This paper reviews seven principles of memory function that set limits on the degree to which any attempt to recover a long-forgotten memory can succeed: encoding, organization, time-dependency, cue-dependency, encoding specificity, schematic processing, and reconstruction. In the absence of independent corroboration, there is no "litmus test" that can reliably distinguish true from false memories, or memories that are based on perception from those that are based on imagination. Practicing clinicians should use great caution when using hypnosis or any other technique to facilitate delayed recall.
Hypnosis, Delayed Recall, and the Principles of Memory
Clinical interest in memory processes has been revived by the claim that a whole host of problems, including anxiety, depression, and eating disorders, have their origins in childhood experiences of incest and other sexual trauma, abuse, neglect, and deprivation -- memories of which have been repressed by the patient. Therefore, many therapists seek to recover these memories, and bring them into conscious awareness, so that the patient can deal with them more adaptively. Sometimes, hypnosis plays a role in this process. However, the very nature of memory places strict limitations on the degree to which any attempt to recover memories from the distant past will succeed. In this paper, I wish to discuss some of these principles and their clinical implications (for fuller explication of these principles, and documentation of their sources in the experimental literature, see Kihlstrom & Barnhardt, 1993; for a further discussion of ostensibly repressed memory, see Kihlstrom, 1993).
Seven Principles of Remembering
In analyzing the causes of remembering and forgetting, it is common to characterize memory processing as consisting of three stages (Crowder, 1976): encoding, the process by which a trace of current experience is laid down in memory; storage, by which an encoded memory trace remains available over time; and retrieval, the process by which information is recovered from storage and used in ongoing cognitive activity. Logically, a memory cannot be retrieved from storage unless it was encoded in the first place, or if it has been lost from storage. But even the encoding of a memory, and its availability in storage, does not guarantee that a veridical memory of the original event will be retrieved. In order to understand how this is so, it is important to consider the manner in which memory operates.
The encoding of memory appears to be governed by two principles, elaboration (Anderson & Reder, 1979; Craik & Tulving, 1975) and organization (Bower, 1970; Mandler, 1967):
The Elaboration Principle: The memorability of an event increases when that event is related to pre-existing knowledge at the time of encoding.
The Organization Principle: The memorability of an event increases when that event is related to other events at the time of encoding.
The point here is that events can be remembered to the extent that they were encoded at the time they occurred, and that proper encoding does not occur automatically, but rather requires active, cognitive effort. This fact accounts, in large part, for the familiar phenomena of infantile and childhood amnesia: most adults remember so little of childhood because children lack both the knowledge base and the information-processing capacity to encode such memories in retrievable form. There are also other factors involved, such as whether the event in question is the topic of conversation between the child and other people; see Fivush & Hudson, 1990; Nelson, 1993); of course, such social interactions are likely to facilitate elaborative and organizational processes. Elaboration and organization, are highly relevant in the case of ostensibly repressed memories, most of which appear to be for events occurring in early childhood, and cover events which by their very nature are not shared with other people.
Assuming that a memory has been encoded, it remains available for subsequent retrieval and use -- at least in principle. But in practice, memories seem to fade over time, an observation that has been enshrined in another principle (Ebbinghaus, 1885):
The Time-Dependency Principle: The memorability of an event declines as the length of the storage interval (i.e., between encoding and retrieval) increases.
Of course, there are instances in which knowledge appears to be preserved in rich detail over long periods of time, as in so-called "flashbulb" memories for emotionally arousing events (Winograd & Neisser, 1992), and the "permastore" of factual knowledge (Bahrick, 1984). But these are the exceptions that test the rule, and in any event careful studies have shown that many flashbulb memories are highly inaccurate, and the notion of a permastore generally refers to factual knowledge rather than memory for particular events. Again, the principle of time-dependency is clearly relevant when we are talking about the adult recollection of childhood events.
The experimental literature does contain a number of studies of so-called hypermnesia, in which memory appears to grow rather than fade with time (Erdelyi, 1984; Kihlstrom & Barnhardt, 1993; Payne, 1987). This phenomenon is potentially relevant to the problem of repressed memories, but does not contradict the principles governing encoding and storage. Hypermnesia occurs when items were subject to elaborative and organizational activity at the time of encoding, and most hypermnesia is accomplished on the first few attempts at retrieval, relatively soon after encoding has taken place.
The retrieval process itself is governed by a large set of principles. One of these relates to the fact that, by and large, recognition tests produce more memory than recall tests, and cued recall produces more memory than free recall (Tulving, 1974):
The Cue-Dependency Principle: The memorability of an event increases with the amount of information supplied by the retrieval cue.
