It is the stuff of movies -- being able to erase a bad memory while keeping others intact. But members of a team that includes a Medical College of Georgia researcher say they've been able to do just that in specially bred mice.

MCG scientist Joe Z. Tsien cautions, however, that it won't work in people and that clinical treatments are a long way off.

In a study published today in the journal Neuron , Dr. Tsien and collaborators at East China Normal University in Shanghai were able to selectively take out key components of certain memories. They used mice specially bred to have high levels of a mutated enzyme called alpha-CaMKinase II. Those mice perform poorly on memory tests.

But when they're given a chemical that inhibits that enzyme, normal memory function is restored. As researchers manipulated levels of the enzyme at specific times during the memory testing, the mice were able to experience an event normally, to learn, but when they tried to recall it with elevated enzyme levels, the memory failed. Specific aspects could not be recalled later, even when enzyme levels were kept low to restore normal memory functions.

For example, the mice might be put in a chamber where a low tone is played and then they are given a mild foot shock. This test creates two types of memory -- a contextual memory tied to a place or location and a cued memory tied to the sound. When returned a day later or a month later, normal mice would show a fear response to the chamber and to the tone. But the specially bred mice who had experienced either the chamber or the tone separately in an altered state wouldn't respond to that part of the scenario, even after normal memory function returned. But they would still recognize the other parts: Those who now don't remember the tone still react to the chamber; and those that don't recognize the chamber react to the tone when they hear it in another room.

"It was very surprising actually," said Dr. Tsien, who heads up the Brain and Behavioral Discovery Institute at MCG. "You can actually erase one kind of memory while preserving the other. These are the same kind of memory but different aspects. You can go in there and actually selectively erase them."

It is theorized that learning increases the linking between certain sets of neurons and retrieving that memory strengthens those bonds, Dr. Tsien said. But when the enzyme is elevated in the transgenic mice during initial recall, the opposite seems to happen, he said.

"Somehow, under the overexpression condition, these connections are weakened by the initiation of the recall of the memory," he said.

Seth D. Norrholm, a fear and anxiety researcher at Emory University, cautioned that it is very difficult to determine whether a memory has truly been erased.

"When you use a term like erasure, it is possible that the memory trace is there at some threshold level that can't be retrieved," said Dr. Norrholm, who was not involved in the research.

Still, "I thought it was a really elegant study," he said. "I would just interpret the results with caution."

Dr. Tsien also urges caution, noting it is not the real-life equivalent of the movie Eternal Sunshine of the Spotless Mind, in which a couple have memories of each other erased.

"It's fun when people joke about that stuff, but I don't think it's possible given the circumstances," Dr. Tsien said. "It's not really possible to achieve this in humans because the memory is such a complex problem. We are barely at the foot of the mountain here."

What the research does do is give them possible pathways to look for other proteins and enzymes that might be affected and might be better targets for drugs, he said.