PHI 313 Theory of Knowledge S2007

Bas C. van Fraassen
Room 219, 1879 Hall       E-mail: fraassen []

Topics will include the history of the problem of induction, contextual epistemology, and subjective probability,
with special reference to issues in philosophy of science.

  1. Critique of Traditional Epistemology
    A. 'Induction' versus 'induction'
    B. History of the problem :
         Method of Hypotheses vs. Induction
    C. False hopes of traditional epistemology
  2. Contextualist Epistemology
  3. Voluntarism and virtue in Epistemology
  4. Scientific Naturalism in Epistemology
  5. Radical Probabilism as an Alternative

  • Grading basis: A-F, P, Aud
  • Evaluation: midterm exam 45%, term paper 45%;
          assignments and attendance in precepts 10%.

  • Prerequisites: Two of 200, 201, 203 or permission of the instructor
  • Readings: obtainable by downloading or on library reserve.
         See bibliography linked to items in the Lecture Outline

'Induction' versus 'induction'

Certainly there is ampliative “reasoning”, in the sense of moving from evidence to opinions that go beyond that evidence . We do that all the time. Augustine mentions as example that he believes in the reality of the ocean, though he has never seen it. We may call this practice "induction" with a small letter "i".
      The so-called Problem of Induction -- associated historically with Hume's question of how induction can be justified -- however, in its contemporary form, starts with the idea that there is a logic to such “reasoning”, a recipe, a set of rules of inference, which is not only humanly or even mechanically followable, but also objectively reliable, even compelled by standards of rationality. (Whatever that is, let us call it "Induction" with a capital "I", to distinguish it from 'induction' which includes any way at all of going beyond our evidence.) That idea of Induction is what is really at issue. Since no such logic or method has ever been successfully formulated, it is an open question whether that idea is not based on a mistake.
     But if the idea is rejected, if we do not believe in Induction, what could still be rational ways of managing our opinion, whether individually or in the process of extending and changing the sciences on which we rely?

History of the problem :
Method of Hypotheses vs. Induction

From a contemporary point of view it is puzzling and surprising that our so-called 'Problem of Induction' is hardly to be found at all before the modern era. We will look at how the advent of modern science changed the philosophical scene, and examine the controversies pitting the 'method of hypotheses' against the 'method of induction' in the seventeenth and eighteenth centuries.
      Focus: Descartes, Newton, the British and Dutch Newtonians in the Enlightenment

False hopes of traditional epistemology

The 'Problem of Induction' and its ramifications played an important role in 20th century epistemology (Quine: "The Humean condition is the human condition"). We will focus on Bertrand Russell's discussion, followed by Hans Reichenbach's "Rationalism and empiricism: an inquiry into the roots of philosophical error"
      Reichenbach maintained that there is no method of Induction that is guaranteed to work under all circumstances,but also that there is a simple rule which will succeed whenever any method at all could succeed. The virtues of his simple rule ("straight rule") as well as this audacious claim were subjected to critique in recent decades. There are also attempts to support it, though in more moderate forms, including the idea that the method could be historically relativized to the family of actually competing hypotheses at a given time.
      The signal lack of success of all (even moderate)versions of Induction motivates the exploration of more 'permissive' views of knowledge and rational belief or opinion, that leave the image of such 'rules of right reason' behind.

Contextualist Epistemology

      Claims to knowledge, and attributions of knowledge, turn out to be highly context-sensitive. A seminal paper on this subject is David Lewis' “Elusive knowledge”

      Skeptical arguments have tended to trade on equivocation made possible by this, and can be countered on that basis. Starting with the offered solutions to skeptical doubts and other such puzzles, contextualist epistemology is a sustained attempt to develop a concept of knowledge that allows for both its value and its fragility.
      But the context-sensitivity of other notions, not just knowledge, but e.g. belief, opinion, evidence, and so forth, is more important -- for other reasons than solving old puzzles about skepticism. It needs to be appreciated, for example, to understand how background to any scientific controversy influences what counts and what does not count as relevant factors in the debate.
      Focus: papers by David Lewis, Keith DeRose, and critical discussions of their work.

Voluntarism and virtue in Epistemology

      Once contextual factors, such as context-specific criteria and presuppositions, are taken into account, the evaluation of how well or how rationally opinion is managed takes on a new look. Emphasis on the role of the will and choice in rational belief already had historical precedents in e.g Augustine, Descartes, and William James. This makes sense only if values other than simply "rightness", "accuracy" and "truth" can play a role in the rational management of opinion -- values such as responsibility, integrity, audacity, prudence, .... (Opinion can include advocating changes in accepted scientific theories, so it is not a merely personal matter!) But doesn't that idea flirt with relativism and irrationalism?
      Focus: William James, Wilfrid Sellars, and writings in virtue epistemology (Ernest Sosa, Lorraine Code)

Scientific Naturalism in Epistemology

     “Naturalism” has several meanings in philosophy; sometimes it is a synonym for "physicalism" or "materialism", and sometimes it refers to the project of naturalized epistemology announced by Quine, to change epistemology into a disciplinary part of cognitive science. But here it will refer to a distinctive sort of view in epistemology according to which any discussion of such a subject as science or mathematics should take for granted that there is real knowledge in the subject, and that it is precisely what the practitioners claim to know.
Roughly, the view is that

Science itself is what we know, and philosophers as much as anyone else should proceed on the basis of what we know. It is mistaken as well as self-defeating for philosophers to discuss science from a vantage point independent of or prior to their actual historical situation, replete with the science of their day.
In the particular version we shall examine, we are seen to proceed to knowledge through ampliative rules, but the reliability of those rules is not demonstrable a priori; it is demonstrable by, or follows from the results of, the empirical sciences.
      That this is indeed so can be put in question, and we shall examine this claim to see whether science alone could provide such a ground for humanly attainable epistemic security, or whether the elements found lacking in the idea of Induction are equally needed to supplement the deliverances of the empirical sciences

Radical Probabilism as an Alternative

     The notion of subjective probability provides a new representation of nuances in our opinion, and makes room for the important role played in our thinking by unformulated prior opinion, forms of expertise that we cannot put into words. Attention to the conditions of coherence can vastly improve the management of opinion, as gauged by objective standards of "calibration".
      There is one strand in probabilism, "orthodox Bayesianism" that prescribes a mechanical rule ("Conditionalizing" -- a sort of generalized Modus Ponens) as sole rational response to new experience. Richard Jeffrey introduced a more liberal form of probabilism in epistemology ("Bayesianism with a human face"), which we shall explore.

Back to top of site