Remembering usually begins with some kind of cue that provides some information about the event which is to be remembered. Cues that are highly informative are more likely to contact available memory traces than those that are not. In some respects, encoding and retrieval are in a complementary relationship: access to well encoded memories generally requires fewer retrieval cues, and lots of retrieval cues can gain access even to very poorly encoded memories. But the success of recognition testing comes with a price: false alarms. Subjects may remember more correct information on a recognition test than on a recall test, but they also remember more incorrect information. In the laboratory, we can tell which is which. But in real life, corroborating evidence may be entirely lacking -- which is why the problem of false identification in police lineups is so difficult (Malpass & Devine, 1984).
Cues are important, but they must also supply the right kind of information, not just the right amount (Tulving & Thomson, 1973):
The Encoding Specificity Principle: The memorability of an event increases when the information processed at the time of retrieval was also processed at the time of encoding.
The manner in which an event is encoded -- the meaning of the event, how it is perceived and interpreted and categorized -- determines which retrieval cues will be successful in gaining access to that event. Encoding specificity appears to underlie the phenomenon of state-dependent memory -- whether "state" is defined in physiological, emotional, or environmental terms. It is certainly relevant in the case of ostensibly repressed memories. For example, memories that have been processed during states of high emotional arousal may only be accessible if the same affect is reinstated at the time of retrieval (Bower, 1981; Eich & Metcalfe, 1989).
Moreover, it turns out that memory is determined by the degree to which an event conforms to our expectations and beliefs (Hastie, 1981):
The Schematic Processing Principle: The memorability of an event increases when that event is relevant to expectations and beliefs about that event.
The general principle is straightforward enough, but the details may be a little surprising. If memory is plotted as a function of the degree to which the target events can be predicted on the basis of pre-existing knowledge, represented in the form of organized knowledge structures known as schemata, it turns out that events that are highly congruent with expectations are highly memorable; but events that are highly incongruent with active schemata are even more memorable. The U-shaped function apparently reflects the operation of two different principles: events that are inconsistent with preexisting schemata are surprising and draw more attention, and receive more elaborative and organizational activity at the time of encoding; and at the time of retrieval, the schema provides additional cue information that can facilitate access to relevant memories. Events that are irrelevant to the schema get neither advantage, and so are poorly remembered.
The role of cognitive schemata is also underscored by another principle (Bartlett, 1932):
The Reconstruction Principle: The memory of an event reflects a blend of information retrieved from specific traces encoded at the time of that event with knowledge, expectations, and beliefs derived from other sources.
In describing how memory works, we often resort to the metaphor of a library: memory traces are books that must be purchased and catalogued; the prospective user must look up the book in the catalog in order to know where to find it; and in order for the search to succeed, the book must not have been eaten by worms, or displaced by a careless user. The library metaphor will take us a long way, but the notion of memory retrieval obscures the fact that memories can be distorted, biased, and otherwise changed by changes in perspective and other events that occur after the time of encoding. In the final analysis, memory is not so much like reading a book as it is like writing one from fragmentary notes. The reconstruction principle is of utmost importance in the present context, because it means that any particular memory is only partly derived from trace information encoded at the time of the event: in the process of remembering, trace information combined with knowledge, beliefs, and inferences derived from other sources.
Is there any way in which we can determine the source of the information contained in memory? And, in the absence of independent corroboration, is there any way in which we can determine which memories are valid representations of past experience and which are not? Some data in this respect come from studies on source memory and on reality monitoring. Source memory refers to the person's ability to remember the circumstances under which he or she acquired certain items of information. Anecdotal evidence strongly indicates that source memory is not perfect: we have all wondered whether something really happened, or whether we dreamed or imagined it; we have all confused what we did with what we intended to do, or wish that we had done. Sometimes we don't know whether we saw a movie in the cinema or on television. And we have all forgotten who told us a particular joke, or let us in on a particularly juicy rumor.
Johnson and her colleagues have studied this problem extensively. In a representative experiment (Johnson & Raye, 1981), subjects viewed a set of words and pictures representing familiar objects. For half the words, the subjects are asked to count the number of vowels; for the other half, they are asked how long it would take to draw a picture of the object represented by the word. Similar questions were asked about the pictures: how long it would take to draw, and how many vowels are in the word corresponding to the picture. Later, the subjects were read the words, and asked whether they saw the item as a picture or as a word. Johnson observed a number of confusions, especially for those words for which the subjects made a visual judgment, and those pictures for which they made a lexical one.
