This is a set of seven "true stories" by an American professor of psychiatry who treats children who have suffered trauma, and who has figured as an "expert witness" in some notable court cases in America. We are promised that the "stories illustrate how we forget childhood trauma and . . . how and why these memories return. They also illustrate exactly what goes wrong with memory and what parts of a memory sometimes turn false." Do they?

Most space is devoted to the first account, of the supposedly sudden return of a memory of an adult daughter about her father, George Franklin, of whom she had a vivid image of his murdering her close childhood friend some 20 years earlier. Until that time, there had been no such memory. The author was an expert witness at the trial, advancing the credibility of the return of a repressed memory. Largely on such evidence George Franklin was sentenced to life imprisonment for murder. All the other stories, in one way or another, illustrate the supposed return of fragments or whole episodes of which the subject's were said not to have been fully conscious until they resurfaced. Of the seven pieces, one is a false story of a child who was pressured by her mother into making persistent accusations against her two therapists. It is hinted that the child may just have been a good actress. Another is about the claim by a former Miss America that as a child she had been repeatedly sexually abused by her socialite father. The weakest account is of a man struggling to remember something of his brother before he died some 30 years earlier when he was four years old.

All these make engaging and easy reading. But Lenore Terr's object is not merely to tell stories, but to use them as props on which to hang two kinds of fabric. One of these is the early Freudian notion of repression -- that memories of traumatic events can be actively held out of consciousness for years, even decades, and then suddenly resurface. Dissociation is a related form of escape from the mnemonic pain of trauma. The other fabric is of the "scientific basis" of memory research.

Regrettably, both fabrics are ragged. The most dramatic evidence, the case of George Franklin and his daughter, has had a very deep influence on the "recovered memory" debate. The case has been analysed by others, and especially in some detail in a recent book by Richard Ofshe and Ethan Watters (Making Monsters, 1994). It would seem that Terr is economical with the facts and also gets some of them wrong. She writes as though Franklin's daughter, Eileen, "suddenly" saw a "picture of her childhood friend in a state of terror, with her father gripping a rock held high above his head and targeted on the child." She does not dwell on the fact that Eileen had been under the care of two therapists who testified that they taught Eileen about such theories, and according to one of them, Eileen's memories gradually developed over the course of the sessions. One need not doubt the sincerity of Eileen's belief in her image; the question is where the belief came from, and whether it was historically true. Unfortunately, similar hints of important background events not fully exposed, or fully explored, run through some of the other stories. Corroboration of historical truth is not given high priority. To be fair, there is the occasional acknowledgment that memories might be "falsely implanted", but Terr asserts, without further evidence, that "many more memories lend themselves to confirmation than one might think".

ÁñÁ«ÊÓƵ

As for the science of memory, there are frequent allusions to contemporary research -- on retrograde amnesia, on cyclic AMP, on learning by fruit flies and sea snails, on "spatial memories being handled by relatively simple presynaptic and postsynaptic connections in the hippocampus", on "habit memory" being handled by the monkey's basal ganglia, but also "memory" being located all over the brain", on infantile amnesia, on implicit and explicit memory, on self-recall by Marigold Linton. If the recital were not quite so transparent, we might be blinded by it. The difficulty, alas, is that precious little of it has any bearing whatsoever on Terr's claims. There is nothing about cyclic AMP, or any of the other assembled evidence that tells one anything at all about normal forgetting, let alone why it might support anything as controversial as repression. Virtually all of the cited research relates to mechanisms of storage of new information, and not to its fate or its availability once stored.

An important question is how repression, which is a theoretical concept, relates to forgetting, which is a measurable fact. Here the semantics are largely metaphorical, by example rather than by tight definition. If you remember your grandmother's brooch in the bottom of your drawer after a long interval when it was "out of mind", that was forgetting. If, on the other hand, your grandmother died a bloody death with the brooch on, your putting it out of mind was repression.

ÁñÁ«ÊÓƵ

Some of the scientific evidence is simply not understood by Terr. To cite just two examples: from research on the retrograde amnesia seen in amnesic patients, she states that "old memories apparently 'reorganise' themselves many years after they are formed". The evidence supports nothing of the sort. Another surprising assertion: "When rehearsal is prevented in the laboratory, most memory is lost in about 30 seconds." This applies, if it applies at all, to a limited domain like telephone numbers, not "most memory". In any event, even if were true it does not illuminate her own material. Her account of infantile amnesia is too undeveloped and unbalanced.

There is a large body of experimental evidence on the malleability of memory over time, especially in the classical work of Sir Frederick Bartlett and in more recent studies by researchers like Elizabeth Loftus. These demonstrate clearly that memory is not like a hard nugget to be winkled out in pristine form; it is a constructive creation. Terr argues such studies are irrelevant because repression and dissociation are especially to be found with memories of very traumatic events. But, strangely, even her own earlier research on the traumatised children in the "Chowchilla kidnapping" contradicts this. To try to accommodate a mixture of evidence she advances a new hypothesis, that there are two types of trauma, Type I, which is of single traumatic events, and Type II, multiple traumas. The latter but not the former are said to be prone to repression, indeed, impinging immediately upon the advent of each new trauma itself. Even here there has to be special pleading: thus Eileen Franklin, who allegedly forgot the single trauma of the murder scene is Type II because she was repeatedly traumatised by her father and hence forgot the murder scene for 20 years. But we are also told in contradiction that "she always remembered . . . that her father was an unpredictably violent alcoholic -- this she had not forgotten. He threw her and his four other children against walls. He punched her and hit much too hard." In fact, the evidence for these two categories, Type I and Type II, in relation to memory loss is extremely weak; it is candy-floss theorising. The claim is ad hoc, not even post hoc.

Terr is semantically on better grounds regarding dissociation -- the idea that trauma leads to a shift of focus of memory and attention to other less painful thoughts. This is not just "not thinking about unpleasant things". Dissociation is argued to be a deep cleavage of the personality and of cognitive systems, the extreme forms being multiple personalities or the separation between the "night person" and the "day person" of the former Miss America. Terr refers, accurately, to research on "multiple memory" systems, deriving mainly from findings with brain-damaged amnesic patients in whom there is evidence of "implicit memory" -- retention of learned information with no explicit recognition of "remembering" the information. But all of Terr's examples are of failures of explicit memory and its alleged eventual recovery from repression or dissociation. I know of no evidence that tells of a switch of implicitly stored information into a viable explicit memory, and that is what Terr requires.

All this makes engaging "human interest" reading -- until one reflects on it and places it in the context of what is taken to be "expert testimony". Lives turn on such testimony, embroidered by supposed authority dogmatically delivered with weak or non-existent scientific evidence and without corroboration of historical truth. George Franklin is serving a life sentence for murder, for which there is no evidence beyond the "recovered memory". He might have been a pretty unpleasant father, but there is not a tinge of doubt expressed by Terr that there might have been a dreadful miscarriage of justice. But the danger is not only in the courtroom; just as worrying is what takes place in the therapist's consulting room. If clients are led to believe that memory is like an accurate tape recording, and that all that is needed is to find the right button with which to play it back, then they themselves are being deeply misled. In both the therapeutic and legal contexts, memory is too complex and challenging a subject to be analysed with such dogmatic and doctrinaire glibness.

ÁñÁ«ÊÓƵ

Lawrence Weiskrantz is emeritus professor of psychology, University of Oxford.