Computing Science seminar by Franz Berto

Computing Science seminar by Franz Berto
-

This is a past event

Relevant Semantics and Inconsistent Representation

Ordinary epistemic logics, as phrased via possible worlds semantics, face a group of problems gathered under the label of logical omniscience: an epistemic agent is represented as being consistent in its beliefs, and as believing/knowing all the logical consequences of what is believed/known. This makes for an idealized picture of knowledge- and belief-representation, unreal for finite, fallible cognitive beings like us. On the other hand, so-called relevant logics notoriously make Ex contradictione quodlibet (ECQ), the classically valid principle that a contradiction entails everything, fail. And the most popular semantics for relevant logics are world semantics including so-called non-normal or impossible worlds (see [Berto 2009]). These can be thought of as situations where the truth conditions of logical operators are different ([Mares 1997], [Restall 1996]). Non-normal world semantics provide counterexamples to ECQ. One promising interpretation of the framework takes worlds as information states or conduits thereof (see [Mares 2004]). Accordingly, relevant logics are claimed to model "inconsistency-robustness": our capacity of reasoning in the face of inconsistent information without inferring arbitrary consequences.

What is less known is that non-normal worlds can directly model our having inconsistent representations. Arguably, this latter capacity is tightly connected to, if not a precondition of, the former. This talk explores the phenomenon by combining a formal setting with philosophical discussion.

I present a modal semantics for for a first-order language, including non-normal worlds, along the lines of [Priest 2005]. Non-normal worlds are such that, at them, structured formulas can behave anarchically: their truth values are not determined recursively, bu directly assigned in an arbitrary way, as per the [Rantala 1982] impossible worlds frames for epistemic logic. The semantics for the relevant epistemic operator, called the representation operator, is phrased via a binary accessibility relation, R, called representational accessibility (R-accessibility). `wRw1' is read intuitively as the claim that, at world w1, things are as they are represented to be at w: it is represented that A (at w) just in case A is true at all w1 where things are as they are represented.

A longer abstract will be distributed as an attachment to the email announcements.

Franz' URL is http://www.abdn.ac.uk/philosophy/staff/details.php?id=f.berto

Speaker
Dr Franz Berto
Hosted by
Kees van Deemter
Venue
MT203