The Invention of Memory: A New View of the Brain by Israel Rosenfield (Basic Books: $18.95; 224 pages)
Artificial intelligence has failed in its effort to make computers think like people because computers are too rigid. The real world is characterized by imprecision and novelty. No two situations are exactly alike.
Computers are bound by rules, which are never able to accommodate all of the possibilities that might come up. But human brains handle novelty with ease.
For example, no computer has yet been devised that can do a simple thing like recognizing faces. Yes, if you show a computer two identical pictures of a face, it can laboriously compare them pixel by pixel and conclude that they are the same person. But then show the machine a picture of the same face taken from a different angle, or the same face 10 years older, and the computer will not be able to recognize it. Most humans would have no trouble.
Historian and Physician
So the brain must operate differently than computers do. That much seems clear. What isn't so clear is how the brain manages to do all of the things that it does.
Israel Rosenfield is a historian and physician in New York who has written extensively on neuroscience. In "The Invention of Memory," he lays out a theory of the brain that differs radically from conventional wisdom on the subject and explains, en passant , why brains and computers behave so differently.
In Rosenfield's view, the brain is not a static system, like a computer, but a dynamic one. Individual thoughts, memories and ideas are not stored in specific places but emerge in context.
"There are no specific recollections in our brains," Rosenfield writes. "There are only the means for reorganizing past impressions, for giving the incoherent, dreamlike world of memory a concrete reality. Memories are not fixed but are constantly evolving generalizations--re-creations--of the past, that give us a sense of continuity, a sense of being, with a past, a present and a future. They are not discrete units that are linked up over time but a dynamically evolving system."
As a result, he writes, "Human intelligence is not just knowing more, but reworking, recategorizing, and thus generalizing information in new and surprising ways."
In support of this view, Rosenfield reviews many of the 19th-Century case studies on damaged brains that led to the conclusion that the brain is like a filing cabinet with individual memories stores in specific locations. The idea was that the loss of specific memories or abilities resulted from damage to the sites where they resided.
Rosenfield argues instead that brain damage causes a loss in organizing power, which can affect the ability to speak or read in quirky ways. It's not that specific words are lost, he says. Rather it's the ability to organize words and use them in the proper way that is affected.
In putting forth his view of how the brain works, Rosenfield draws heavily on the work of Gerald M. Edelman, the Nobel laureate who is director of the Neurosciences Institute at the Rockefeller University in New York. A large chunk of Rosenfield's book is devoted to Edelman, who also argues that the brain operates as a unified whole rather than as a collection of individual duchies.
'Mapping Other Maps'
In Edelman's view, the brain's neurons are grouped into maps, which organize information in useful and constantly changing ways. "The brain has many different kinds of maps and ways of mapping other maps," Rosenfield says. "The purpose of the maps is to create perceptual categorizations that will permit the animal to act in appropriate ways."
This theory also provides an explanation for dreaming: "We dream when the maps are released from the constraints of sensory order during sleep."
I confess that I do not understand all of this very well (and apparently am not alone, judging by some researchers' reactions to Edelman's theories). But I find the ideas appealing and intriguing nonetheless.
Conventional theories of the brain are too limited. They focus on the wrong part. The brain's power stems not from information but from its ability to put information together in new ways. Life is less a process of being than of becoming.
Though murky in parts, Rosenfield's book is a forceful presentation of these ideas, which deserve a large audience and serious consideration. Where one goes with them is anybody's guess. But from what has been written so far, this theory seems fruitful and promising.