One problem with the idea of modules is that humans tend to oversimplify and tend to make ideas like modules too real. The module idea is metaphoric and should always be interpreted as "brain function X acts something like a module that exists separately and is connected to other modules in a variety of ways.

I have a useful metaphor that compares the brain to a synthesizer that is capable of producing a very large number of sounds in complex patterns. A synthesizer is built from function-specific modules that are linked together to produce the final results.  When I was first studying electronics, I constructed a synthesizer out of modules that each performed one function. The modules had jacks on their front panels so that each could be connected to other modules by using patch cords. In the case of my first synthesizer, each module was a separate physical entity and performed its function without any help from other modules. To generate musical sounds you connected a keyboard controller that sent out a different voltage from each key to a voltage controlled oscillator that generated the sound waves at the appropriate pitch. The oscillator was typically connected to a voltage-controlled amplifier that was in turn controlled by a ramp generator, since musical sounds emerge and decay within an amplitude envelope. To finish a convincing musical sound, many modules would be connected in a single path.

This approach to synthesis provided a simple model of a modular system that is similar to explanations of how the brain works. Functions of different modules are described and their connections are supposed to explain how the brain works. The anatomic view of the brain connects modules but cannot describe  exactly what these connections do. You could imagine that there are different kinds of connections and construct theories of function accordingly. Neuroanatomy describes complex networks of connections that are often parts of webs and meshes more than point to point discrete connections, although point-to point connections do exist and are the most obvious.

The pattern of connections in my early synthesizer was expressed by mazes of patch wires linking the modules and the word "patches" is still used in musical circles to describe sounds made by specific combinations of modules. My first Moog-like synthesizer represents the simplified and concrete idea of modules, helpful for beginners to understand, but not adequate for real understanding of how the brain works. One idea is that living systems, unlike machines, are self-referential and generate their own activity. Rather than just playing the keyboard to make sounds, I spent many hours trying different patch combinations that would allow the synthesizer to generate its own sounds. It is easy to generate monotonously repetitive sounds, but the real interesting experiments lead to sound production that varied continuously or would cycle through sets of variations in interesting ways without the intervention of a human operator.

My current synthesizer, the Korg Trinity, is a magnificent electronic device that contains the equivalent of hundreds of modules and thousands of patch cords. Rather that actually building modules physically and connecting them with real wires, the Trinity simulates modules and patches by using a digital computer to calculate what the output would sound like if you had a set of modules connected in a certain way. You choose the "modules" from menus on a touch sensitive screen, choose values for many parameters that control the modules and connect modules to form the "patch" or sound that emerges in stereo from the output jacks of the synthesizer. The sound can be a single instrument played expressively, an entire string section or a complex and evolving mixture of sound effects suitable for a Star Wars soundtrack. The Trinity with its virtual modules and dense, variable inner connections comes closer to representing a modular interconnection concept that can be applied to studying the brain. You combine the knowledge of what each module does with knowledge of how different connections add and modify the function of individual modules.

Three gifted neuroscientists, Thomas D. Albright, Thomas M. Jessell, Eric R. Kandel, and Michael I. Posner stated in their review of neuroscience at the beginning of the 21st century: ^[i]

“Perhaps the greatest unresolved problem in visual perception, in memory, and, indeed, in allof biology, resides in the analysis of consciousness… a successful solution will require, at aminimum, insight into two major issues that lie at the heart of thestudy of consciousness: (1) awareness of the sensory world and (2) volition, the voluntary control of  thoughts and feelings.As with other problems in biology there are both reductionist and holistic approaches to these components of consciousness. A reductionist  approach would viewthese aspects of consciousness from a genetic, synaptic, and cellular level. However, in the case ofconsciousness it is hard to imagine any solution that would not alsorequire an understanding of the large neural networks that underlie cognition, actions, and emotion…

“The use of neuroimagingmethods during the last decade has made it possible to observe theactivity of large numbers of neurons in human subjects while they arestudied for their awareness of the sensory world and for their voluntarycontrol of thoughts and feelings. Cognitive studies using these imaging methods, such as positron emission tomography(PET) and functional magnetic resonance imaging (fMRI), are based upon changes in blood flowand blood oxygenation that occur in localized regions of the brain whenneurons increase their activity These methods have now been applied with some success to the studyof attentional orienting, visual imagery, and regulation of cognitive and emotional states. In each ofthese domains, the individual functional components have proven to be surprisingly well localized;however, each of the major functions of consciousness—such as attentional orienting to sensorystimuli and volition—involves not one but several functional components. As a result, eachfunction of consciousness appears to involve several networks and these are distributedacross a variety of brain areas. Fortunately, the enormous complexity of the problemshas been made somewhat more tractable by use of appropriate animal models. Inthe best case, as with studies of the visual system, it has proven possibleto relate neural activity studied at the cellular level in nonhuman primatesto the activity of large neural networks studied in the same brainareas but now in human subjects using brain imaging.”

For the scientists who are actually trying to figure out how each module works, the brain becomes an enormously complex puzzle and tempts you in some experimental situations to believe that function is well localized even down to a single neuron that does only one thing. At other times, brain studies suggest the opposite that localized function can move around and that distributed function is required for cognition.

This paradoxical view is similar to the dualistic aspect of quantum physics where one set of observations suggests that light acts like particles and another set that suggest that light acts like waves. In brain terms, we have to allow for events that are both localized and distributed; discrepant observations are not contradictory but rather are complementary.