Computer architecture understands the most efficient information-processing systems, agnostic to hardware implementation. To understand humans in computer architecture terms, one must first ask, "What kind of computer is a human being?", before then asking, "How do humans optimally interact?"
The key distinction known by everyone from hardware engineers to business executives is the core distinction between types of operating system: real-time or flow systems vs. batch or sequential systems. The key difference is the ability for computation to respond quickly to updated input. Real-time systems are primed to respond immediately, while batch systems initiate distinct "programs" having distinct beginnings and ends.
For example, humans, robots, and even Wall Street stock-trading servers are all hard-wired with real-time systems having millisecond reaction-times. On a personal computer, the mouse and/or touchscreen use real-time systems, while the response to a click comes from a sequential system.
But it's difficult to do experiments on real-time systems, and much easier to impose start/stop times on experimental subjects. So much of neuroscience and psychology refer to discrete concepts like categories, decisions, and behaviors. Of course humans can and do engage in start/stop activities, but only as an afterthought (literally): the brain's primary computational purpose is high-bandwidth real-time response, without waiting for anything. Brains are built for flow.
When we see ourselves as flow machines, much about us makes sense, including how our bodies vibrate and communicate by vibration. In fact, in a computer-architecture view of people as flow computers, closely-bonded groups of people in close proximty are collective flow-computers, performing the same sorts of computations as individuals, but exchanging much more data and (usually) computing more reliably. Because we evolved to live in kin-groups, our computational heritage is hive-mind.