Agenda

Claudio Mezzetti (University of Queensland) - Strategic Self-Deception

Abstract:

Empirical evidence from psychology and behavioral economics indicates that people's beliefs are often shaped by their preferences. People make use of motivated reasoning, processing information in a way that aligns with their preferences. They suffer from wishful thinking, believing in the truth or likely occurrence of what they want to be true or wish to occur. More generally, they engage in self-deception, misleading themselves about their situation. The literature in psychology traditionally attributes self-deception to the desire to maintain a more positive self-image.  Self-deception, however, is likely to induce people to venture into situations that may impose large costs on them; e.g., fights, life threatening challenges, etc.

The existence of these costs suggests that a tendency towards making inaccurate inferences about oneself could have been selected by evolution over time only if it comes with some material (and not just psychological) benefits. The biologist Robert Trivers famously proposed one such possible benefit: self-deception may be useful to deceive others if people are more able to convince others when they truly hold the belief they want to convey.

In this paper, we provide a rational micro-foundation for this strategic role of self-deception by showing that it emerges as a way to gain an advantage in competitive interactions. We postulate that the outcome (win or lose) of a competitive interaction between two parties depends on the strength, or type, of one of the parties, Player 1. Player 1 and Player 2 have a common prior belief about Player 1's type, but Player 1 may gather information privately before deciding whether to participate in the competitive interaction. That is, he can strategically select his self-beliefs, constrained by Bayes' law and the cost of gathering information. While Player 2 does not observes Player 1's self-belief, the action of competing or  quitting by Player 1 serves as a signal of Player 1's beliefs about his own strength, as at the prior belief, Player 1 would not want to compete. If Player 1 competes, then Player 2 may accept or refuse to compete.  If both players engage, then Player 1 wins if strong and loses if weak. Both agents prefer to win rather than quit and prefer to quit rather than lose.

We show that, as long as the cost of gathering information is not too high, Player 1 stops gathering information and competes when his belief has reached a stopping belief that is independent of the cost of gathering information. Even if the cost vanishes, Player 1 does not want to strengthen his belief before competing. Thus, as suggested by Trivers Player 1 is only able to convince Player 2 of his strength when he truly holds convenient beliefs that he is likely to be strong and acts upon such self-beliefs. The lower belief at which Player 1 stops gathering information and decides not to compete, on the other hand, depends on the cost of gathering  information. The cheaper it is, the more pessimistic Player 1 has to become before he quits.  We interpret the fixed stopping belief at which Player 1 decides to compete as a Bayesian form of self-deception in the spirit of Sobel (2020} -- who defines deception as a situation where a sender's message is not as informative to the receiver as it could be. In our setting Player 1 is both the sender and the receiver, hence he self-deceives.