Do we have any way of telling which memories are products of perception, and which are the products of imagination? Johnson and her colleagues have proposed a number of attributes which tend to distinguish between the two types of memories (Johnson, Foley, Suengas, & Ray, 1988). For example, memories of percepts contain more sensory detail, are less complex, more realistic, and more likely to contain information about the spatial and temporal context in which the remembered event occurred. But it should be understood that these distinctions are entirely probabilistic: that is, the presence of one or more of these attributes only increases the likelihood that the memory is a product of real experience, rather than of imagination. Moreover, the list of attributes distinguishing between memories of perceived and imagined events from childhood was very short. Thus, there is no litmus test that is 100% reliable, or even close to it.
In another line of research, Raskin, Steller, and their colleagues have been promoting a technique called criteria-based statement analysis (Raskin & Steller, 1989; Steller & Koehnken, 1989, originally developed in Germany by Undeutsch (1984). These investigators have proposed a number of criteria by which the veracity of a witness's statement might be assessed. For example, they assert that factual statements contain a large number of details, especially those that are unusual or superfluous, make reference to unexpected complications, portray the witness's own mental state, and spontaneous corrections, admissions of lack of memory, and expressions of self-doubt. Many of these criteria make intuitive sense, and some of them overlap with those developed by Johnson; but it should be said that the empirical basis of these criteria is rather thin at present. Undeutsch and his colleagues have found that attributes such as these make statements appear credible, but their surface credibility is uninformative as to their actual truth-value. There is little or no evidence that these characteristics actually distinguish between accurate and inaccurate memories.
In the final analysis, then, there do not appear to be any internal criteria -- that is, standards that can be applied to the statements themselves -- that can serve to reliably distinguish between accurate recollections and fabrications and confabulations. Nothing substitutes for external criteria -- that is, the verification of individual statements by objective evidence. In the absence of such independent corroboration, we have no means of reliably distinguishing between fact and fantasy.
Implications for Clinical Practice
The success of any attempt to exhume a memory will depend on why that memory was forgotten. Of the all the principles of remembering and forgetting summarized above, perhaps the broadest is the encoding specificity principle (Tulving & Thomson, 1973), which may be restated as follows: what is encoded depends on what was perceived, and what is encoded determines what subsequently will be retrieved. A variety of factors may conspire against elaborative and organizational processing, but the consequence is always the same: if an event is poorly encoded in the first instance, the prospects for its subsequent retrieval and conscious recollection, especially after a long retention interval, are very poor as well.
To some extent, an extremely rich retrieval environment may compensate for poor encoding received by a memory. However, there are limits to this compensation: even recognition testing, in which the person is cued by a copy of the event in question, often fails. Moreover, attempts to enrich the retrieval environment by providing extra cues increase the risk that the subject will incorporate that information into his or her memory, or at least become confused about whether the source of particular details is in the trace or the query. The result may well be a distortion of memory. In the absence of independent corroboration (which is so often the case in delayed recall), such distortions are arguably worse than no memory at all, because neither the subject nor the questionner know where truth lies.
Nothing about hypnosis changes the rules by which memories are encoded, stored, and retrieved; accordingly, these rules constrain the utility of hypnosis (or any other procedure) as a technique for the enhancement of memory. There is no scientific evidence that hypnosis can improve the retrieval of poorly encoded memories, or those that have been degraded by interference accumulated over long retention intervals. It is possible, at least in principle, that hypnosis could be used to provide an extremely rich set of retrieval cues, overcoming encoding and storage conditions. For example, age regression could be used to structure an hallucinated environment that returns the person to the circumstances, both environmental and emotional, of the forgotten episode: if accurate, this would be the richest retrieval cue imaginable.
Unfortunately, the clinical and forensic circumstances in which hypnosis is used to enhance recall are those in which such information is rarely available. In any event, the suggestions used to reconstruct such a mental representation of the past may also promote confabulation, and in such a manner that neither the therapist nor the patient will be able to determine, with accuracy, whether the recollection is accurate. For this reason, hypnosis should be used, if indeed it is to be used at all in forensic and therapeutic situations, only in the generation of hypotheses to be corroborated by independent evidence. Hypnotically refreshed memories should never be allowed to stand alone.
Paper prepared for a special issue of the International Journal of Clinical & Experimental Hypnosis on "Hypnosis and Delayed Recall". The point of view taken in this paper is based on research supported in part by Grant #MH-35856 from the National Institute of Mental Health. I thank Terrence Barnhardt, Lawrence Couture, Jennifer Dorfman, Elizabeth Glisky, Martha Glisky, Victor Shames, Michael Valdisseri, Susan Valdisseri, and James Wood for their comments.
Correspondence should be addressed to: John F. Kihlstrom, Amnesia & Cognition Unit, Department of Psychology, University of Arizona, Tucson, Arizona 85721. Internet: email@example.com. Bitnet: kihlstrm@arizvms.
