Madness Explained: Psychosis and Human Nature

by Richard P. Bentall

Typically, the strategy we adopt when testing a hypothesis depends on the nature of the hypothesis. When some kind of choice is believed to have resulted in a good outcome (for example, when we believe that a cake is particularly nice because we used honey instead of sugar), it is sensible to test the hypothesis by looking for confirmatory evidence (for example, by baking another cake with honey but changing some of the other ingredients). This is because the result of the test is likely to be a further positive outcome (another nice cake). However, if the outcome is negative (for example, if we think that a cake tastes awful because we used margarine instead of butter) it is sensible to devise a disconfirmatory test (for example, by baking a cake with butter) as this will reduce the possibility of another disappointing result.