The brain has become a digital computer; yet we are still trying to make our machines intelligent. Should those machines be modeled on the brain, given that our models of the brain are performed on such machines?

Rodney Brooks, Professor of Robotics, MIT

We humans are adept at finding metaphors and analogies to help us understand the unknown in terms what we already know, and we applaud those who are particularly gifted at this -- we call them poets and novelists. Not only are our brains good at finding metaphors for worldly things ("Shall I compare thee to a Summer's day?"), but they see themselves in metaphorical terms. The "Brains are like..." game goes back a long way, and tends to reflect on the most complicated technology known at the time. So for instance:

* The Egyptians compared the wrinkled surface of the brain with that of slag, the by-product of their new-found ore-extraction technology. (Back then, the brain was a secondary organ, and the heart was the seat of thinking and emotion. That's why Latin-based cardiac and credible, for instance, are cognate.)

* Roman brain metaphors followed along with their feats of hydraulic engineering. The great second century BCE physician Galen, for instance, posited that a pure "animal" fluid flowed from the liver to the brain, thence back throughout the nervous system, creating consciousness and rationality.

* Nearly 2,000 years later, Rene Descartes (1596-1650) codified Galen's idea by identifying the pineal gland ("seat of the soul") as the mediator and controller of this flow.

* The Industrial Revolution brought a slew of new metaphors involving new-fangled steam-driven mechanical systems. Around 1720, for instance, Gottfried Leibniz visualized the brain as a vast "mill" chock full of complex machinery. (As late as 1948, English brain pioneer Charles Sherrington -- he coined the word synapse -- pictured the brain as "an enchanted loom where millions of flashing shuttles weave a dissolving pattern ...")

* Telegraph and telephone networks provided whole new ways of conceiving the brain, which by the 1920s had become a complex metaphorical switchboard of inputs and outputs, patch cords, amplifiers and colored wires.

* Starting in the 1940s, the computer age has given us metaphors for the brain that endure today: central processing units, read-only and random-access memories, computation speeds, parallel processing, networks and cloud computing.

Not only is the "brain as computer" metaphor commonplace today, but its converse, "computer as brain," has inspired a generation of artificial intelligence gurus to attempt to model "aware computers" (think HAL and Lieutenant Commander Data) based on their understanding of the brain's hardware. I suspect that our gooey, soggy, cellular, pulsating brains are too quirky, too spontaneous, too self-aware to be simulated by machines made of silicon and copper. Hardware matters. Today's brain metaphors are surely closer than those of the Egyptians and Romans, but I'm not sure we can move forward on either track -- understanding brains or building intelligent computers -- until we move beyond the "brains are like computers are like brains ..." loop.

Barry Evans' ([email protected]) brain gets claustrophobic inside his skull, even as his Field Notes compendia wait patiently for impulse-buying brains at Eureka Books and Northtown Books.