The concept of ‘module’ in neuroscience (meaning sufficient for a function, given gas-in-the-tank background conditions) invariably causes more confusion than clarity. The problem is that any neuronal business of any significant complexity is underpinned by spatially distributed networks, and not just incidentally but essentially—and not just cortically, but between cortical and subcortical networks. This is true, for example, of motion perception and pattern recognition, as well as motor control and reinforcement learning, not to mention feelings such as mustering courage to face a threat or deciding to hide instead of run. It is true of self-control and moral judgment. It is likely to be true of conscious experience. The output of a network can vary as the activity of the network’s individual neurons varies. What is poorly understood is how nervous systems solve the coordination problem; i.e. how does the brain orchestrate the right pattern of neuronal activation across networks to get the job done?
This is not all that is amiss with ‘module’. Traditionally, modules are supposed to be encapsulated –aka insulated. But even the degree to which an early sensory area, such as primary visual cortex (V1) is encapsulated has been challenged. Visual neurons in V1 double their firing rate if the animal is running, no matter the identity of the visual input across conditions, to take but one example. To add to the ‘module’ mess, it turns out that specialization in an area such as the V1 appears to be dependent to some nontrivial degree on the statistics of the input. Visual cortex is visual largely because it is connected to the retina and not the cochlea, for example. Notice that in blind subjects, the visual cortex is recruited in reading braille, a high-resolution spatial—and somatosensory—task. As would be expected if specialization depends on the statistics of the network’s input, infant brains have much more plasticity in regional specialization than do mature brains. Doris Trauner and Elizabeth Bates discovered that human infants with a left hemispherectomy can learn language quite normally, whereas an adult who undergoes the same surgery will have severe language deficits.
I think of ‘module’ in the way I think of ‘nervous breakdown’—mildly useful in the old days when we had no clue about what was going on under the skull, but of doubtful explanatory significance these days.