Research Group
in Analytic Philosophy

Fallibility, Rational Belief, and Knowledge (FARBEK)

Duration: 2014 - 2017

Code: FFI2013-45968-P

Principal Investigator

Sven Rosenkranz (

All researchers

Sven Rosenkranz (ICREA-UB)

Esa Diaz-Leon

Patrick Greenough

Hannes Leitgeb 

Aidan McGlynn

Moritz Schulz

Crispin Wright




We are fallible epistemic agents. The question is what admission of the fact
rationally requires. The received view is that admission of one's fallibility
wrt a given area rationally precludes assignment of subjective probability 1 to
propositions in that area. At the same time, it is common ground that admission
of one's fallibility wrt a given area, unlike endorsement of scepticism about
it, does not rationally impede one's aspiration to knowledge in that area. In
one's attempt to qualify as someone who knows H by believing it, one trusts
that one's total evidence E suffices for knowledge of H, and hence that the
only possibilities that E leaves open are ones in which H holds. It would
follow that, in the light of one's own fallibility, the space of possibilities
over which one's credence function is defined exceeds the space of
possibilities consistent with what one takes oneself to know. Accordingly, in
trusting E to suffice for knowledge of H, one ought nonetheless to assign E, or
the proposition that E suffices for knowledge of H, a subjective probability
less than 1. The credence assigned to H will thus be constrained by the
possibility of one's having misjudged the force of the evidence one actually
has. This suggests that, according to the received view, credences to degree 1
are construed as claims to Cartesian certainty rather than knowledge, and hence
as being hardly ever rational. The present project aims to explore, in a
constructive spirit, an alternative to the received view. Its initial
hypotheses are (A) that the interval over which credences are defined, spans
the spectrum between outright belief and disbelief, where its poles might be
construed as claims to knowledge rather than Cartesian certainty, so that
credences can naturally be taken to aim at evidential probabilities; (B) that
admission of one's fallibility might after all impose no rational constraints
on the assignment of credences - and so, in particular, not preclude
assignments of subjective probability 1 - but merely call for one's readiness
to stand corrected in the light of future evidence, whose availability one has
no means of ruling out; and (C) that the trust invested in one's evidence, when
one seeks to qualify as someone who has knowledge on its basis, though
delimiting the space of possibilities over which one's credence function is
defined, might best not be construed in terms of belief whose rationality would
in turn require evidence for its truth.

Total budget: €66550
1 FPI fellow


  • Sven Rosenkranz. 2018

    The Structure of Justification

    Mind 127 (2018), 309-38