Can Your Total Evidence Mislead About Itself?

It’s fairly uncontroversial that you can sometimes get misleading higher-order evidence about what your first-order evidence supports. What is more controversial is whether this can result in a situation where your total evidence is misleading about what your total evidence supports: that is, where your total evidence is misleading about itself. It’s hard to arbitrate on purely intuitive grounds whether any particular example of misleading higher-order evidence is an example of misleading total evidence. This chapter tries to make progress by offering a simple mathematical model that suggests that higher-order evidence will tend to bear more strongly on higher-order propositions about what one’s evidence supports than it does on the corresponding first-order propositions; and then by arguing that given this, it is plausible that there will be some cases of misleading total evidence.

Evidence of Evidence in Epistemic Logic

The slogan ‘Evidence of evidence is evidence’ is obscure. It has been applied to connect evidence in one situation to evidence in another. The link may be diachronic or interpersonal. Is present evidence of past or future evidence for p present evidence for p? Is evidence for me of evidence for you for p evidence for me for p? The chapter discusses intra-perspectival evidential links. Is present evidence for me of present evidence for me for p present evidence for me for p? Unless the connection holds between a perspective and itself, it is unlikely to hold between distinct perspectives. Evidence will be understood probabilistically, using formal models from epistemic logic. Bridge principles between first-level and higher-level epistemic conditions often imply versions of controversial principles, such as positive and negative introspection. Formalizations of intra-perspectival principles that evidence of evidence is evidence have similarly implausible connections.

Introduction

Normally, when evidence speaks for or against believing some proposition, it does so by virtue of speaking for or against the truth of that proposition. If, for example, I look out the window and see that the sky is darkening, the evidence I have thereby acquired speaks in favor of believing that it will rain by virtue of indicating that it will, in fact, rain....

Return to Reason

This chapter discusses responses to the author’s “Rationality’s Fixed Point (or: In Defense of Right Reason).” Among other things, the chapter: explains how the author understands rationality; explains why akrasia is irrational; intuitively overviews the argument from the Akratic Principle to the Fixed Point Thesis; explains why you can’t avoid this argument by distinguishing the rational from the reasonable, ideal rationality from everyday rationality, or substantive from structural norms; responds to the suggestion that misleading higher-order evidence creates rational dilemmas; explains why the Fixed Point Thesis doesn’t assume objectivist or externalist notions of rationality; dismisses complaints about agents who aren’t able to “figure out” what’s rational; then responds to an objection that peer disagreement undermines doxastic justification. Finally, the chapter modifies the author’s steadfast position on peer disagreement to take into account cases in which peer disagreement rationally affects an agent’s first-order opinions without affecting higher-order ones.

Escaping the Akratic Trilemma

Much of the recent literature on higher-order evidence has revolved around the following three theses. First, one’s credence in any given proposition p should rationally reflect one’s evidence e bearing on the truth of p. Second, one’s credence in any given higher-order proposition p′ (concerning the evidential relation between e and p) should rationally reflect one’s evidence e′ bearing on the truth of p′. Third, it is epistemically irrational to have a high credence in p based on e, while having a high credence that e does not support p, or that one’s processing of e is somehow faulty (The Non-Akrasia Requirement). All three theses are prima facie plausible, yet they jointly lead to inconsistencies. This is what might be called The Akratic Trilemma. This chapter assesses two recent responses to The Akratic Trilemma (Titelbaum 2015; Lasonen-Aarnio 2014), argues that both responses fail, and offers a novel way out of the Trilemma.​Keywords

Predictably Misleading Evidence

Evidence can be misleading: it can rationalize raising one’s confidence in false propositions, and lowering one’s confidence in the truth. But can a rational agent know that her total evidence supports a (particular) falsehood? It seems not: if we could see ahead of time that our evidence supported a false belief, then we could avoid believing what our evidence supported, and hence avoid being misled. So, it seems, evidence cannot be predictably misleading. This chapter develops a new problem for higher-order evidence: it is predictably misleading. It then examines a radical strategy for explaining higher-order evidence, according to which there are two distinct epistemic norms at work in the relevant cases. Finally, the chapter suggests that mainstream accounts of higher-order evidence may be able to answer the challenge after all. But to do so, they must deny that epistemic rationality requires believing what is likely given one’s evidence.

Evidence of Evidence as Higher-Order Evidence

In everyday life and in science we acquire evidence of evidence and based on this new evidence we often change our epistemic states. An assumption underlying such practice is that the following EEE Slogan is correct: ‘evidence of evidence is evidence’. We suggest that evidence of evidence is best understood as higher-order evidence about the epistemic state of agents. In order to model evidence of evidence the chapter introduces a new powerful framework for modelling epistemic states, Dyadic Bayesianism. Based on this framework, it then discusses characterizations of evidence of evidence and argues for one of them. Finally, the chapter shows that whether the EEE Slogan holds, depends on the specific kind of evidence of evidence.

Higher-Order Defeat and the Impossibility of Self-Misleading Evidence

Two attractive theses about epistemic rationality—evidentialism and the enkratic principle—jointly imply that a certain sort of self-misleading evidence is impossible. That is to say, if evidentialism and the enkratic principle are both true, one’s evidence cannot support certain false beliefs about what one’s evidence supports. Recently, some epistemologists have argued that self-misleading evidence is possible, since misleading higher-order evidence doesn’t have the strong defeating force needed to rule out this possibility. The goal of this chapter is to offer an account of higher-order defeat that does indeed render self-misleading evidence impossible. Central to the account is the idea that higher-order evidence acquires its normative significance by influencing which conditional beliefs it’s rational to have. What emerges is an independently plausible view of higher-order evidence that allows us to reconcile evidentialism and the enkratic principle.

Whither Higher-Order Evidence?

A widespread view is that higher-order evidence makes a difference to whether it is rational for a person to believe a proposition. This chapter considers in what way higher-order evidence might do this. More specifically, it considers whether and how higher-order evidence plays a role in determining what it is rational to believe distinct from that which first-order evidence plays. To do this, it turns to the theory of reasons, and tries to situate higher-order evidence within it. The only place for it there, distinct from that which first-order evidence already occupies, is as a practical reason, that is, as a reason for desire or action. One might take this to show either that the theory of reasons is inadequate as it stands or that higher-order evidence makes no distinctive difference to what it is rational to believe. The chapter tentatively endorses the second option.

The Puzzles of Easy Knowledge and of Higher-Order Evidence

The goal of this chapter is to provide a unified solution to two widely discussed epistemological puzzles: the puzzle of easy knowledge and the puzzle of higher-order evidence. The chapter begins by setting out each of these two puzzles. It then briefly surveys some of the proposed solutions to each puzzle, none of which generalizes to the other. Finally, the chapter argues that the two puzzles arise because of a widespread confusion concerning the relation of substantive and structural constraints of rationality: or, in the epistemic domain, the relation of evidence and coherence. Clearing up this confusion allows us to clear up both puzzles at once.