You Must Remember This

NEITHER ROSS CHEIT NOR EILEEN Franklin remembered the horror until years later. For Cheit, the trigger was a 1992 dream, a nighttime reverie that blossomed into a conscious recollection of a camp counselor molesting him in the summer of 1968. He had not thought about, or even remembered, the incident for decades. But the ex-counselor, confronted with Cheit's accusation, admitted the abuse, and witnesses supported Cheit's recollection. For Franklin, the key was a glimpse of her daughter, in 1989, in a certain pose. Her childhood friend Susan, she remembered, had been in just that posture when she was brutally murdered in 1969. All of a sudden, Eileen claimed, she had a flashback: she saw her father killing 9-year-old Susan, a crime that had remained unsolved. In 1990 George Franklin was convicted of first-degree murder, the first time a ""recovered memory'' had ever been the basis of a criminal prosecution. But gradually evidence trickled in that it was therapy, and allegedly hypnosis, that created Eileen's memory, which remained uncorroborated. In 1995 Franklin's conviction was overturned. And last week California declined to retry him.

Is there any way of distinguishing an accurate memory from a mistaken one? That is what Harvard University psychologist Daniel L. Schacter asks in his new book ""Searching for Memory'' (398 pages. Basic Books. $27). ""The answer, unfortunately, is no,'' he concludes. Which is not to say that scientists are not trying. The last two years have seen an explosion of research in how traumatic memories are stored, lost and sometimes found and in how false memories can form. In the latest experiment, researchers led by Schacter report the first images ever captured of the brain remembering something it heard and then ""remembering'' something it only thought it heard. The study, to be published in the August issue of Neuron, ""gives us a model of how the brain produces accurate and false memories,'' says coauthor Dr. Eric Reiman of the University of Arizona.

In the experiment, at Good Samaritan Regional Medical Center in Phoenix, a dozen female volunteers heard lists of 20 semantically related words (chart). Then the women were shown other words, on a computer screen, and asked which were on the list. The volunteers had many false memories: 58 percent of the test words they ""remembered'' had not been on the list. Most of the false hits were clever decoys, words related to those on the list.

To find out what the volunteers' brains were up to, the scientists PET-scanned them. PET scans measure changes in blood flow, and blood flow indicates neural activity. In both true and false recognition, parts of the brain around the hippocampus (a seahorse-shaped structure involved in memory) became active. Apparently, false memories seem so real because they tap the same neural structures as accurate memories. But true memories also lit up areas behind the left ear, in the superior temporal lobe; this region processes information about sounds of recently heard words. In illusory memories, it stayed quiet. ""Accurate memories appear to retrieve sensory details that are not available to us in illusory memories,'' says Reiman. Finally, false memories were marked by more fireworks in the frontal cortex and cerebellum, which struggle to recall the context of an event.

PET scans are not likely to emerge as lie detectors in courtroom battles over recovered memory, however. For starters, the brain differences picked up by PET described the 12 volunteers in aggregate, but ""might not hold up in any one case,'' cautions Schacter. Also, false memories may be just as likely as true ones to ignite the brain's sensory apparatus. Perhaps more so: when someone imagines a pseudo-event over and over, she often implants sensory data about it in the mind. She can actually see or hear or feel an event that never occurred.

Memory, so fragile and easily fooled, is clearly much stranger than psychologists thought when, just two years ago, some dogmatically asserted that traumatic memories are never lost and recovered memories are always fabricated. Now a more balanced viewpoint is emerging. Some victims of childhood abuse do forget single episodes and perhaps multiple incidents. Last year J. Douglas Bremner of the West Haven (Conn.) Veterans Administration Hospital found a small reduction in the size of the memory-forming hippocampus in 17 survivors of childhood trauma. But there is little or no evidence of indefinitely forgetting years of abuse. There are also a few cases of recovered, true memory. But hundreds of people have retracted claims of recovered memory. The memory wars rage -- more than 800 civil and criminal cases based on recovered memory have been filed -- and it will be a long time before science imposes a ceasefire.

ORIGINAL LIST

1. Thread

2. Pin

3. Eye

4. Injection

5. Syringe

6. Sewing

7. Sharp

8. Point

9. Hurt

10. Knitting

11. Prick

12. Thimble

13. Haystack

14. Pain

When asked to identify works on the original list,many volunteers incorrectly picked decoy words like needle

TEST LIST

1. Sugar

2. Needle

3. Thread

4. Sweet