Attention · about 4 minutes
The Gorilla Room
One room ago something changed in front of you, repeatedly, and you couldn't find it. That was a failure of comparison -- you saw both versions and couldn't tell them apart.
This is worse. This time the thing will be in plain view, large, slow, and unhidden, and there will be nothing to compare it to. You will simply not see it.
Before we start. Small shapes move around a box for about fifteen seconds -- smooth motion, no flashing, nothing sudden. The counting task is genuinely the task; do it properly and the exhibit works. Do it badly and it also works, just for a duller reason.
Some shapes are dark and some are pale. They bounce around the box. Your job:
Count how many times a PALE shape touches an edge of the box.
Only pale ones. Only edges. It goes quickly and the count matters, so concentrate -- you will be asked for the number.
Ready
How many times did a pale shape touch an edge?
Now. Apart from the shapes you were counting -- did anything else cross the box?
What just happened
This is inattentional blindness, and it is the most famous result in the psychology of attention because of one experiment in 1999. Daniel Simons and Christopher Chabris filmed people passing basketballs, asked viewers to count the passes made by the team in white, and sent a woman in a full gorilla suit strolling through the middle of the shot. She stops. She beats her chest. She walks off. She is on screen for nine seconds.
About half of viewers do not see her. Not "forget her" -- do not see her, and frequently refuse to believe the replay is the same footage.
The crucial distinction from Exhibit № 6 is that nothing here was hidden, brief, or interrupted. The gorilla was large, central, slow, and unobstructed, in a scene you were staring straight at with your eyes open. Eye-tracking studies find that many people who miss it looked directly at it. The light landed on the fovea. The image was formed. And no one was home to receive it.
The part that stops being funny
In 2013 Trafton Drew and colleagues took the joke to work. They inserted a small image of a gorilla -- about the size of a matchbox, some forty-eight times larger than the nodules they were looking for -- into the last slides of a chest CT scan, and gave the scans to twenty-four expert radiologists to search for lung nodules.
Twenty of the twenty-four missed the gorilla. Eye-tracking showed that most of them had looked straight at it.
These were not careless people; they were specialists doing the thing they are best in the world at, and their expertise was precisely the problem. They had a superbly tuned filter for nodules, and a gorilla is not a nodule. The same effect has been found with radiologists, air-traffic controllers, and anyone whose job is to look hard for one specific thing.
Why your brain does this on purpose
It is tempting to file this under human error. It is closer to human design.
Far more arrives at your senses each second than could possibly be turned into experience -- Exhibit № 29 counts some of it. Something has to choose, and choosing means discarding, and discarding competently means discarding whatever does not match the task. A system that let every unexpected gorilla through would not be more conscious. It would be unable to count anything.
So attention is not a spotlight illuminating a dark room. It is much closer to a budget. And Exhibit № 28 is about what happens at the door where that budget is spent -- the fact that the same door can be busy for a fifth of a second is the small version of what just happened to you for fifteen.
The bit that should worry you slightly
Every previous room in this wing showed you a gap in your experience. This one shows you something else: you cannot tell, from the inside, when you are having one.
Your experience of the last fifteen seconds was not of a partial scene with a hole in it where a gorilla ought to be. It was of a complete, unremarkable box with some shapes in it. Seamless. Fully furnished. Wrong.
That is the same trick as Exhibit № 1, run at a higher level: not filling in a missing patch of retina, but filling in a missing event, and reporting the result as the whole of what happened. You will walk out of here confident that you know roughly what you have been aware of today. You have precisely as much evidence for that as you had two minutes ago.
Take it further
Simons and Chabris's The Invisible Gorilla is the book, and it is generous about how badly the authors themselves are fooled by related effects. The radiology study is Drew, Võ & Wolfe (2013), The invisible gorilla strikes again. The ancestor of all of it is Ulric Neisser's selective-looking work in the 1970s, which used superimposed video of people playing games and a woman with an umbrella, and which the field largely ignored for twenty years -- an oversight with a certain poetry to it.
One honest note: our shapes are a cheap replica of a much better experiment. We chose the counting task and the crossing object to give the effect its best shot, and a single trial on one visitor proves nothing either way. If you saw the gorilla, that is not evidence of superior attention; roughly half of everybody does.