Anderson, J.R., & Reder, L.M. (1979). An elaborative processing explanation of depth of processing. In L.S. Cermak & F.I.M. Craik (Eds.), Levels of processing in human memory (pp. 385-403).
Bahrick, H.P. (1984). Semantic memory content in permastore: 50 years of memory for Spanish learned in school. Journal of Experimental Psychology: General, 113, 1-29.
Bartlett, F.C. (1932). Remembering: A study in experimental and social psychology. Cambridge: Cambridge University Press.
Bower, G.H. (1970). Organizational factors in memory. Cognitive Psychology, 1, 18-46.
Bower, G.H. (1981). Mood and memory. American Psychologist, 36, 129-148.
Craik, F.I.M., & Tulving, R.S. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology: General, 11, 268-294.
Crowder. (1976). Principles of learning and memory. Hillsdale, N.J.: Erlbaum.
Ebbinghaus, H. (1885). Memory: A contribution to experimental psychology (H.A. Ruger & C.E. Bussenues, trans.). New York: Teachers College, Columbia University. Translation published, 1913.
Eich, E., & Metcalfe, J. (1989). Mood dependent memory for internal versus external events. Journal of Experimental Psychology: Learning, Memory, & Cognition, 15, 443-455.
Erdelyi, M.H. (1984). The recovery of unconscious (inaccessible) memories: Laboratory studies of hypermnesia. In G.H. Bower (Ed.), The psychology of learning and motivation (Vol. 18, pp. 95-127). New York: Academic.
Fivush, R., & Hudson, J.A. (Eds.). (1980). Knowing and remembering in young children. Emory Symposia in Cognition, Vol. 3. New York: Cambridge University Press.
Hastie, R. (1981). Schematic principles in human memory. In E.T. Higgins, C.P. Herman, & M.P. Zanna (Eds.), Social cognition: The Ontario Symposium (Vol. 1, pp. 39-88). Hillsdale, N.J.: Erlbaum.
Howe, M.L., & Courage, M.L. (1993). On resolving the enigma of infantile amnesia. Psychological Review, 113, 305-326.
Johnson, M.K., & Raye, C.L. Reality monitoring. Psychological Review, 88, 67-85.
Johnson, M., Foley, M.A., Suengas, A.G., & Raye, C.L. (1988). Phenomenal characteristics for perceived and imagined autobiographical events. Journal of Experimental Psychology: General, 117, 371-376.
Kihlstrom, J.F. (1993, April). The recovery of memory in laboratory and clinic. Invited address presented at the joint annual meeting of the Rocky Mountain Psychological Association and Western Psychological Association, Phoenix.
Kihlstrom, J.F., & Barnhardt, T.M. (1993). The self-regulation of memory: For better and for worse, with and without hypnosis. In D.M. Wegner & J.W. Pennebaker (Eds.), Handbook of mental control (pp. 88-125). Englewood Cliffs, N.J.: Prentice-Hall.
Malpass, R.S., & Devine, P.G. (1984). Research on suggestion in lineups and photospreads. In G.L. Wells & E.F. Loftus (Ed.), Eyewitness testimony: Psychological perspectives (pp. 64-91). Cambridge: Cambridge University Press.
Mandler, G. (1967). Organization and memory. In K.W. Spence & J.T. Spence (Eds.), The psychology of learning and motivation, Vol. 1 (pp. 327-372). New York: Academic Press.
Nelson, K. (1993). The psychological and social origins of autobiographical memory. Psychological Science, 4, 7-14.
Payne, D.G. (1987). Hypermnesia and reminiscence in recall: A historical and empirical review. Psychological Bulletin, 101, 5-27.
Raskin, D.C., & Steller, M. (1989). Assessing credibility of allegations of child sexual abuse: Polygraph examinations and statement analysis. In Wegener, H., Losel, F., & Haisch, J. (Eds.), Criminal behavior and the justice system: Psychological perspectives (pp 290-302). New York: Springer-Verlag.
Steller, M., & Koehnken, G. (1989). Criteria-based statement analysis. In d.C. Raskin (Ed.), Psychological methods in criminal investigation and evidence (pp. 217-245). New York: Springer.
Tulving, E. (1974). Cue-dependent forgetting. American Scientist, 62, 74-82.
Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80, 352-373.
Undeutsch, U. (1984). Courtroom evaluation of eyewitness testimony. International Review of Applied Psychology, 13, 51-67.
Winograd, E., & Neisser, U. (1992). Flashbulb memories. Emory Symposia in Cognition, Vol. 4. Cambridge: Cambridge University Press.