Confabulation has a bad press in philosophy, often identified with the main obstacle to attaining self-knowledge and described as an obvious instance of epistemic irrationality. In earlier work I thought about the current definitions of confabulation, which focus on the surface features of the phenomenon, and can be divided into two broad categories: those who define confabulations as false beliefs, and those who define confabulations as ill-grounded beliefs.
In this paper though, after a brief introduction, I leave aside how confabulation should be defined, and focus instead on its costs and benefits. In particular, I ask what costs and benefits it has for the acquisition, retention, and use of information that is relevant to us. Are we epistemically worse or better off when we confabulate?
Does confabulation really compromise self-knowledge? Does it really count as an instance of epistemic irrationality? I argue that confabulatory explanations of one's attitudes and choices do not threaten self-knowledge as correct mental-state self-attribution (that is, we know what our attitudes and choices are); but they are an instance of epistemic irrationality in the sense that we "tell more than we can know", as Nisbett and Wilson (1977) famously put it. For instance, we put forward an explanation of our choice when we lack sufficient evidence relevant to the causal process behind that particular choice. As a result, our explanation is ill-grounded and, on the basis of it, we may adopt further ill-grounded beliefs.
But that is only one side of the story. Confabulation has also a wealth of benefits for our epistemic agency that are often neglected if we just focus on truth and justification as the primary epistemic goods. Primarily, the benefits of confabulation are psychological. First, offering explanations about our attitudes and choices makes us feel more competent and enables us to build links and connections between the different things we may value and choose. A sense of competence and coherence will enhance our perceived agency, that is, the sense that we do not do things randomly or under the influence of uncontrollable environmental cues, but act in accordance to our values striving to attain the goals we set for ourselves.
Another psychological benefit of confabulation is that by offering an answer to a request for an explanation we exchange information with other people, and socialisation might be enhanced as a result. Socialisation contributes to both wellbeing and cognitive performance, but also allows us to receive feedback on our explanations and, in some circumstances, build some critical distance from them. Our explanations are likely to be false, as they are not based on the relevant information, but by being "out there", as an object of conversation and discussion, they may become a source of reflection and bring knowledge eventually, either about our attitudes and choices or about other things.
This does not mean that confabulation is all things considered good for us or should be encouraged. Rather it means that, when we take steps to reduce confabulation, and tell stories that are better grounded, we should also think about how our new and improved stories support our sense of agency, so that we don't throw the baby out with the bathwater.