Login |

Conditionals and Information Transfer




PI: Prof. Dr. Wolfgang Spohn, Department of Philosophy, University of Konstanz

Postdoc: Dr. Eric Raidl, Department of Philosophy, University of Konstanz


The general goal of the renewal project "Conditionals and Information Transfer" is both to address two questions from the first phase that have remained partially unanswered (Part 1) and to tie the responses with a complex of questions on the information transfer effected by conditionals (Part 2). This second complex of questions and issues has not yet been explored during the first project phase.

The first sub-goal of Part 1 consists in checking the accuracy of two important hypotheses about the philosophically most contested truth-evaluability of conditionals: the hypothesis that the truth conditions of conditionals are given by the background assumptions on which the acceptance of the conditional rests (which by itself need neither be true nor false), and the further hypothesis that the conditional relation itself is truth-evaluable to the extent described by objectification-theory for ranking functions.

The second sub-goal is closely related to the issue of truth-evaluability. It consists in investigating the syntactic and semantic conditions under which conditionals can be embedded in logically complex sentences. This investigation is to be negotiated with the linguistic sisterprojects P2 and P8. It harks back to the common observation that conditionals like "if it rains, the game will be cancelled" cannot be felicitously embedded under connectives like "or" or even as the antecedent of a more complex conditional.

The exploration of these open questions from the first phase will be complemented, in Part 2, by a broad array of questions concerning the information transfer "facilitated" by conditionals. This array of questions will be organized in the following way:

The first question to be addressed is whether a hearer processes conditional information the same way in which she processes factual information. There is evidence that the dynamics of forming new (factual) beliefs and changing degrees of existing (factual) beliefs differs when the input belief is a conditional belief, but the exact nature of the impact of conditional beliefs on a given body of beliefs is unclear and yet to be investigated.

Second, building on the result of this prior investigation possible ways of translating the processing of conditional information into an algorithm shall be explored, with the hope that this knowledge can be put to good use in the design of artificially intelligent agents. Ideally, one could eventually build robots, feed them certain bits of conditional information and observe how this affects the way in which they carry out their tasks.

Third, there will be research on the impact conditional information has on the common ground, the common ground being the space of assumptions the parties to a conversation make for the sake of their conversation. Depending on the results of the investigation, our conception of the common ground will have to be altered or enriched. For example, it is an open question whether the common ground should also contain probabilistic information. Fine-tuning a philosophically more satisfying conception of the common ground and integrating that conception with a so-called dynamic or update semantics for conditionals is going to constitute the project’s conclusion.