Force-based Attitudes

Chapter 11 presents a new metaphysics of confidence. The view maintains that all types of confidence—credence and thick confidence alike—are built from mixtures of cognitive force: intellectual attraction, intellectual repulsion, intellectual indifference. It is argued that such forces are recognizable elements of our everyday conception of mind, and it is shown how to model the reduction of confidence to them (with open rays in a three-dimensional volume). The resulting view is then used to reduce belief, disbelief, and suspended judgement. It is shown that the resulting view has all the good-making features of its credal-based cousin as well as good-making features of its own. Specifically, it is shown that the force-based reduction of confidence can give a much better explanation of suspended judgement than its credal-based cousin, and likewise show how the view can make good sense of mental ‘compartmentalization’.

The Bayesian Transition Theory

Chapter 4 discusses the Bayesian transition theory. The distinction is drawn between dynamics and kinematics, and it’s argued that the theory of rational inference belongs to the former rather than the latter. It’s shown that Jeffrey’s rule is thus not a rule of rational inference. Credence lent to a conditional is explained and compared to conditional credence. Two problems for Bayesian kinematics then come into focus: conditional credence is never changing in the model, nor is it ever the contact-point of rational shift-in-view. A natural conception of conditional commitment is then put forward and used to solve both these problems. Along the way it’s argued that modus-ponens-style arguments do not function in thought as logical syllogisms, since modus-ponens-style arguments specify obligatory paths forward in thought.

Inference and Rationality

A theory of rational state transition must answer four questions: are shifts within its domain brought about by agents or do they simply happen to them? Is the approach part a theory’s dynamics or kinematics? Does the approach make use of everyday or ideal rationality? Are the mental states involved coarse- or fine-grained? The questions are used to generate a sixteen-fold classification of rational shift-in-view. It is then argued that rational inference leads to the idea of a coordinated epistemic reason: roughly, a reason where causal-efficacy and evidential-relevance fuse together. This idea is illustrated with everyday examples and it is then argued that the theory of rational inference turns crucially on the non-ideal rationality of agential dynamics. The chapter closes by developing a theory of rational inference and a take on the human mind to go with it.

Puzzling about Epistemic Attitudes

Chapter 8 discusses the relation between coarse- and fine-grained attitudes. It begins with two questions: how do elements in a given attitudinal space relate to one another? and how do elements across attitudinal spaces do so? The chapter argues that coarse- and fine-grained attitudes are sui generis operators within their own spheres of influence. A familiar contradiction is generated by placing natural-looking norms for belief and credence together with a plausible story about how the attitudes fit together. A space of reactions is laid out and its eliminativist elements are rejected. Two options are then discussed: belief-first epistemology says that belief is fundamental, and confidence-first epistemology says that confidence is fundamental. Matching psychologies for these pictures are discussed.

Critical Discussion of the Belief Model

Chapter 6 argues that most of the reductive assumptions put forward by the Belief Model are wrong. It rejects the view that suspended judgement is nothing but the absence of belief and disbelief, and likewise rejects the view that disbelief is nothing but belief-in-negation. Thought experiments demonstrate that all three types of coarse-grained attitude are self-standing elements of our rational architecture. It is shown that norms for belief put forward by the Belief Model do not morph automatically into norms for disbelief or suspended judgement. Then it is shown that the model’s main transition rule does not function at all like rational shift of belief in ordinary life.

The Bayesian Model

Chapter 2 introduces the Bayesian Model of rational credence. The mathematical properties of classic probability are explained—their linearity, cardinality, and so on—as well as the use of probability functions to model mental states. A game with balls is created and diagrammed to make visually vivid how probability works. Venn diagrams and truth-tables are used to illustrate everything for the absolute beginner. Conditional Credence is explained with intuitive examples, and the standard ratio formula is introduced. A game with marbles is then created and diagrammed to pump intuition about change of probability, which leads directly to Jeffrey-style kinematics. The chapter closes by describing a notionally possible agent, Creda, whose psychology matches the Bayesian Model.

Credence-First Epistemology

Chapter 10 discusses the credence-first approach to coarse-grained attitudes. It is explained how the view underwrites a robust realism about the attitudes and why it bolsters the view that belief, disbelief, and suspended judgement are self-standing states. It is explained how credence-first epistemology dovetails with how we ordinarily describe coarse- and fine-grained attitudes and how it makes good sense of ways in which coarse- and fine-grained attitudes march-in-step in their production and rationalization of action. Further, it is explained how credence-first epistemology fits with our use of reductio-based arguments. So it’s argued that there is something deeply right in the credence-first approach. But the chapter closes with a pair of problems for the approach: credence is very often absent in the presence of coarse-grained attitudes—as a matter of descriptive fact—and credence is very often misplaced in the presence of everyday evidence—as a matter of normative fact.

Conditional Commitment and the Ramsey Test

Chapter 7 is a critical discussion of conditional commitment in both the Bayesian and the Belief model. Both use their treatment of conditional commitment as something to connect their theory of states with their respective transition theory. In doing so both models are immediately hit with technical difficulties. The Belief Model generates the ‘impossibility theorem’ first proved by Peter Gärdenfors, and the Bayesian model generates ‘triviality results’ first proved by David Lewis. Each of these technical areas is explained from scratch and diagnosed philosophically. It is argued that the bombshells discussed are best seen as showing that the Binary-Attitude Assumption is false when it comes to conditional commitment, and that there is no essential tie between conditional commitment and rational shift-in-view. Throughout the discussion the 3-place theory of conditionality is related back to Chapter 4’s restricted-vision approach to conditionality.

Belief-First Epistemology

Chapter 9 discusses two strategies for reducing credence to belief. One grounds credence in the transition of belief, the other grounds credence in the content of belief. Since the former elides the important distinction between being in a state and changing a state, the view is rejected. It is argued, though, that the latter strategy is more robust than one might suppose, e.g. the view makes good sense of reasoning with states of credence. But in clarifying its explanatory resources, the view must either reference the signature function of credence or fail to do so. In the former case the approach fails to reduce credence to belief after all. In the latter case the approach entails that credal states do not function normally. The strength-as-content view also conflicts with the manner in which belief and credence march-in-step when producing action.

Force-Based Confidence at Work

Chapter 12 puts the force-based view to systematic explanatory work. The so-called ‘dilation’ of probability sets is explained from scratch and placed within a larger theoretical setting. Systematic and puzzling patterns of intuition about the acceptability of the representor-based model’s consequences are unearthed. It is shown that the force-based view of confidence explains the intuitions, as well as those surrounding a new type of counter-example to the representor approach. The full story exposes deep links between the force-based metaphysics of confidence and three big issues: the forward-looking nature of transition theory, the manner in which confidence should match the character of one’s evidence, and the situations in which confidence is subject to content-based accuracy. It is shown that each of these topics can be handled gracefully by the force-based view.