The psychology of abductive reasoning: a preliminary approach

In a recent essay entitled, The mind doesn’t work that way, Jerry Fodor affirms that we will truly understand the phenomenon of cognition only after we solve what artificial intelligence researchers call the "frame" problem. He states that solving this problem will require "global inferences", otherwise known as abductive inferences.
Epistemology has long been interested in abduction, notably with C.S. Peirce (CP 2. 619-644 et CP 5.180-212). Today, AI is particularly interested in this type of reasoning and we are beginning to see an attempt at integration with induction, the other major form of non demonstrative reasoning (Flach & Kakas 2000). It seems to be becoming clearer that inferential processes that underly the great majority of our reasoning are plausible, revisable, and uncertain. In psychology, tradition dictates a focus on deduction, in the classic sense of the word, but it also seems to be starting to give more attention to plausible reasoning.

I will begin by discussing the problem of plausible reasoning by examining both the work on "defeasible" reasoning from the field of argumentation (Toulmin 1958 and current followers) and the work on non-monotone logic in AI (Krauss, Lehmann, Magidor 1990). We will then look at the fundamental abduction scheme that can first be described in a general fashion (Gabbay and Woods 2005) followed by a more specific and diverse explanation. This will lead us to finally establish an initial typology of abductive reasoning. I will finish by presenting preliminary experimental results from the area of conditional reasoning that tend toward the validation of the connection between individuals' inferences and the logic of plausible arguments. This presentation is particularly interested in the famous fallacious scema of consequence affirmation that we can tie to a simplified form of abductive reasoning.


Click here to view the presentation.

(If you are having trouble viewing, see technical advice)