At the end of the 20th century, it became possible to observe the structure of a living human brain. Subsequently, observing its activa-tions became possible too. Because of that, it appeared that neuroscientists could see the "brain at work". They were able to "see" what parts of the brain were active while the subject was experiencing cer-tain mental contents. From there, the view originated that, through neu-roimages, we could have access to the areas of the brain that produced our mental life. Even if one is convinced, as I am, that the mind is a product of the brain, the notion that neuroimages allow us "to see the brain at work" is misleading. The first reason of skepticism is that neuroimages rest on the assumption of "modularity". This is the idea that mental functions are mapped "one-to-one" into brain areas. However, modularity is not the only plausible way of ascribing mental functions to brain areas. Secondly, modularity becomes useful only on condition that the researcher possesses ways for determining with certainty what brain areas are active at a given moment. As of now, that is not yet the case. Also, the characteristics of the experimental task must be known in detail. This is not yet the case either: The nature of the operations that constitute a mental function and their time course are basically unknown. The most important reason of skepticism is that, even assum-ing that the conditions listed above are met, what we would have achieved is simply to map mental functions into brain areas. However, localizing does not mean explaining. Even if it were possible to map specific mental functions into well-localized brain areas (which is not yet the case) the explanatory value of localization would be doubtful. In order to explain why and how a given mental function occurs, knowing where in the brain it takes place is of little help. A true expla-nation requires being precise and explicit about the mechanisms that cause that mental function.
Keywords: Neuroimages, brain, mind, modularity.