another obvious application is in evaluating eyewitness testimony. However, Shermer's Scientific American column draws a more metaphorical moral. He suggests that we can think of science as a cognitively demanding monitoring task, whose practitioners may therefore be blind to all sorts of gorillas and umbrella women that happen to be cognitively distant from the things they're focused on. I'm sure that this is true, and will offer some examples from the linguistic sciences in later posts. However, in my experience, "inattentional blindness" in science is more complex. Everyone actually sees the metaphorical gorilla, but they've collectively decided that it's not interesting or relevant, so that it's not examined with any care, and it's ignored in framing descriptions and crafting explanations.

Shermer closes with an upbeat assessment, which I share:

... the power of science lies in open publication, which, with the rise of the Internet, is no longer constrained by the price of paper. I may be perceptually blind, but not all scientists will be, and out of this fact arises the possibility of new percepts and paradigms. There may be none so blind as those who will not see, but in science there are always those whose vision is not so constrained.


Posted by Mark Liberman at February 28, 2004 08:23 AM