Multiple Memory Systems & PRS

1997

pdf version

Memory lies behind much of behavior and creates the substrate by which one adapts to environmental stimuli.  In terms of evolutionary design, it is highly unlikely that a single, unified system of memory exists in higher animals.  It is more likely that multiple systems of memory work in unison at times and separately at others to produce the entire range of functions and structures that science terms memory.

In current neuropsychological literature two primary modes of analysis are presented to account for memory.  The processing theories, especially as highlighted in the work of Blaxton (1995), and the systems theories as delineated in Gabrieli (1995), forward two opposing viewpoints.

Systems theorists, using amnesic subjects, make a clear distinction between implicit (i.e. memory of an 'unconscious' type; as in learned motor skills) and explicit memory (memory such as directed recall) and link this distinction to specific neural structures (Sherry & Schacter, 1987).  In particular, global amnesia, as caused by damage to medial-temporal lobes and/or diencephalic structures, leads to a deficit in explicit memory while retaining fully capable implicit memory ability.  The prototypical case of this dissociation is the subject known simply as H.M. (Milner, 1962).  H.M., as a result of surgery meant to alleviate his epilepsy, lost both his hippocampi, amygdala, and entorhinal and parahippocampal gyri (collectively deemed the medial-temporal lobes).  While H.M. suffers from nearly absolute global anterograde amnesia, he remains capable of learning feats of procedural, or implicit, memory.  Again, this memory is the kind that produces motor sequences and learned movements.  Gabrieli, et al., in one experiment, used tracing skills to exhibit the retention of this memory in amnesic subjects (1993).  The correlation between these explicit memory deficits and lesions to these specific structures seems rather strong.

Processing theorists, through studies with normal subjects, have found a dissociation between memory performance they term perceptual and conceptual (Blaxton, 1995).  Perceptual memory processes stimulus format while conceptual memory processes content based on the semantic meaning of stimuli.  Thinkers of this camp criticize systems thinkers as conflating these distinct processes.  Blaxton, in particular, illustrates how many systems experiments do not make a proper distinction between implicit/explicit memory and test only conceptual memory abilities (1992).  Furthermore, Blaxton contends that systems theory is unable to account for the dissociability of perceptual and conceptual memory in normal subjects (1989).

Processing theorists, though raising several key points, generally fail to validate either their critique or their positive claims.  For instance, amnesics cannot be characterized as having global impairment in all their semantically driven conceptual memory processing (Gabrieli, 1995).  Both the explicit/implicit dichotomy and the perceptual/conceptual stance encounter challenges.  The opposing schools have done good service to each other by pointing out the limitations of their respective models.  The dissociability of perceptual and conceptual measures in normal subjects, if anything, only corroborates the broader philosophical underpinnings of a systems theory that pairs key neural structures with differing types of memory.  Blaxton does not seem to have  a problem with accepting the neural bases of these memory processes, but lobbies for more precise experimentation and standardization among the study of amnesics (1992).

As Gabrieli predicts at the end of his article, A systematic view of human memory processes (1995), an aggregate of the features of both perspectives will lead to a more fruitful description of memory processes.  Since, he proposes, neither system can account for all the phenomena, a combination of various elements of each better describe and predict the phenomena.   Schacter (1990) outlines an explanation that, by combining some different elements of the aforementioned theories, more completely covers the range of phenomena that arise in memory research.

Schacter proposes a theory of perceptual representation systems (PRS) (1990).  Careful analysis of priming phenomena (such as stem completion and word identification) reveals that a kind of context-dependent perceptual memory does exist.  In these experiments, subjects are given words in a study list and later tested with an assortment of words, some of which had been included in the study list.  Stem completion exercises (in which the subject is given part of a word that has already been seen) and word identification (that is, identifying a word as one of the ones already seen) show much higher incidence of familiarity by subjects.  In amnesics, this parallels implicit memory utilization and in normals, perceptual memory.  Simply identifying a task with a particular system does not really illuminate the nature of the phenomena in a germane fashion.  Importantly, Schacter illustrates that in these priming experiments various representations of the words can be dissociated.

In promulgating the PRS theory, Schacter portrays subjects with unusual impairments in which highly individual aspects of words cause priming effects.  Apparently, the storage of words can be done in various ways including one visually based system known as the visual word form system (Warrington & Shallice, 1980).  In this verbal subsystem, a word's visual representation is selectively stored; a representation that is wholly separate and dissociable from the word's semantic content, or meaning.  This faculty can be individually impaired without affecting other aspects of the word's representation!  In other words, in the brain, words are approached from many different angles.  The closest analogy would be something akin to a hologram in which various angles diffract to create the representation of a three-dimensional image.  Even with one of the vectors oblated a hologram is still tenable though diminished.  If such a visual word form system truly exists, which a host of research suggests, then priming effects may be explained by activation of a word's visual form alone.  In one interesting experiment (Marshall & Newscombe, 1973) it was determined that the word form system can be damaged in a highly selective fashion.  Patients with surface dyslexia are able to read normally except that they read irregular words as if they were regular (using a grapheme-to-phoneme conversion strategy).  This indicates that a specific element can be lost while the rest of the word form system remains completely intact.  Research such as this suggests that no semantic processing (such as in conceptually-driven processes) nor conscious recognition and recall (as in explicit memory systems) need occur in order to stimulate a priming effect since specific parts of a word activate recognition that may be 'blind' to the rest of the word form system.  This may help to explain the implicit memory effects exhibited in amnesics and normals alike.  If the visual word form system lies outside of those regions typically injured in amnesics' brains then the implicit memory phenomena should remain as in normals.

