QR6.2.1 Growing Processing

A transistor can’t grow into a computer but a cell can grow a brain, as we all did, so building information processing isn’t the same as evolving it. Evolution found a path from cells to a brain and embryo brains grow by following that path. We can build a computer but we can’t grow one as nature does.

Brains and computers both use electricity to power on/off units that process data, so brain neurons are logic gates that process data just as computer transistors do (McCulloch & Pitts, 1943). Sensorimotor channels also mirror computer input-output channels so brain-computer theories propose that nerves process the senses to give muscle output as computers process input and output (Churchland & Sejnowski, 1992). But the comparison misleads because growing and building a processor are entirely different challenges (Whitworth, 2008).

Figure 6.2 Von Neumann Architecture

We build a computer at leisure then switch it on but an evolving brain is always on because life never stops. Our computers use the Von Neumann design of a central processing unit (CPU) with memory that processes input to give output (Figure 6.2) because then it always knows what to do next but if the CPU fails, everything does. Biological parts fail regularly, so a system that fails if a part does is too fragile to survive. Evolution had to find another way so brains don’t have a central processing unit.

To understand the brain, one must understand evolution. Darwin’s natural selection is that traits gradually change over time to select what survives (Figure 6.3).

Figure 6.3 Evolution is Gradual

For a brain, the requirements are variability, change and survival:

1. Variability. Brains vary because nerves have autonomy, the ability to act based on internal direction. If neurons didn’t grow and act as they chose, the brain couldn’t evolve so neural freedom allows evolution while absolute central control denies it, hence the brain had to decentralize control.

2. Change. Evolution occurs in a step-wise manner so brains had to evolve in the same way. A brain can’t string together neurons in a series of steps that eventually gives value, as programmers do, because each step has to give value. The result is a nested hierarchy of processing , where each step adds value alone and leads to the next. It is layer upon layer, where each layer evolved while the previous one was still operating.

3. Survival. To survive, a brain must add value, say by moving a body towards the light, but even a simple sense like light detection is useless if it isn’t acted upon. To survive, a brain must control the feedback loop between sensory input and muscle output.

Decentralized control, nested hierarchies and feedback control are the brain principles that allowed it to evolve. They explain it better than any computer analogy because the information processor that nature evolved had different requirements than the computers we build.