Dreaming is unlearning, Francis Crick told the National Academy of Sciences yesterday.

Explaining a controversial theory about dreams that is gaining popularity, Crick said that recent experiments with simple networks -- such as electrical grids in which each wire is linked to the others -- can demonstrate the mechanics and the purpose of dreaming.

Crick, who has a Nobel prize for his work deciphering the structure of deoxyribonucleic acid, now works on the biology and mathematics of brain function. He told the gathering yesterday that the brain can be understood only in terms of what groups of neurons can do together, as "neural nets" in his phrase.

Brain cells, called neurons, fire electrical signals. When and how they fire is determined by the ways in which they are stimulated -- for example by seeing, hearing and associated thinking.

Crick's idea is that so many associations are sparked by experiences -- each neuron connecting to so many others in the network -- that the system can become overloaded. Normally, an array of neurons learns "facts" by learning firing patterns that can be recalled as needed.

But each neuron can be associated with different memories and can trigger connections between one memory and another. The associations can proliferate and create hybrid memories, or fantasies and dreams. They are mixes of elements from real memories, Crick says.

The memories made during a day need to be separated from spurious associations, the mixing up of bits and pieces from different memories.

Dreaming achieves this separation, Crick says. He says that the events of the day -- and virtually all dreams comprise material brought up from the day's mental activity -- are replayed many times, and during the replaying the legitimate memories are strengthened while the random associations are weakened.

Crick calls it "reverse learning." He said he has two slogans to characterize the idea: "We dream in order to forget" and, slightly more accurately, "We dream in order to reduce fantasies and obsession," which he says are mixed, hallucinatory types of recall.

Crick said his theory, one of several now being offered to explain dreaming, is undoubtedly too simple to be completely valid, but he suggested it as a starting point.

Elsewhere at yesterday's meeting, Sheldon Segal of the Rockefeller Foundation said that the solution to overpopulation will come through science, when a cheap, simple and long-lasting contraceptive is devised.

He said several new contraceptives are in final research stages or on the market that should greatly boost the chances of contraception being used more widely throughout the world.

For example, in all but a few countries of sub-Saharan Africa more than 90 percent of couples use no contraception. The figure is about 30 percent in developed nations. In some developing nations, such as Thailand, effective government programs have increased contraceptive use from a small percentage to rates equivalent to those in developed nations.

Segal said that difficulties in distributing contraceptives and getting people to use them, coupled with the ineffectiveness of several methods, have contributed to the problem.

While oral contraceptives and intrauterine devices are 95 to 99 percent effective, contraceptives such as spermicides, diaphragms and condoms are only 65 to 80 percent effective. Natural family planning, which involves avoiding intercourse during ovulation, is about 50 percent effective, he said.

Segal has done research on the implantable contraceptive, which is now reaching the world market. It is a tube about 1 1/2 inches long and one-tenth of an inch wide that contains a hormone such as is present in birth control pills. This hormone is exuded slowly from the tube over five years or more, giving contraceptive protection.

Segal said tests have shown that the device is about 10 times as effective as the most effective contraceptives now on the market. Its main disadvantages are that it must be implanted and removed in a clinic, which can be a problem in a developing country, and that a substantial number of women who use it experience irregular menstrual bleeding.