Title: Imaging and Imagining

Abstract:

I investigate the workings and epistemic credentials of counterfactual imagination: the activity of supposing that P in order to investigate what would be the case if P was the case. The mainstream way of drawing the distinction between imagination in the indicative and in the subjunctive or counterfactual mood has it that one can only suppose that P in the former way when one gives nonzero chance to P. I argue that that’s wrong. Instead, both kinds of suppositional thinking work by simulated belief revision; but while the former is governed by Bayesian conditionalization, the latter is governed by Lewisian imaging. So understood, counterfactual imagination can be rationally justified by considerations concerning belief accuracy: a specific kind of imaging, namely Laplacian generalized imaging, minimizes expected inaccuracy as measured by the Brier score. So if one endorses the claims that (expected) accuracy is the fundamental virtue of epistemic attitudes, and that accuracy is adequately measured by the Brier score, imagining via imaging is virtuous indeed.