Subliminal

experimental psychology class were each given five of those creatures, a T-shaped maze, and a seemingly simple assignment.⁶ One arm of the T was colored white, the other gray. Each rat’s job was to learn to run to the gray side, at which time it would be rewarded with food. The students’ job was to give each rat ten chances each day to learn that the gray side of the maze was the one that led to food and to objectively record each rat’s learning progress, if any. But it was actually the students, not the rats, who were the guinea pigs in this experiment. The students were informed that through careful breeding it was possible to create strains of maze- genius and maze-dummy rats. Half the students were told that their rats were the Vasco da Gamas of maze explorers, while the other half were told that theirs had been bred to have no sense of direction at all. In reality, no such selective breeding had been performed, and the animals were effectively interchangeable, except perhaps to their mothers. The real point of the experiment was to compare the results obtained by the two distinct groups of humans, to see if their expectations would bias the results achieved by their rats.

The researchers found that the rats the students thought were brilliant performed significantly better than the rats believed to be on the dumb side. The researchers then asked each student to describe his or her behavior toward the rats, and an analysis showed differences in the manner in which students in each group related to the animals. For example, judging from their reports, those who believed their rats to be high achievers handled them more and were gentler, thereby communicating their attitude. Of course, that might have been intentional, and the cues we are interested in are those that are unintentional and difficult to control. Luckily, another pair of researchers shared that curiosity.⁷ They essentially repeated the experiment but added an admonishment to the students that a key part of their task was to treat each rat as they would if they had no prior knowledge about its breeding. Differences in handling, they were warned, could skew the results and, by implication, their grade. Despite these caveats, the researchers also found superior performance among the rats whose handlers expected it. The students attempted to act impartially, but they couldn’t. They unconsciously delivered cues, based on their expectations, and the rats responded.

It’s easy to draw analogies with how unconsciously communicated expectations might also affect human performance, but are they accurate? One of the researchers in the rat study, Robert Rosenthal, decided to find out.⁸ His plan was to again have his students conduct an experiment, but this time they would experiment on people, not rats. That, of course, involved altering the experiment to be better suited to human subjects. Rosenthal came up with this: he asked the student experimenters—who were themselves the true subjects of the experiment—to show their subjects photographs of people’s faces and request that they rate each face on the degree of success or failure they felt it reflected. Rosenthal had pretested a large set of photos, and he gave his students only those photos that had been judged as neutral. But that’s not what he told them. He said he was trying to duplicate an experiment that had already been performed, and he told half the experimenters that their stack of photos depicted faces that had been rated as successful, and the other half that theirs were rated as failures.

In order to make sure the student experimenters did not use any verbal language to communicate their expectations, Rosenthal gave them all a written script to follow and warned them not to deviate from it in any way or speak any other words. Their job was merely to present the photos to their subjects, read the instructions, and record their subjects’ responses. One could hardly take stronger precautions to discourage experimenter bias. But would their nonverbal communication nevertheless flag their expectations? Would the human subjects respond to these cues just as the rats had done?

Not only, on average, did the students who expected their subjects to accord high success ratings to the photos obtain such ratings but, in addition, every single student who had been led to expect high ratings obtained higher ratings from their subjects than did any of those expecting low ratings. Somehow they were subliminally communicating their expectations. But how?

A year later, another set of researchers repeated Rosenthal’s study, with a twist.⁹ During the course of that study, they recorded the experimenters’ instructions to their subjects. Then they conducted another experiment, in which they eliminated the human experimenters and instead communicated the instructions to the subjects using the tape recordings, thus getting rid of all cues other than those that could be transmitted through the sound of the voice. Again the results were biased, but only about half as much. So one important way the experimenters’ expectations were communicated was through the inflection and tonal quality of their voices. But if that is just half the story, what’s the other half? No one knows for sure. Over the years, many scientists have tried to find out by doing variants of the experiment, but though they confirmed the effect, none was ever able to specify any more precisely just what the other nonverbal signals were. Whatever they were, they were subtle and unconscious and probably varied considerably among the individuals.

The lesson learned has obvious applications in our personal and professional lives, with regard to our family, our friends, our employees, our employers and even the subjects being interviewed in a marketing focus group: whether or not we wish to, we communicate our expectations to others, and they often respond by fulfilling those expectations. You can probably think of expectations, whether stated or not, that you have regarding most people you interact with. And they have expectations of you. That’s one of the gifts I received from my parents: to be treated like the Vasco da Gama rats, to be made to feel as if I could navigate my way to success in whatever I set out to do. It’s not that my parents talked to me about their belief in me, but I somehow felt it, and it has always been a source of strength.

Rosenthal went on to study precisely that—what expectations mean for our children.¹⁰ In one line of research he showed that teachers’ expectations greatly affect their students’ academic performance, even when the teachers try to treat them impartially. For example, he and a colleague asked schoolkids in eighteen classrooms to complete an IQ test. The teachers, but not the students, were given the results. The researchers told the teachers that the test would indicate which children had unusually high intellectual potential.¹¹ What the teachers didn’t know was that the kids named as gifted did not really score higher than average on the IQ test—they actually had average scores. Shortly afterward, the teachers rated those not labeled gifted as less curious and less interested than the gifted students—and the students’ subsequent grades reflected that.

But what is really shocking—and sobering—is the result of another IQ test, given eight months later. When you administer an IQ test a second time, you expect that each child’s score will vary some. In general, about half of the children’s scores should go up and half down, as a result of changes in the individual’s intellectual development in relation to his peers or simply of random variation. When Rosenthal administered the second test, he indeed found that about half the kids labeled “normal” showed a gain in IQ. But among those who’d been singled out as brilliant, he obtained a different result: about 80 percent had an increase of at least 10 points. What’s more, about 20 percent of the “gifted” group gained 30 or more IQ points, while only 5 percent of the other children gained that many. Labeling children as gifted had proved to be a powerful self-fulfilling prophecy. Wisely, Rosenthal hadn’t falsely labeled any kids as being below average. The sad thing is that such labeling does happen, and it is reasonable to assume that the self-fulfilling prophecy also works the other way: that branding a child a poor learner will contribute to making the child exactly that.

HUMANS COMMUNICATE VIA a rich linguistic system whose development was a defining moment in the evolution of our species, an innovation that remade the character of human society. It’s an ability that seems to be unique.¹² In other animals, communication is limited to simple messages, such as identifying themselves or issuing warnings; there is little complex structure. Had Hans, for example, been required to answer in complete sentences, the gig would have been up. Even among primates, no species naturally acquires more than a few signals or combines them in anything but a rudimentary manner. The average human, on the other hand, is familiar with tens of thousands of words and can string them together according to complex rules, with hardly any conscious effort, and without formal instruction.

Scientists don’t understand yet how language evolved. Many believe that earlier human species, such as Homo habilis and Homo erectus, possessed primitive language-like or symbolic communication systems. But the development of language as we know it probably didn’t occur until modern humans came into the picture. Some say language originated one hundred thousand years ago, some later; but the need for sophisticated communication certainly became more urgent once “behaviorally modern” social humans developed, fifty thousand years ago. We’ve seen how important social interactions are to our species, and social interactions go hand in hand with the need to communicate. That need is so powerful that even deaf babies develop language-like gesture systems and, if taught sign language, will babble using their hands.¹³