Brain's function analog, digital

CU researchers develop computer models of the mind's inner workings

The brain was long considered an amorphous web with internal machinations too complex to understand. Then came the cognitive revolution in the middle of the past century, when the advent of digital computers convinced scientists that the brain worked like something sold by IBM.

Then in the mid-1980s, the tide turned. The brain, says University of Colorado psychology professor Randy O'Reilly, was thought of as an analog processor - "parallel, gooey networky type of thing that doesn't work at all like a digital computer."

O'Reilly, 39, specializes in modeling the biological basis of cognition. He describes himself as a product of the analog school, but he argues in a paper published in today's journal Science that the brain operates in ways at once both digital and analog. He says scientists trying to model the brain would be best served incorporating both schemes.

The analog-digital distinction is not trivial. O'Reilly and other brain researchers say knowing how the mind actually works - and then developing computer models capable of simulating it - could have profound impacts on treating cognitive problems or mental illness.

The prefrontal cortex, the gray matter behind the forehead, is the part O'Reilly says acts digital. By way of triggers including baths of the neurotransmitter dopamine and interaction with inner-brain structures called basal ganglia, parts of the prefrontal cortex can be held in an "on" or "off" state like a classic light switch.

An analog system, by comparison, acts like a dining-room dimmer switch, reacting to stimulus in infinitely subtle grades.

This black-or-white processing can come in handy. The prefrontal cortex "houses our internal 'executive' that decides what we want to do and keeps us focused on these goals," O'Reilly writes in the paper. Without it, our lives would be like "the crazy world of dreams," where one flails powerlessly from one random situation to the next.

The prefrontal cortex is switched off during dreams.

Analog processes, characterized by neurons firing in greater or lesser unison in the posterior cortex toward the back of the brain, have their role, too. O'Reilly says analog processing is the intuitive, "You know it when you see it" thinking. He gave the example of a funky object in a furniture store that one recognizes to be an expensive and probably uncomfortable chair.

A 2001 experiment involving monkeys and images of cats, dogs and digitally hybridized combinations of the two showed how the digital and analog processing regions may interact, said Todd Braver, a psychology professor at Washington University in St. Louis who has collaborated with O'Reilly.

A monkey pushed one button if it recognized a dog and another if it recognized a cat. When a monkey saw an image that was 80 percent dog, 20 percent cat, the neurons in the brain's analog regions showed graded signals, with more dog than cat. The digital prefrontal cortex cut to the chase and the monkey pressed "dog."

The aim of such work is to produce computer models sophisticated enough to simulate the brain's inner workings. Today, scientists don't understand the complex interactions of those systems, Braver said. Rather than considering symptoms and inferring what in the brain might be misbehaving, scientists could program in malfunctions to a simulated brain and see what symptoms emerge, Braver said.

Jonathan Cohen, a psychology professor and director of Princeton University's Neuroscience Institute, called O'Reilly's work "pace-setting and visionary" in the effort to form a mathematical basis for neuroscience and psychology, which has lagged compared to fields such as chemistry or meteorology.

"Advances in science come in precise computational and mathematical terms," Cohen said. "The brain is arguably the most complex device in the universe, and Randy's work is a hugely important and valuable endeavor."