We are likely to overestimate our capacities and make exceedingly rosy predictions about our future. This widespread bias towards optimism is a robust finding in psychology. It is also a clear case of epistemic irrationality which has serious implications for risk assessment. According to a recent article, unrealistic optimism played an important role in our estimate of the spread of COVID-19. At the end of January 2020, leading epidemiologists predicted that the virus could infect 40-70% of the world population. But in February citizens of many European countries thought that their risk of infection was around 1%.
In 2015-2016 I worked with Anneli Jefferson on a project on the costs and benefits of unrealistic optimism. I marvelled at the endless powers psychology attributes to optimistic agents. Their optimism allegedly leads to enhanced productivity and mastery, altruism, better physical and mental health, resilience, and the capacity for effective coping. Much like confabulations, even when they are not supported by evidence and are not responsive to counter-evidence, optimistic beliefs strengthen a person’s sense of agency.
Optimism and epistemic success
But do they have epistemic benefits as well? My belief that I am better than average in a number of domains contributes to my cognitive performance by reducing anxiety and self-doubt. So does the belief that I can control external factors to my advantage. Self-enhancing beliefs increase my chances to attain epistemic goals by sustaining my motivation to pursue those goals in the face of obstacles. And optimistic predictions behave as self-fulfilling prophecies in some contexts. Suppose I believe that my romantic relationship will be long-lasting and satisfying, and idealise my partner’s qualities. Those beliefs make me more likely to react constructively when a crisis strikes, contributing to the success and duration of the relationship.
In February 2020 I believed that my chance of getting COVID-19 was very small. My belief did not take into account compelling epidemiological evidence and caused me to underestimate the risk of infection. But the belief probably had some indirect epistemic benefits. Instead of adjusting my plans in light of the risk of infection, I attended a conference in Denmark as I had arranged to do. I learnt a lot from it. A less irrational belief about my risk of infection might have led me to change my plans. My overly optimistic belief qualifies for epistemic innocence.
Belief versus hope
In her tightly argued commentary on my book, Anna Ichino proposes a revision of the conditions for epistemic innocence which achieves two objectives. First, Anna wants to extend epistemic innocence to mental states that are not beliefs (non-doxastic states). Second, she wants to find a way to make epistemic innocence more demanding so that its conditions are not too easy to meet.
In the revised version, a state s is epistemically innocent if (1) s has epistemic costs; (2) s has epistemic benefits; (3) there is no other state with fewer costs than s that can deliver the same benefits as s. Now consider again my belief in February 2020 that my chances of COVID-19 infection were small. The belief continues to have epistemic costs and benefits, but no longer qualifies as innocent. That’s because a less epistemically costly state that would deliver the same benefits was available to me. I could have hoped that my chance of infection was small.
High risk, high gain
My hoping that my chance of infection was small in February 2020 was not irrational. There was no tension between my hope and the epidemiological evidence. Indeed, it is not clear that my hope had any epistemic costs. But would my hope deliver the same benefits as my belief that the chance of infection was small? Hope has known positive effects on mood and health. It also motivates: agents are motivated to act so that their hoped-for outcome turns into an actual outcome. But my hope that the chance of getting COVID-19 is small is a less powerful reason not to cancel my trip to Denmark than a belief with the same content.
Doxastic commitment is what makes epistemic innocence interesting. Switching from belief to hope reduces the epistemic risks but also the gains. That said, the idea of broadening the scope of epistemic innocence is worth pursuing further. It certainly makes sense to apply epistemic innocence to those mental states that can be the object of epistemic evaluation. Hopes may fall into that category (McCormick thinks that hope can be irrational) but I am not sure imaginings and other non-doxastic states do.