When Less is More: Evolutionary Dynamics of Deception in a Sender-Receiver Game

TL;DR

This paper uses evolutionary game theory to analyze deception in sender-receiver interactions, revealing a counterintuitive threshold where the role with greater payoff differences tends to lose over time, with implications for understanding disinformation dynamics.

Contribution

It introduces a novel evolutionary model of deception in sender-receiver games, highlighting a threshold phenomenon and the impact of payoff differences on long-term strategy success.

Findings

The role with larger payoff differences is more likely to lose in the long run.

Slower evolving roles can exploit faster roles' strategies, affecting fixation probabilities.

Finite populations show that less frequent strategies of the slower role tend to fixate.

Abstract

The spread of disinformation poses a significant threat to societal well-being. We analyze this phenomenon using an evolutionary game theory model of the sender-receiver game, where senders aim to mislead receivers and receivers aim to discern the truth. Using a combination of replicator equations, finite-size scaling analysis, and extensive Monte Carlo simulations, we investigate the long-term evolutionary dynamics of this game. Our central finding is a counterintuitive threshold phenomenon: the role (sender or receiver) with the larger difference in payoffs between successful and unsuccessful interactions is surprisingly more likely to lose in the long run. We show that this effect is robust across different parameter values and arises from the interplay between the relative speeds of evolution of the two roles and the ability of the slower evolving role to exploit the fixed strategy of the faster evolving role. Moreover, for finite populations we find that the initially less frequent strategy of the slower role is more likely to fixate in the population. The initially rarer strategy in the less-rewarded role is, paradoxically, more likely to prevail.