The brain (and, by implication, mind) have been compared to latest technological innovation in every generation. The computer metaphor is now in vogue. Computer hardware metaphors were replaced by software metaphors and, lately, by (neuronal) network metaphors.
Metaphors are not confined to philosophy of neurology. Architects and mathematicians, for instance, have lately come up with structural concept of "tensegrity" to explain phenomenon of life. The tendency of humans to see patterns and structures everywhere (even where there are none) is well documented and probably has its survival value.
Another trend is to discount these metaphors as erroneous, irrelevant, deceptive, and misleading. Understanding mind is a recursive business, rife with self-reference. The entities or processes to which brain is compared are also "brain-children", results of "brain-storming", conceived by "minds". What is a computer, a software application, a communications network if not a (material) representation of cerebral events?
A necessary and sufficient connection surely exists between man-made things, tangible and intangible, and human minds. Even a gas pump has a "mind-correlate". It is also conceivable that representations of "non-human" parts of Universe exist in our minds, whether a-priori (not deriving from experience) or a-posteriori (dependent upon experience). This "correlation", "emulation", "simulation", "representation" (in short : close connection) between "excretions", "output", "spin-offs", "products" of human mind and human mind itself - is a key to understanding it.
This claim is an instance of a much broader category of claims: that we can learn about artist by his art, about a creator by his creation, and generally: about origin by any of derivatives, inheritors, successors, products and similes thereof.
This general contention is especially strong when origin and product share same nature. If origin is human (father) and product is human (child) - there is an enormous amount of data that can be derived from product and safely applied to origin. The closer origin to product - more we can learn about origin from product.
We have said that knowing product - we can usually know origin. The reason is that knowledge about product "collapses" set of probabilities and increases our knowledge about origin. Yet, converse is not always true. The same origin can give rise to many types of entirely unrelated products. There are too many free variables here. The origin exists as a "wave function": a series of potentialities with attached probabilities, potentials being logically and physically possible products.
What can we learn about origin by a crude perusal to product? Mostly observable structural and functional traits and attributes. We cannot learn a thing about "true nature" of origin. We can not know "true nature" of anything. This is realm of metaphysics, not of physics.
Take Quantum Mechanics. It provides an astonishingly accurate description of micro-processes and of Universe without saying much about their "essence". Modern physics strives to provide correct predictions - rather than to expound upon this or that worldview. It describes - it does not explain. Where interpretations are offered (e.g., Copenhagen interpretation of Quantum Mechanics) they invariably run into philosophical snags. Modern science uses metaphors (e.g., particles and waves). Metaphors have proven to be useful scientific tools in "thinking scientist's" kit. As these metaphors develop, they trace developmental phases of origin.
Consider software-mind metaphor.
The computer is a "thinking machine" (however limited, simulated, recursive and mechanical). Similarly, brain is a "thinking machine" (admittedly much more agile, versatile, non-linear, maybe even qualitatively different). Whatever disparity between two, they must be related to one another.
This relation is by virtue of two facts: (1) Both brain and computer are "thinking machines" and (2) latter is product of former. Thus, computer metaphor is an unusually tenable and potent one. It is likely to be further enhanced should organic or quantum computers transpire.
At dawn of computing, software applications were authored serially, in machine language and with strict separation of data (called: "structures") and instruction code (called: "functions" or "procedures"). The machine language reflected physical wiring of hardware.
This is akin to development of embryonic brain (mind). In early life of human embryo, instructions (DNA) are also insulated from data (i.e., from amino acids and other life substances).
In early computing, databases were handled on a "listing" basis ("flat file"), were serial, and had no intrinsic relationship to one another. Early databases constituted a sort of substrate, ready to be acted upon. Only when "intermixed" in computer (as a software application was run) were functions able to operate on structures.
This phase was followed by "relational" organization of data (a primitive example of which is spreadsheet). Data items were related to each other through mathematical formulas. This is equivalent of increasing complexity of wiring of brain as pregnancy progresses.