of unspecified authenticity. Next, you sit down in front of a computer screen where you’re presented with a square divided into two triangles by a diagonal line. The trial starts, and for one second, twenty randomly scattered dots flash within the square (see the diagram below). Then the dots disappear, leaving you with an empty square, the diagonal line, and two response buttons, one marked “more-on-right” and the other marked “more-on-left.” Using these two buttons, your task is to indicate whether there were more dots on the right-hand or left-hand side of the diagonal. You do this one hundred times. Sometimes the right-hand side clearly has more dots. Sometimes they are unmistakably concentrated on the left-hand side. Other times it’s hard to tell. As you can imagine, you get pretty used to the task, as tedious as it may be, and after a hundred responses the experimenter can tell how accurately you can make these sorts of judgments.
Next, the computer asks you to repeat the same task two hundred more times. Only this time, you will be paid according to your decisions. Here’s the key detail: regardless of whether your responses are accurate or not, every time you select the left-hand button, you will receive half a cent and every time you select the right-hand button, you will receive 5 cents (ten times more money).
With this incentive structure, you are occasionally faced with a basic conflict of interest. Every time you see more dots on the right, there is no ethical problem because giving the honest answer (more on the right) is the same response that makes you the most money. But when you see more dots on the left, you have to decide whether to give the accurate honest answer (more on the left), as you were instructed, or to maximize your profit by clicking the more-on-right button. By creating this skewed payment system, we gave the participants an incentive to see reality in a slightly different way and cheat by excessively clicking the more-on-right button. In other words, they were faced with a conflict between producing an accurate answer and maximizing their profit. To cheat or not to cheat, that was the question. And don’t forget, you’re doing this while still wearing the sunglasses.
As it turned out, our dots task showed the same general results as the matrix task, with lots of people cheating but just by a bit. Interestingly, we also saw that the amount of cheating was especially large for those wearing the fake sunglasses. What’s more, the counterfeit wearers cheated more across the board. They cheated more when it was hard to tell which side had more dots, and they cheated more even when it was clear that the correct answer was more on the left (the side with the lower financial reward).
Those were the overall results, but the reason we created the dots task in the first place was to observe how cheating evolves over time in situations where people have many opportunities to act dishonestly. We were interested in whether our participants started the experiment by cheating only occasionally, trying to maintain the belief that they were honest but at the same time benefitting from some occasional cheating. We suspected that this kind of balanced cheating could last for a while but that at some point participants might reach their “honesty threshold.” And once they passed that point, they would start thinking, “What the hell, as long as I’m a cheater, I might as well get the most out of it.” And from then on, they would cheat much more frequently—or even every chance they got.
The first thing that the results revealed was that the amount of cheating increased as the experiment went on. And as our intuitions had suggested, we also saw that for many people there was a very sharp transition where at some point in the experiment, they suddenly graduated from engaging in a little bit of cheating to cheating at every single opportunity they had. This general pattern of behavior is what we would expect from the what-the-hell effect, and it surfaced in both the authentic and the fake conditions. But the wearers of the fake sunglasses showed a much greater tendency to abandon their moral constraints and cheat at full throttle.*
In terms of the what-the-hell effect, we saw that when it comes to cheating, we behave pretty much the same as we do on diets. Once we start violating our own standards (say, with cheating on diets or for monetary incentives), we are much more likely to abandon further attempts to control our behavior—and from that point on there is a good chance that we will succumb to the temptation to further misbehave.
IT SEEMS, THEN, that the clothes do make the man (or woman) and that wearing knockoffs does have an effect on ethical decisions. As is the case with many findings in social science research, there are ways to use this information for both good and ill. On the negative side, one can imagine how organizations could use this principle to loosen the morality of their employees such that they will find it easier to “fake out” their customers, suppliers, regulators, and competitors and, by doing so, increase the company’s revenue at the expense of the other parties. On the positive side, understanding how slippery slopes operate can direct us to pay more attention to early cases of transgression and help us apply the brakes before it’s too late.
Up to No Good
Having completed these experiments, Francesca, Mike, and I had evidence that wearing counterfeits colors the way we view ourselves and that once we are painted as cheaters in our own eyes, we start behaving in more dishonest ways. This led us to another question: if wearing counterfeits changes the way we view our own behavior, does it also cause us to be more suspicious of others?
To find out, we asked another group of participants to put on what we told them were either real or counterfeit Chloe sunglasses. Again, they dutifully walked the hall examining different posters and views from the windows. However, when we called them back to the lab, we did not ask them to perform our matrix or dots task. Instead, we asked them to fill out a rather long survey with their sunglasses on. In this survey, we asked a bunch of irrelevant questions (filler questions) that are meant to obscure the real goal of the study. Among the filler questions, we included three sets of questions designed to measure how our respondents interpreted and evaluated the morality of others.
The questions in set A asked participants to estimate the likelihood that people they know might engage in various ethically questionable behaviors. The questions in set B asked them to estimate the likelihood that when people say particular phrases, they are lying. Finally, set C presented participants with two scenarios depicting someone who has the opportunity to behave dishonestly, and they were asked to estimate the likelihood that the person in the scenario would take the opportunity to cheat. Here are the questions from all three sets:
Set A: How likely are people you know to engage in the following behaviors?
Stand in the express line with too many groceries.
Try to board a plane before their group number is called.
Inflate their business expense report.
Tell their supervisor that progress has been made on a project when none has been made.
Take home office supplies from work.
Lie to an insurance company about the value of goods that were damaged.
Buy a garment, wear it, and return it.
Lie to their partner about the number of sex partners they have had.
Set B: When the following lines are uttered, how likely is it that they are a lie?
