Gerstenberg, T; (2013) Making a Difference: Responsibility, Causality and Counterfactuals. Doctoral thesis (PhD), UCL (University College London).

Preview

Text 1383490_gerstenberg.pdf Download (8MB) | Preview

Abstract

In this thesis, I develop a general framework of how people attribute responsibility. In this framework, people's responsibility attributions are modelled in terms of counterfactuals defined over a causal representation of the situation. A person is predicted to be held responsible to the extent that their action made a difference to the outcome. Accordingly, when attributing responsibility we compare what actually happened with the outcome in a simulated counterfactual world in which the person's action had been different. However, a person can still be held responsible for an outcome even if their action made no difference in the actual situation. Responsibility attributions are sensitive to whether a person's action would have made a difference in similar counterfactual situations. Generally, responsibility decreases with the number of events that would have needed to change from the actual situation in order to generate a counterfactual situation in which the person's action would have been pivotal. In addition to how close a person was to being pivotal, responsibility attributions are influenced by how critical a person's action was perceived prior to the outcome. The predictions derived from this general framework are tested in a series of experiments that manipulate a person's criticality and pivotality by varying the causal structure of the situation and the person's mental states. The results show that responsibility between the members of a group diffuses according to the causal structure which determines how individual contributions combine to yield a joint outcome. Differences in the group members' mental states, such as their knowledge about the situation, their expectations about each other's performance as well as their intentions, also affect attributions. Finally, I demonstrate how this general framework can be extended to model attributions for domains in which people have rich, intuitive theories that go beyond what can be expressed with simple causal models.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item