researchers showed volunteers a colour swatch of the sort one might pick up in the paint aisle of the local hardware store and allowed them to study it for five seconds.21 Some volunteers then spent thirty seconds describing the colour (describers), while other volunteers did not describe it (non-describers). All volunteers were then shown a lineup of six colour swatches, one of which was the colour they had seen thirty seconds earlier, and were asked to pick out the original swatch. The first interesting finding was that only 73 per cent of the non-describers were able to identify it accurately. In other words, fewer than three quarters of these people could tell if this experience of yellow was the same as the experience of yellow they had just a half-minute before. The second interesting finding was that describing the colour impaired rather than improved performance on the identification task. Only 33 per cent of the describers were able accurately to identify the original colour. Apparently, the describers’ verbal descriptions of their experiences ‘overwrote’ their memories of the experiences themselves, and they ended up remembering not what they had experienced but what they had said about what they experienced. And what they had said was not clear and precise enough to help them recognize it when they saw it again thirty seconds later.
Most of us have been in this position. We tell a friend that we were disappointed with the house chardonnay at that trendy downtown bistro, or with the way the string quartet handled our beloved Bartok’s Fourth, but the fact is that we are unlikely to be recalling how the wine actually tasted or how the quartet actually sounded when we make this pronouncement. Rather, we are likely to be recalling that as we left the concert, we mentioned to our companion that both the wine and the music had a promising start and a poor finish. Experiences of chardonnays, string quartets, altruistic deeds and banana-cream pie are rich, complex, multidimensional and impalpable. One of the functions of language is to help us palp them–to help us extract and remember the important features of our experiences so that we can analyse and communicate them later. The New York Times online film archive stores critical synopses of films rather than the films themselves, which would take up far too much space, be far too difficult to search and be thoroughly useless to anyone who wanted to know what a film was like without actually seeing it. Experiences are like movies with several added dimensions, and were our brains to store the full-length feature films of our lives rather than their tidy descriptions, our heads would need to be several times larger. And when we wanted to know or tell others whether the tour of the sculpture garden was worth the price of the ticket, we would have to replay the entire episode to find out. Every act of memory would require precisely the amount of time that the event being remembered had originally taken, which would permanently sideline us the first time someone asked if we liked growing up in Chicago. So we reduce our experiences to words such as happy, which barely do them justice but which are the things we can carry reliably and conveniently with us into the future. The smell of the rose is unresurrectable, but if we know it was good and we know it was sweet, then we know to stop and smell the next one.
Perceiving Differences
Our remembrance of things past is imperfect, thus comparing our new happiness with our memory of our old happiness is a risky way to determine whether two subjective experiences are really different. So let’s try a slightly modified approach. If we cannot remember the feeling of yesterday’s banana-cream pie well enough to compare it with the feeling of today’s good deed, perhaps the solution is to compare experiences that are so close together in time that we can actually watch them change. For instance, if we were to do a version of the colour-swatch experiment in which we reduced the amount of time that passed between the presentation of the original swatch and the presentation of the lineup, surely people would have no problem identifying the original swatch, right? So what if we reduced the time to, say, twenty-five seconds? Or fifteen? Ten? How about a fraction of one And what if, as a bonus, we made the identification task a bit easier by showing volunteers a colour swatch for a few seconds, taking it away for just a fraction of a second, and then showing them one test swatch (instead of a lineup of six) and asking them to tell us whether the single test swatch is the same as the original. No intervening verbal description to confuse their memories, no rival test swatches to confuse their eyes, and only a sliver of a slice of a moment between the presentation of the original and test swatches. Gosh. Given how simple we’ve made the task, shouldn’t we predict that everyone will pass it with, um, flying colours?
Yes, but only if we enjoy being wrong. In a study conceptually similar to the one we just designed, researchers asked volunteers to look at a computer screen and read some odd-looking text.22 What made the text so odd was that it alternated between uppercase and lowercase, so that it lOoKeD lIkE tHiS. Now, as you may know, when people seem to be staring directly at something, their eyes are actually flickering slightly away from the thing they are staring at three or four times per second, which is why eyeballs look jiggly if you study them up close. The researchers used an eye-tracking device that tells a computer when the volunteer’s eyes are fixated on the object on the screen and when they have briefly jiggled away. Whenever the volunteers’ eyeballs jiggled away from the text for a fraction of a second, the computer played a trick on them: it changed the case of every letter in the text they were reading so that the text that lOoKeD lIkE tHiS suddenly LoOkEd LiKe ThIs. Amazingly, volunteers did not notice that the text was alternating between different styles several times each second as they read it. Subsequent research has shown that people fail to notice a wide range of these ‘visual discontinuities’, which is why filmmakers can suddenly change the style of a woman’s dress or the colour of a man’s hair from one cut to the next, or cause an item on a table to disappear entirely, all without ever waking the audience.23 Interestingly, when people are asked to predict whether they would notice such visual discontinuities, they are quite confident that they would.24
And it isn’t just the subtle changes we miss. Even dramatic changes to the appearance of a scene are sometimes overlooked. In an experiment taken straight from the pages of Candid Camera, researchers arranged for a researcher to approach pedestrians on a college campus and ask for directions to a particular building.25 While the pedestrian and the researcher conferred over the researcher’s map, two construction workers, each holding one end of a large door, rudely cut between them, temporarily obstructing the pedestrian’s view of the researcher. As the construction workers passed, the original researcher crouched down behind the door and walked off with the construction workers, while a new researcher, who had been hiding behind the door all along, took his place and picked up the conversation. The original and substitute researchers were of different heights and builds and had noticeably different voices, haircuts and clothing. You would have no trouble telling them apart if they were standing side by side. So what did the Good Samaritans who had stopped to help a lost tourist make of this switcheroo? Not much. In fact, most of the pedestrians failed to notice–failed to notice that the person to whom they were talking had suddenly been transformed into an entirely new individual.
Are we to believe, then, that people cannot tell when their experience of the world has changed right before their eyes? Of course not. If we take this research to its logical extreme we end up as extremists generally do: mired in absurdity and handing out pamphlets. If we could never tell when our experience of the world had changed, how could we know that something was moving, how could we tell whether to stop or go at an intersection, and how could we count beyond one? These experiments tell us that the experiences of our former selves are sometimes as opaque to us as the experiences of other people, but more important, they tell us when this is most and least likely to be the case. What was the critical ingredient that allowed each of the foregoing studies to produce the results it did? In each instance, volunteers were not attending to their own experience of a particular aspect of a stimulus at the moment of its transition. In the colour-swatch study, the swatches were swapped in another room during the thirty-second break; in the reading study, the text was changed when the volunteer’s eye had momentarily jiggled away; in the door study, the researchers switched places only when a large piece of wood was obstructing the volunteer’s view. We would not expect these studies to show the same results if burnt umber became fluorescent mauve, or if this became t h a t, or if an accountant from Poughkeepsie became Queen Elizabeth II while the volunteer was looking right at her, or him, or whatever.