Abstracts

Thought Experiments, Fiction, and Free Logic

Sören Häggqvist Stockholm University

Philosophical thought experiments are hypothetical cases used for different purposes. One important purpose is to test various theories or claims. To this end, philosophers stipulate a particular scenario, supposed to elicit a judgement concerning some target property or aspect.

A number of proposals have recently been offered concerning the logical form of such thought experiments (e.g. Williamson 2007; Ichikawa and Jarvis 2009; Malmgren 2011). These proposals have focussed on Gettier cases, and tried to display their form using quantified modal logic. Thus Malmgren (2011) offers the following suggestion as what she calls “the real content” of Gettier judgement:

POSSIBILITY ∃x∃p (GCx,p & JTBx,p & ¬Kx,p).

Here, the variable “x” ranges of persons/subjects while “p” ranges over propositions; the predicates are, respectively, “x stands to p exactly as in the stated scenario”, “x has justified, true belief in p” and “x knows that p”. According to Malmgren, a thinker judges this in response to a (typical) Gettier scenario.

Notably, POSSIBILITY contains no names or singular terms. One of Malmgren’s motivations is her claim that “fictional characters are essentially fictional whereas characters in philosophical problem cases are not” (2011: 299; her italics). Yet, as Williamson observes, the vignettes describing thought-experimental scenarios are “short fictional narratives … with fictional uses of proper names” (Williamson 2007: 183). Moreover, some such vignettes, although hypothetical, use names of actual people or pronouns apparently referring to the author or reader. This seems to present no additional difficulties for someone contemplating those vignettes. So ideally, we should want an account of thought experiments to have a semantics that is continuous across actual and non-acual domains.

I argue that we have reason to construe judgements concerning philosophical thought experiments as using terms, rather than quantified variables (as both Williamson and Malmgren do). But I also argue that we must allow terms without referents in the actual domain. Hence, we have reason to employ free modal logic in analyzing thought experiments. I defend this proposal against claims made by Malmgren, Williamson and Ichikawa and Jarvis.

I also sketch how my proposal extends to other philosophical thought experiments beyond Gettier cases, and how it may shed light on our thinking about imaginary scenarios, including fiction more generally.