Retrieval of Memories:
A Psychobiological Approach
Norman E. Spear
State University of New York, Binghamton
I. INTRODUCTION
Personal experience has familiarized us with what happens to a memory after it has been acquired (that is, âstoredâ)âretention or forgetting may occur and, at any given moment, retrieval of the memory may be easy, hard, or impossible. The terms âretention,â âforgetting,â âretrieval,â âmemory,â and related labels are defined later, when the reader may discover that these familiar terms are used here in a manner slightly different from that expected.
Why do acquired memories sometimes control behavior dramatically and yet in other cases seem to be quite ineffective as if they had not really been established originally? The famous Russian neuropsychologist, Luria, became interested in the psychological characteristics of a young newspaper reporter, Mr. S., who reportedly forgot little or none of what he learned (Luria, 1968). A series of tests soon brought Luria to the astounding conclusion that the capacity of Mr. S. for remembering events could not be quantified because it apparently had no limit! Luria found Mr. S. to be perfectly accurate in recalling, for example, the precise location and identity of 20 randomly chosen numbers in matrix form that had been presented to him once 16 years earlier.
In contrast, consider the case of another young man, H. M., who after ten years of suffering from epilepsy agreed to have the temporal portion of both sides of his brain removed in an attempt to alleviate his uncontrollable seizures. Following the operation, H. M. had an IQ not greatly different from that of Mr. S., was in fine health, and could carry on a quite normal conversation. H. M. could recall accurately events a few seconds after they occurred, but thereafter his memorial representations seemed to evaporate:
⌠after his operation, his family moved into another house on the same street; he could never remember the new address nor his return to the old house. His parents moved again a few years later, and, although he seemed to know that he had moved, he could not remember the address. He could not remember where things belonged; he mowed the lawn regularly but had to ask his mother each time where the mower was kept. He did the same puzzles day after day and reread the same magazines. He did not recognize nor know the names of his neighbors, and invariably treated them as strangers [Barbizet, 1970, p. 60].
The cases of Mr. S. and H. M. are patently abnormal, yet, there are indications of nearly as wide variations in forgetting and retention in normal persons. Perhaps the most dramatic instance of severe forgetting that we all experience is in regard to specific episodes from early childhood. While this forgetting is difficult to specify systematically, especially by the experimental method, it is clear that the extent of forgetting for events that occur between birth and approximately three years of age is greater than that for any other comparable period in a humanâs life. Similar inflated forgetting has been found for immature animals as compared to mature animals (see Campbell & Spear, 1972). However, there are indications that some memories acquired in early childhood may be retrieved under certain circumstances. Tompkins (1970) describes an experiment in which adults are asked to stand in front of a group of persons and shout loudly, âNo, I wonât!â; according to Tompkins, 95% of those who comply report their spontaneous recall of apparently forgotten events of childhood or adopt childlike postures. Reports are common of hypnotists or psychotherapists who apparently can stimulate recall in normal adults of long-forgotten childhood events using certain, largely verbal, techniques. The validity of these particular phenomena can hardly be said to be well establishedâcertainly the processes or mechanisms underlying the techniques are not understood and sometimes not even identifiedâand only recently have they come under careful experimental scrutiny (see Reiff & Scheerer, 1959, as an example of one experimental approach, and OâConnell, Short, & Orne, 1970, for further experiments and a critique of the Reiff-Scheerer study). Nevertheless, these phenomena provide interesting examples of the variations possible in the retention and forgetting in normal individuals and of some circumstances which may determine this variation.
I take the view that the âcircumstancesâ which influence these wide variations in forgetting are definable largely in terms of the similarity between what is noticed during memory storage and what is noticed when memory retrieval is required. Before I become more explicit about how these terms are used, some heuristic value may be obtained from another example which has potential interest but again lacks hard facts and solid concepts. Investigators at the University of Michigan (Blum, Graef, Hauenstein, & Passini, 1971) observed a subject recall, with dramatic accuracy, a hypnotic dream which reportedly had occurred two years earlier. Blum et al. hypothesized that the critical determinant of this excellent retention was the âdistinctive mental contextâ under which the memory in question had been acquired and was reinstated (for the first time) when recall was assessed. To test this hypothesis, an individual was hypnotized, provided with verbal suggestions intended to create a âdistinctive mental state,â and then was told of an imaginary episodeâcomplex and bizarre in some cases, quite simple and rational in others. Each distinctive mental state was accompanied by a distinctive episode. Unknown to the subject at the time, retention of these episodes was to be tested 143 days later. A specific instance of a context-and-episode combination is useful for illustration. To create a distinctive mental state, the subject was given the hypnotic suggestion that he was floating two feet above the couch on which he was lying, was surrounded by a âbeautiful blue mistâ and could âfeel a light breeze blowing.â The episode then described to the subject was fairly long and detailed. Essentially, the subject was to imagine President Johnson (the date was Spring, 1967) with the face of Sammy Davis, Jr., in cowboy hat and boots, leading a peace march in Hanoi. After the 143-day interval, the subject was placed under a similar hypnotic state in the laboratory and allowed to hear a tape recording of the initial training session; by this means the âmental contextâ of floating in a blue mist and feeling a light breeze was resuggested to the subject. The subject was then instructed to describe the episode that had been presented in conjunction with that particular âmental stateâ during the earlier session. Blum et al. report that with somewhat halting speech, and sometimes incredulous tone to his voice, the subject proceeded to describe President Johnson wearing a Texas-type hat, looking like Sammy Davis, Jr., giving a speech against the war in Viet Nam.
