Stronger evidence isn't always better: A role for social inference in evidence selection and interpretation

Much of what we know comes from other people, and the quantity of information provided is often constrained by time or space. For a communicator, what information they choose to convey depends not just on the nature of their topic, but also on the social inferences their listeners will make about them based on what they say. For the listener, their interpretation of information given to them depends not just on the information itself, but also on what inferences they make about the bias and motivations of the communicator they received it from. In this paper we explore how and whether these social factors interact with the “true” nature of the information being communicated. We find that stronger evidence does not always lead to stronger conclusions and often leads to increased perceived bias. Communicators, perhaps for this reason and perhaps for others, often modulate the evidence they present to be less unanimous than warranted. This has implications for real-world situations, like communicating about climate change: in such situations, both communicators and listeners behave in what may be individually rational ways, but the end result is that the underlying truth gets distorted.