80-821 Formal Epistemology


Instructor: Kevin T. Kelly.

E-mail kk3n@andrew.cmu.edu.

Phone: X8567.

Time: 1:30 AM to 3:50 AM  Wed

Room DH 4303

Office 135 K BH.

Office hours: 1:00 – 2:00 PM Tuesday and Thursday or by appt.

Text: online scanned papers linked to this page.


Read, listen, talk, and write a final paper.


Epistemology is traditionally the study of the nature of knowledge and associated concepts including belief, justification, evidence, truth, testimony, and coherence.  Most contemporary work in epistemology is already logically structured, up to a penumbra of vagueness that is sometimes intentional and that sometimes represents a lack of contact with logical and probabilistic details.  Formal epistemology aims to improve the situation in Epistemology via rigorous application of potentially relevant mathematical tools, most notably, Bayesian decision theory and conditioning, epistemic modal logic, belief revision theory, and formal learning theory.  Traditional epistemologists are curious whether and to what extent the excursion into mathematical detail will really improve the situation, rather than merely shifting the focus from genuine questions to merely "technical" ones. 

Since there is no graduate course on Epistemology in the department, we will begin with a rapid review of some classic papers in epistemology.  Then we will review some recent, central results in formal learning theory.  We will start with probabilistic belief and Bayesian epistemology.  Then we will consider propositional belief states, modal epistemic logic, and belief revision theory.  Finally, we will examine attempts to connect probabilistic with propositional belief, focusing on the lottery paradox and theories of acceptance.

One major issue that has received less attention in formal epistemology than in standard epistemology is the relationship between justification and truth.  Standard epistemic logic and Bayesian credal states say little about how justification relates to truth.  I am interested in addressing that lacuna by paying more attention to the concept of truth-conduciveness, itself.

I. Short Introduction to Epistemology


Analysis of knowledge: Matthias Steup
Internalist/Externalist debate:  George Pappas
Foundationalist/Coherentist debate:  Jonathan Kvanvig  Richard Fumerton.
Skepticism: Matthias Steup
Epistemic paradoxes:  Roy Sorensen


The need for a "fourth condition" for knowledge and a causal account.  Edmund Gettier, Alvin Goldman

A truth-tracking account. Robert Nozick.
An explanatory chain account.Alan Goldman
An indefeasibility account. Keith Lehrer

Foundationalism Alston.

Coherentism Bonjour

Reliabilism Goldman

Contextualism Annis

Anti-skeptical positions G. E. Moore,  Norman Malcolm,  Robert Nozick

I. Probabilistic Belief States

Diachronic Dutch book:

van Fraassen
Wlodek Rabinowicz
Skyrms (focus on this one)

Confirmation and Coherence

William Talbott

Jim Joyce

Hayek and Joyce
Bovens and Hartmann

III.  Connecting Propositions with Probabilistic Beliefs

O.K., I can count, but Hannes is coming, so III precedes II.  

Last time we looked at Hannes' stuff.  I fixed the link to his talk.

Hannes Leitgeb paper
Hannes Leitgeb talk

Next week is Lin and Kelly

Following week will be guest appearance by Paul and Horacio

Related stuff:

Conditional logic:  Arlo-Costa

Lottery paradox:  Wikipedia

Wheeler, Kyburg, Teng

II.  Propositional Belief States

OK, let's look at chapter 5 of Gierasimczuk's thesis, where learning and tense logic are mixed.  That is something we definitely wanted to look at. 


Epistemic logic and belief revision:  Vincent Hendricks and John Symons
Belief revision:  Sven Ove Hansson.

Learning and Epistemic Logic

Learning theoretic logic: Gierasimczuk's new dissertation, Chapters 1-3.
Learning theoretic logic: Gierasimczuk's new dissertation, Chapter 4.
Learning theoretic logic: Gierasimczuk's new dissertation, Chapter 5.
Learning theoretic logic: Gierasimczuk's new dissertation, Chapter 6.
Learning theoretic logic: Gierasimczuk's new dissertation, Chapter 7.

Final Session:  Non-luminosity and KK.

Chapter 5 Knowledge and its Limits
Improbable Knowledge
Very Improbable Knowledge