The point of this anecdote is purely as an illustration. It is impossible to conclude much of anything from such demonstrations, and my further concerns will be with facts of a good deal more substance. Questions raised by this demonstration are of some interest, however. Was it necessary for the stimulus âmentalâ) context to be so unusual? Was the hypnotic trance necesary? Was the critical portion of the stimulus context primarily verbal or primarily some nondescript, internal âfeeling,â hormonal or neural in origin? Need the episode itself have been so bizarre? Would any episode be retained with equal efficiency in that particular context? Was this particular subject uniquely suited for this episode or this combination of context-plus-episide? Under the supposition that memory retrieval was indeed facilitated dramatically by reinstatement of the associated mental context, we might ask whether the instrucions themselves were critical or whether it was the subjectâs reaction to the instructions, whether reinstating only isolated aspects of the âmental contextâ would be as effective in alleviating forgetting whatever the source of the forgetting, and whether a similar context-mechanism functions for all sorts of organisms under all sorts of ânaturalâ circumstances. It is with these sorts of issues that I am concerned.
II. MEMORY AND MEMORY PROCESSING:
SOME BACKGROUND
I use the term âmemoryâ here to refer to an individualâs internal representation of what he or she has learned; by âmemory processingâ I refer to the series of operations, changes instate, form, or accessibility, that may occur between the acquisition and behavioral manifestation of a memory.
In current theories, a memory is most commonly viewed as multidimensional, a set of identifiable attributes (components), each of which might define an isolated event or characteristic of an event. At the present stage of physiological knowledge, these attributes are hypothetical representations of elements noticed during an event. An animal placed into a situation where it must jump on a ledge within ten seconds after a tone occurs, or be shocked in the feet, will acquire some information about the contingency between the tone and the foot shock and jumping as attributes of the memory for that episode. In addition, there may be attributes representing many other aspects of the animalâs environment, such as the odor of the experimental apparatus, the texture of the experimenterâs glove or hand, the ambient noise or illumination level, as well as aspects of the animalâs internal state, whether fatigue, hunger, fear, etc. The attributes of our memory of a single word may include its acoustical and orthographic characteristics, its frequency of occurrence in a language, and the kinds of words with which it ordinarily is seen (see Underwood, 1969). A college sophomore instructed to remember a list of words may have as attributes of this memory the features of the room in which the words were presented, how they were presented (for example, acoustically or visually), the appearance of the experimenter, or his own feelings or mood. However, although many attributes combine to form a single memory, they probably function independently, to some extent. The college sophomore may have perfect recall of the list of words one week after initial learning, but may have forgotten some features of the room (how many chairs? was there a blackboard?) or what the experimenter had been wearing, even though he originally registered this information. The differential forgetting of attributes that collectively formed the sophomoreâs memory of the initial learning supports the notion that the attributes function independently. There is no reason to believe otherwise, but this independence does remain an empirical question which now is studied systematically (for example, Galbraith, 1975).
What is gained by viewing a memory as multidimensional? First, this view is helpful in our empirical analysis of retention and forgetting because it provides a reminder that the discriminative or conditioned stimulus is not the only determinant of retention. Second, a theoretical analysis that includes the multidimensional view of memory recognizes the impact of contextual determinants of retention and discourages the notion of a memory as a single âkernelâ located somewhere in the brain, a unitary âtraceâ whose existence alone determines whether an episode will later influence an organismâs behavior. On the other hand, the multidimensional view of memory does bring problems. There is first the problem of infinite regress. For example, if we view the experimental room as a contextual attribute of the memory of a learning episode, ought we not to consider instead the color of the room, the nature of the lighting fixture, and the crack in the wall near the window? Such regress probably can be controlled by defining events psychologically, in terms of perceptual groupings and attentional factors, for example. Unfortunately, this control is difficult and may not be sufficient. A second danger is that of considering too many attributes unnecessarily. Underwood (1972) has warned against this danger in the investigation of verbal memory processing, and the danger readily applies to other classes of memory processing. At present, the advantages of a multidimensional view of memory outweigh the disadvantages, but we may hope that in future years the prefix âmultiâ will be qualified appropriately.
A. Retention and Forgetting
I find it useful to define âretentionâ and âforgettingâ operationally in terms of the measuring device selected to index the memory. Typically, the index of a memory is some behavior that reflects specific learning and usually is selected on the basis of convenience of measurement, or because of precedents or some theoretical or pretheoretical bias of the experimenter. For both human and animal research, this index usually is some fairly obvious response, but it is important to remember that this is a selected index of an acquired memory, not the only one possible and perhaps not even the best. With human subjects, this index may be supplemented by simple interviewing techniques âWhat else do you remember about the episode?â) Supplementary indices with animalsâphysiological changes, for exampleâusually are less convenient. The important point is that indices of memories are arbitrary and often used merely because it has been the custom to do so. A more reasonable criterion would be tied to the particular problem being studied.
I shall define forgetting and retention in an operational sense: âforgettingâ is a decline in the index of the memory relative to that of original learning. Generally, degree of original learning is defined by performance in the absence of the source of forgetting under investigation. When this source is a retention interval, the extent of original learning is defined by performance immediately following the last training trial, with the understanding that the intervening circumstances do not differ from circumstances during original learning. For example, the interval and ensuing events between the last training trial and the test to establish original learning should be no different than that between other learning trials. âRetentionâ is defined by the extent to which evidence of the memory is clearer when original learning occurred than when it did not. Retention may be determined operationally in terms of differences in the index of the memory among individuals presented all the contingencies or contiguities needed for learning and the corresponding index among other individuals treated similarly but without the contingencies or contiguities. Forgetting, or decrement in retention, implies nothing about the source or permanence of the behavioral change.
B. Memory Storage and Memory Retrieval
âStorageâ of a memory is the hypothetical process through which events are represented and established collectively as a memory for potential future reference. âRe...