Active Epistemology

Abstract

Orthodox epistemology deals with what we might call "passive agents:" agents that merely respond to what's given to them. But we care most about are active agents: agents that can sing, dance, seek evidence, make conjectures, etc. In this dissertation, I explore three ways in which this observation is relevant to the traditional questions of epistemology. I argue that it can shine a new light on the nature of knowledge, help us make sense of standard scientific practices, and help solve some famous puzzles in epistemology.

Chapter 1 asks how the fact that we know something is related to the question of whether we should seek more evidence. This problem has been discussed in the philosophical literature under the name of "the dogmatism puzzle." In this chapter, I use the multi-armed bandit model to argue for a new solution to the dogmatism puzzle: knowledge often requires proper maintenance.

Chapter 2 presents a novel account of significance testing, one of the most important practices in science. It shows that we can make sense of this practice if we accept the claim that predictions are better than accommodation. I then use this account to answer some of the many objections to significance testing.

Finally, chapter 3 argues that what we know often depends on our choices, and not merely on what is given to us. We can gain knowledge by choosing something, whereas if we fail to choose, or attempt to choose too many things, we fail to gain knowledge. I then use this idea to offer a new solution to the lottery paradox, to help understand inductive knowledge, and, once again, to make sense of significance testing and related practices.