How to avoid the fallacies of cognitive subtraction in brain imaging

Functional neuroanatomy of cognitive processes is generally derived by subtraction of a control task from an experimental task. We show that this method is logically untenable and propose to replace it with an adaptation of the specific-effect method and of the additive-factor method. The basic flaw that undermines the subtractive method is the comparison between different tasks. We argue that the specific-effect method should be used when investigating activations produced by different levels of a qualitative variable, whereas the additive-factor method should be used for quantitative variables. The use of these methods allows one to distinguish between parallel and serial stages of processing and between local and distributed processing.