Is the Brain Really a Computer? Comparisons That Do Real Work

Explaining something unfamiliar requires connecting it to something familiar, and in science the unfamiliar thing is frequently invisible, abstract or enormously complicated. A comparison imports a whole structure of relationships at once, so describing a cell as a factory conveys division of labour, inputs, outputs, assembly lines and waste disposal in a single move, which a list of facts would take far longer to deliver. That efficiency is why such comparisons appear throughout textbooks and research papers alike. They are not decoration added to finished understanding but part of how the understanding is built, and several have shaped entire research programmes for decades.

A handful of comparisons have done extraordinary amounts of work:

• •The heart as a pump, which reframed circulation as a mechanical problem
• •The brain as a telephone exchange, then as a computer, then as a network
• •The gene as information, a code, a blueprint and a programme
• •The atom as a solar system, which is wrong and still taught first
• •Natural selection as a designer, which biologists spend considerable effort disclaiming
• •The ecosystem as an economy, with producers, consumers and budgets

Every comparison carries over features that do not apply, and the trouble comes from the parts nobody notices being imported. Describing genes as a blueprint suggests a detailed plan of a finished structure, which they are not, since nothing in a genome corresponds to a picture of an organism and the relationship between a sequence and a body is nothing like the relationship between a drawing and a building. Treating the brain as a computer imports a separation between hardware and software, and between storage and processing, that the brain does not appear to respect. The atom compared to a solar system implies orbits that electrons do not have. In each case the comparison is useful at one level and generates confident wrong predictions at another.

Several of these have left the laboratory and now shape how the public argues about things, which is where the imported errors do the most damage. The selfish gene, introduced as a precise statement about the level at which selection acts, is routinely read as a claim that organisms are selfish, which is the opposite of what the argument says about cooperation. Survival of the fittest is read as a moral endorsement rather than a description, with a long and ugly history of political use. The brain as a computer supports confident claims about uploading minds. Chemical imbalance, used loosely to describe depression, implies a precision that the underlying science does not support and has been criticised heavily by the researchers whose work it supposedly summarises.

Comparisons change as technology changes, which is a revealing pattern in itself. The brain has been compared to whatever the most sophisticated available machine happened to be, having been a hydraulic system, a clockwork mechanism, a telegraph network, a telephone exchange, a digital computer and most recently a statistical learning system, with each generation confident the current one is not merely a comparison. Heredity was described in terms of blending fluids, then of particles, then of information, following the same pattern. The honest position is that a comparison is a tool with a working range, that stating what it does not cover is part of using it properly, and that the next one will look as dated as the hydraulic brain does now.

Comparisons import a whole structure of relationships at once, which is why explanation runs on them and why several have shaped entire fields. The blueprint comparison for genes and the computer comparison for brains both generate confident wrong predictions at the level where the imported features stop applying. The brain has been compared to whatever the most advanced machine of the day happened to be, and the current version will date the same way.