The motor skills that amnesics perform (Gabrieli, 1993) are probably a separate subsystem.  Under the general rubric of implicit memory remain a number of varying items.  The visual word form system illustrates one specific process while procedural skills reveal another.  Also, structural representations of physical objects find a separate, though related, explanation.  In many forms of visual agnosia, subjects are unable to identify or categorize objects or even distinguish whether an object is a 'possible' object or an 'impossible' one (as defined by geometric representation).  Given semantic information, though, these same subjects find it easy to identify or describe an object and show strong priming effects in subsequent measurements (Schacter, 1990).

A natural response to Schacter's observations is that these random agnosias have little to do with memory proper.  Schacter himself argues that the presence of independent processing modules is not prima facie evidence for multiple memory systems as these modules could conceivably all run into a common memory system or at least utilize the same rules as other modules (thus allowing them to perform in parallel) (1990).  While generalized claims cannot be corroborated either for or against such a notion, the specific claims that Schacter considers (the visual word form system and structural  systems) provide forceful evidence that at least some elements of 'memory' are independent, dissociable, and employ unique (though compatible) rules.  As such, a case in favor of a multiple memory system that is composed of, and draws on, a number of subsystems finds much more credence than a unified mechanism.

Perhaps the finest point raised in Schacter's theory of modular processing in different memory subsystems is his conceptualization of memory as representation (1990).  It is important to make this distinction for memory is not the direct reproduction of objective items, processes, and pairings but rather the representation of these different objects to the mental states of the organism.  In simpler terms, representation implies multiple modalities of storage in the brain and veers away from a simple, one-to-one reproduction of the outer world to the inner world.  While this point may seem insignificant it forcefully substantiates the notion of a multiplicity of memory systems as logically necessary.  The groundwork by which it accomplishes this lies in the very prickly topic of consciousness and agency which will be generally avoided here.  Notably, though, consciousness, while ostensibly unified, most certainly is not.  To use Dennett's language (1991), there is no "Cartesian theater."  In other words, the notion of reproduction entails a central processing mechanism by which processing theories are made applicable; a notion which finds no true viability.  Representation, however, implies different representational means and, ultimately, different usages of perceptual information by different brain regions.  In a large analysis then, memory processes, as studied in normal subjects and in subjects with various impairments, must be analyzed as different modalities and subsystems which can be dissociable both in anatomy and performance.  Schacter's attempt to discover the specific elements of different subsystems both theoretically, and practically, best makes sense of the range of memory phenomena in normals and amnesics.  Underlying all these items must be a host of specific neural structures and the study of patients with memory dysfunction will undoubtedly point to an increasingly small number of specific neural substrates.

Memory and representation grow together inextricably.  To speak of memory one must consider the nature of representation.  Representation implies a plurality of modes that know relevance in terms of specific functions.  The natural tendency of the human mind is to unify and simplify the causes and nature of a phenomenon.  Even the term 'memory' does injustice to a host of functions that the conscious subject knows as many different abilities.  While the recognition of multiple memory systems does not provide the hope of a simplification of research and eventual understanding it does provide a practical way to go about the goal of understanding.  To study memory the researcher must study all its various attributes.  As such, the dichotomies engendered in processing theory are just as valuable as systems theory because they provides measures of memory processes.  However, the more general framework of processing theory cannot support the multiplicity of features that memory research has already brought to light.  Studying normals is valuable but will not provide the causative sequence of memory as clearly as studying subjects with unusual memory impairments.  Multiple memory systems which feed into faculties both explicit and implicit, perceptual and conceptual, even conscious and unconscious, yield a much more realistic and accurate portrayal of the labyrinthine complexity of that phenomenon succinctly known as memory. 

REFERENCES

Blaxton, T.  1995.  A process-based view of memory.  Journal of the International Neuropsychological Society, 1, 112-114.

Blaxton, T.  1992.  Dissociations among memory measures in memory- impaired subjects: Evidence for a processing account of memory.  Memory and Cognition, 20, 549-562.

Blaxton, T.  1989.  Investigating dissociations among memory measures: Support for a transfer-appropriate processing framework.  Journal of    Experimental Psychology: Learning, Memory, and Cognition, 15, 657-668.

Dennett, D.  1991.  Consciousness explained.  Little, Brown & Co.  Boston.

Gabrieli, J. D. E.  1995.  A systematic view of human memory processes.  Journal of the International Neuropsychological Society, 1, 115-118.

Gabrieli, J.D.E., Corkin, S., Mickel, S.F., & Growdon, J.H.  1993.  Intact acquisition and long-term retention of mirror-tracing skill in Alzheimer's disease and global amnesia.  Behavioral Neuroscience, 107, 899-910.

Marshall, J.C., & Newsombe, F.  1973.  Patterns of paralexia.  Journal of Psycholinguistic Research, 2,  175-199.

Milner, B.  1962.  Les troubles de la memoire accompagnant des lesions hippocampiques bilaterales.  Psychologie de l'hippocampique.  Paris: Centre National de la Recherche Scientifique.

Schacter, D. L.  1990.  Perceptual representation systems and implicit memory: Toward a resolution of the multiple memory systems debate.  In A. Diamond (Ed.), Development and Neural Bases of Higher Cognitive Function, Annals of the New York Academy of Sciences, 608, 543-571.

Sherry, D.F. & Schacter, D.L.  1987.  The evolution of multiple memory systems.  Psychological Review, 94, 439-454.

Warrington, E.K. & Shallice, T.  1980.  Word-form dyslexia.  Brain, 103, 99-112.