The Epistemic Innocence of Irrational Beliefs

Silver linings

The cover of The Epistemic Innocence of Irrational Beliefs (Oxford, 2020)I took this photograph in a place that means a lot to me, the lagoon in Nora, Sardinia. It is a wetland which is home to amazing wildlife and also to a rescue centre for injured sea turtles. To me, it is a picture of hope. There are some dark clouds in the sky. But we can see the proverbial silver linings of those clouds, where the diffraction of sunlight gives rise to a range of colours, from purple to gold. There is no human in sight, and nature clearly dominates. But a boat is floating on the water, and boats conjure ideas of safety within exploration and adventure.

The photograph conveys the spirit of my latest book, The Epistemic Innocence of Irrational Beliefs (OUP 2020), an attempt to make sense of the idea that our irrationality supports us when we navigate the physical and social world. We sometimes succeed not in spite of, but thanks to our irrational beliefs. If that is the case, then we need a framework that helps us sort irrational beliefs. Which beliefs can we discard without a second thought, because they have no epistemic benefits? Which beliefs should we replace with care, because they have significant epistemic benefits?

Why epistemic innocence?

Brief animation explaining the notion of epistemic innocence

In order to help with the sorting, I developed the notion of epistemic innocence. The notion builds on previous work and is inspired by the legal concept of an innocence defence. In some contexts, an apparent offence is acceptable if it prevented serious harm. Self-defence is the classic example. To save my own life, I hit on the head a man who is about to stab me. If the innocence defence succeeds, then the act of hitting the man on the head is no longer an offence, because it averted my likely death.

We can apply an analogue of the innocence defence to those beliefs that are irrational but are the only way for an agent to bring about an epistemic benefit. Epistemic benefits are contributions to the agent’s capacity to pursue and fulfil epistemic goals. Take the belief that I am a brilliant public speaker. The belief is ill-grounded, given the available evidence about my past performances. But the belief brings about a significant epistemic benefit if, without it, I would lack the confidence to address the audience in front of me. Without the belief, I would be unable to deliver my talk and receive the feedback which I seek. So, the belief qualifies as epistemically innocent.

Innocence is not justification

Anne Meylan: “Epistemic Innocence and how to weigh irrationality”

Epistemic innocence is a weaker notion than epistemic justification because it does not take away the undesirability of the belief. What does it mean that a belief is epistemically justified within a consequentialist framework? It means something like this: the belief has good and bad epistemic consequences and the good consequences outweigh the bad. As a result, the belief is epistemically good overall.

If we view epistemic innocence in the same way, we encounter significant problems. Anne Meylan explains why in her thoughtful commentary. We cannot weigh the epistemic irrationality of the belief against its epistemic benefits. Consider again the belief that I am a brilliant public speaker. Is the ill-groundedness of the belief outweighed by the benefits of receiving feedback from an expert audience? No matter how much I gain from the belief, its ill-groundedness remains a problem.

No redemption for irrationality

An epistemically innocent belief remains irrational notwithstanding its potential benefits. It is ill-grounded if it lacks evidential support prior to its adoption. It is impervious to counter-evidence if it fails to be responsive to evidence after its adoption. And the belief’s irrationality remains a bad thing. If I could adopt another belief that would enable me to face my audience without being ill-grounded, that is the belief that I should adopt. Similarly, the act of hitting my assailant on the head remains undesirable, even in self-defence. If a non-violent way to stop the assailant from stabbing me were available, it would be preferable.

In the rest of this week’s posts, I will offer some examples of epistemically innocent beliefs. I will talk about clinical delusions, confabulatory explanations of our behaviour, and the biases affecting our self-related beliefs and predictions. I will also respond to a number of challenges to how I developed and applied the notion of epistemic innocence.


    • Dear Doug, thank you for your comment. I am not well versed in credibilism but based on Robin’s definition of it, the idea is that there are common sense beliefs that are (at least temporarily) indubitable. In the book I do not discuss indubitability as such, though it is of course possible that, for some of the types of beliefs I review, the agent is (at least temporarily) unable to reject them or that it is very difficult for the agent to reject them. That’s not because the beliefs are foundational, but because they present the only viable explanation for a phenomenon that the agent needs to make sense of.

  1. Pingback: Mini-Heap - Daily Nous

  2. Doug Robinson

    Thanks, Lisa! I was mainly thinking of isomorphisms between the etiologies of EI and credibilism, that both are taken on board cognitively not through rational sifting and vetting but through a mysterious process whose instigating factors are hard to trace (at least by the agent).

    • Good point. That sounds like a potential isomorphism between epistemically innocent beliefs and common sense beliefs: I do not say much about aetiology in the book (at least not explicitly) but it is plausible that a variety of factors contribute to the formation of those beliefs.

Leave a Reply