Truth-telling under Oath

Sometimes people lie. Societies have responded to these falsehoods with ex post punishments and ex ante institutions designed to commit people to truth-telling. In antiquity, a merchant who claimed to have been robbed during his journey was asked to take an oath to avoid retaliation from those who were supposed to receive the goods involved. From a rational perspective, however, an oath might be considered useless. A rational person who sees veracity as a moral obligation does not need to swear an oath. Yet someone who wouldn’t think twice about telling a lie in public will confirm it by an oath. Still, oaths are used in many modern societies. Very surprisingly however, little or no attention has been given to the oath in experimental social sciences. Our research program fills this gap. 

In this paper, we explore whether people who voluntarily sign a solemn truth-telling oath are more committed to sincere behavior when offered the chance to lie. Based on the existing literature on the non-monetary incentives to tell the truth, an oath can be expected to change behavior through two possible channels: a tighter link between words and actions, and a reminder of the moral issue involved in lying. Our aim is to separate and identify these two channels. 

To that aim, we examine how the oath affects truth-telling in two different contexts: a neutral context replicating the typical experiment in the literature, and a context in which subjects are reminded that a lie is a lie. This treatment introduces moral reminders of ethical standards by explicitly labeling untruthful communication as a "lie" and truthful communication as "truth." Our main findings are twofold. First, we find that "loading" the decision problem with an explicit mention of its truth-telling content drastically affects a subject’s willingness to lie: half of the original lies disappear. Second, a truth-telling oath strongly reinforces the non-monetary incentives to tell the truth: another third of the lies are eliminated under oath. What is more surprising, and puzzling, is that the oath works only in the loaded environment. We see no difference in behavior when the oath is carried out in a neutral environment. 

We formulate two assumptions to explain why people do not respond to the oath in the neutral environment. First, subjects don’t realize that the task they are performing in the lab is about lying. The oath is therefore not effective because it is irrelevant. The second assumption states that subjects in our neutral treatment do understand that it is about lying but, because lying is an available choice, lying is not problematic. It is allowed by the experimenter. In other words, the neutral environment could give subjects more "wiggle" room to rationalize lying behavior.  

Although these two assumptions lead to the same behavior, i.e. lying under oath, they have testable consequences on response times‚ the time taken by subjects to make their decision on which message to send. If the first explanation holds, response times in the neutral environment should be of comparable magnitude whether subjects are under oath or not. The oath should make no difference because subjects would not consider untruthful communication as lies in the neutral environment. In contrast, longer response times under oath would suggest the second explanation: subjects under oath face a greater moral dilemma. We show that subjects who lie under oath in the neutral environment do realize what their action means in terms of honesty, because when they are under oath they take more time to decide. Yet they still choose to lie. The choice in the neutral communication game is therefore seen as a lying decision, but subjects view that kind of lying as something allowed by the experimenter, which makes the oath ineffective.