counterfactual reasoning

Terms from Artificial Intelligence: humans at the heart of algorithms

Counterfactual reasoning is when you reason about something that you believe to be false, often about different possible ways the world could be or culd have been. This might be about past events, "if it hadn't rained it would have been a good day", or about enitirely fiictional sitiuations "if mice could fly ...". In classical logic this leads to contradicitons, unless one is careful to ensure that the true contradictory facts are not part of our reasoning process. This is used in mathematics in proof by contradiction. In day-to-day counterfactual reasoning, we are more able to compartmentalise, so that in a story about "if mice could fly", we would avoid thinsg that depended on the (true) fact that mice don't fly. In regret we use counterfactual reasoning to weight up the potential advanategs or costs if we had chosen alternative actions "if only I hadn't walked under the ladder ...".

Used on pages 372, 538