QR6.2.2 Decentralized Control

We like to control things so our early networks used centralized control, until we found that decentralized networks like Ethernet were ten times faster. They also degrade gradually under load while centralized networks collapse entirely. When the Internet was first proposed, it was expected to fall into chaos without central control but in fact, decentralization allowed it to survive.

Figure 6.4 Schacter’s Brain Model

Early brain models expected a central executive, as in Figure 6.4, where an executive decides what to do after a conscious awareness unit accesses sensory knowledge modules, memory and higher reasoning (Schacter, 1989).

The cortex is a folded layer wrapped around the midbrain and hindbrain (Figure 6.5). It handles perceptions, voluntary acts, memory, thought, planning, attention and language, so the executive was assumed to be in the cortex.

Figure 6.5 The Cortex is the folds around the brain

Yet how can the most advanced part of the brain have an executive when its two hemispheres split the work between them? The left hemisphere directs the right side of the body and the right hemisphere the left and usually the left specializes in language and the right in spatial analysis. How can two hemispheres result in one executive?

The answer, revealed by the treatment of epilepsy, is that they don’t. Epilepsy is an illness where an electrical disturbance in one hemisphere spreads to incapacitate the cortex. An 800 million nerve bridge called the corpus callosum joins the hemispheres and cutting it in animals didn’t seem to harm them, so surgeons tried the same in epileptics to stop the epilepsy spreading and let them lead normal lives. The treatment worked but while serious side effects were expected, split-brain patients still spoke and acted normally! So little changed that some thought that the corpus callosum was just a structural support. Further studies revealed an unexpected result.

Figure 6.6 How Visual Processing is Shared

Each hemisphere moves the opposite hand but vision is different. Instead of the left hemisphere getting right eye input, it gets input from the right side of both eyes and the right hemisphere gets input from the left side of both eyes (Figure 6.6).

In split-brain studies, each eye saw half of a split screen, so in Figure 6.7, the left eye saw the snow image and the right the claw image. The corpus callosum was cut so the left hemisphere saw only the claw and the right hemisphere saw only the snow.

.

Figure 6.7 Split-brain study set-up

When subjects were asked to point to a picture that matched what they saw, the right hand picked a chicken but the left hand picked a shovel! The left hemisphere that saw a claw used the right hand to point to a chicken and the right hemisphere that saw snow used the left hand to point to a shovel, and neither was aware of the other’s choice. The two hemispheres were able to receive and send data as if each was a brain in itself, so the cortex had no central executive.

When asked why his left hand chose a shovel, a subject said “you need a shovel to clean up after chickens”. The verbal left hemisphere had no idea why the shovel was chosen as it didn’t see the snow but instead of saying I don’t know, it made something up. It tried to interpret events as best it could:

These findings all suggest that the interpretive mechanism of the left hemisphere is always hard at work, seeking the meaning of events. It is constantly looking for order and reason, even when there is none – which leads it continually to make mistakes. It tends to overgeneralize, frequently constructing a potential past as opposed to a true one.(Gazzaniga, 2002) p30

Interpreter theory sees the language and thought of the cortex as more servant than master of the brain. If the brain is a federation of agents(Minsky, 1986), the left cortex is head of human relations not the CEO as some suggest (Kaku, 2014). It acts like a diplomat to explain a system that others run.

Perhaps human intellect expanded when we formed tribes because those who better explain themselves survive to reproduce, as the animal most likely to harm a human is another human. Inventing acceptable reasons after the fact may be the evolutionary basis of our famous intellect. Logical thought, building one idea upon another in a rational way to reach an unforeseen conclusion, probably isn’t what our intellect originally evolved to do.

The left hemisphere usually specializes in language but the right hemisphere isn’t illiterate. One study of a split-brain boy (Wolman, 2012) asked the left hemisphere “Who is your favorite?” but flashed “Who is your favorite girlfriend?” to the right hemisphere. The left hemisphere made no verbal reply, as didn’t see the word girlfriend, but a nervous giggle revealed that the right hemisphere understood. The left hand the right hemisphere controlled then selected scrabble tiles from a set to spell out L-I-Z, a cute girl in his class. The right hemisphere had no vocal control but it could still read and spell. Both hemispheres are conscious in any way you care to define it:

Everything we have seen indicates that that the surgery has left these people with two separate minds, that is, two separate spheres of consciousness. What is experienced in the right hemisphere seems to lie entirely outside the realm of awareness of the left hemisphere. This mental division has been demonstrated in regard to perception, volition, learning and memory.” (Sperry, 1966)p299.

Figure 6.8 Phineas Gage

Evolution favors decentralization because if part of the brain is lost, the rest can still carry on. In a famous case, an iron rod pierced the middle and left cortical lobes (Figure 6.8) of a railway worker called Phineas Gage, who shortly after walked off, conscious and speaking. He showed disturbed behavior but lived for 13 more years and died of unknown causes. Now imagine banging a nail through a mother-board! The brain duplicates the cortex for the same reason that planes double up critical control units – to increase reliability. This duplication answers Von Neumann’s question:

How could a mechanism composed of some ten billion unreliable components function reliably while computers with ten thousand components regularly fail?

As information goes into the brain, it makes sense to peel away the processing layers to find an “I” from which all proceeds but doing so reveals no central executive in the cortex. If the body is a ship run by a decentralized brain, it has no “Captain” even at the highest level:

Studies of the structural and functional organization of the brain have shown that this organ is, to a large extent, decentralized, and processes information in parallel in countless sensory and motor subsystems. In short, there is no single homunculus in our brains that controls and manages all these distributed processes.(Singer, 2007)

No-one searches the Internet to find its “center” so why expect sense data to funnel into a brain center? The answer is that we experience one “I”, but neuroscience doesn’t support this:

In contrast to this first-person experience of a unified self, modern neuroscience reveals that each brain has hundreds of parts, each of which has evolved to do specific jobs – some recognize faces, others tell muscles to execute actions, some formulate goals and plans, and yet others store memories for later integration with sensory input and subsequent action.(Nunez, 2016) p55.

Dennett argues that the conflict between one “I” and a decentralized brain means the first fact is wrong (Dennett, 1991) but science doesn’t solve conflicts by picking a preferred answer. It accepts both, so part 3 of this chapter asks how a decentralized brain can create one observer?

Next