Cognitive Therapy and Research, VoL 4, No. 1, 1980, pp. 1-25
The Self as Memory Processing James C. Mancuso ~ and Seth G. Ceely State University of New York at Albany
Different areas o f theoretical investigation are briefly discussed in relation to competing paradigms and a theory o f self. Clarification o f these areas is taken to be contingent on resolution o f two theoretical issues. Is the self to be treated as a "'caused" or as a "'causing" event? What is the nature o f the storage and retrieval processes related to the enactment o f different self-relevant behaviors? In regard to the issue o f cause, interpretations in terms o f two major theoretical paradigms, mechanism and contextualism, are briefly outlined and the validity o f contextual theory is discussed. In regard to the issue o f storage, examination o f current theories o f memory lead to a consideration o f self-representation as involving both generic and episodic memory structures that possess a number o f organizational properties relevant to an understanding o f consistency and variability in a person's enactment o f self-relevant behavior. In light o f these considerations, the advantages o f contextual formulations o f self-theory are reconsidered and current, related research is noted. People, functioning in the role of naive psychologists, use conceptions by which the origins of thought and action are attributed to internal processes. Psychologists, as formal theorists, have developed propositions about these internal processes. Naive psychologists (see Wegner & VaUacher, 1977) and formal psychologists (Bandura, 1978; Stuart, 1977) have invoked a concept of self to discuss the regulation of the internal processing. Yet, in a scientific ambience in which prediction and control are established as highly positive criteria of an effective science, propositions about a process regulating self will be critically scrutinized (Wolpe, 1978). To place the self in control of the self undermines the probability that the behavior scientist can predict behavior by gaining control of the vari'Address all correspondence to James C. Mancuso, Department of Psychology, State University of New York at Albany, Albany, New York 12222. 1 0 1 4 7 - 5 9 1 6 / 8 0 / 0 3 0 0 - 0 0 0 1 5 0 3 . 0 0 / 0 © 1980 Plenum Publishing Corporation
2
Mancuso and Ceely
ables, particularly stimulus variables, that are assumed to be causally connected to behavior outcomes. If a self may transform carefully calibrated input, then the scientist does not have complete control of response outcomes. The theoretical troubles created by the foregoing considerations are observable in many parts of the behavior science territory.
THE SELF IN PLANNED BEHAVIOR CHANGE PROGRAMS Despite a major effort in that direction, applied psychology could not adhere to strict mechanistic principles to delete internal information processing from programs for behavior change interventions. (The terms mechanist and contextualist, as used throughout this paper, follow from Pepper's [1942] masterful application of the root metaphor method of analyzing cognitively valid metaphysical systems.) Kazdin and Wilcoxin (1976), for example, show that dozens of investigations into the efficacy of "systematic desensitization" have failed to confirm the role of the basic principles that have been assumed to account for the change process. Instead, the effects of the desensitization interventions might best be explained by referring to the expectations of the target person, which somehow transform the meaning of the input. Indeed, work like that of Maltzman (1977) has raised doubts about the conditioning model, the most basic concept of the strict mechanistic "behavior modification" movement. "It now seems apparent that classical conditioning in normal adults, as ordinarily studied in the laboratory, is a consequence of thinking rather than vice versa" (p. 112). Major revisions in theory must follow from accepting the assumption that something internal, and not the external conditioned stimulus, is a necessary antecedent of behavior. Theorists (for example, Bandura, 1978; Mahoney & Arnkoff, 1979) have turned toward building assumptive structures that can explain the means by which people can be taught to self-regulate behavior.
DECEIVING THE SELF
Mixon's (1972, 1977) analysis of the implications of his role-playing replication of Milgram's (1963) "obedience study" have raised a storm over some of social psychology's most familiar experimental procedures. Why should an investigator build a superstructure of deception if he can get comparable results by the direct procedure of asking subjects to enact a role, pretending to respond as if the whole expensive, complicated subterfuge had been erected? But, if a self is asked to operate directly on the input, does that self transform the input in ways that it would not be trans-
Selfas MemoryProcessing
3
formed if the input were " f e d " directly into the self?. Those who object to the role-playing approach to data gathering seem to be saying: "The selves created in a role-playing situation, that is, where the participants are asked to engage in 'an illusion' cannot be the true selves."
SELF-OBSERVATION OF THE TRUE SELF Studies of self-attribution processes may be interpreted as studies of "theory of self" (see Epstein, 1973). Recent trends in this kind of investigation are reflected in discussions of causal attribution (Nisbett & Wilson, 1977; Wortman, 1976), attribution of personal freedom (Harvey, 1976), and self-perception of intrinsic motivation (Ross, 1976). Such studies interpret the person's perceptions of the motives that account for the enactments of his own behaviors. Nisbett and Wilson (1977) propose that people attribute cause to a stimulus by applying their personal causal theories relative to that stimulus: "Thus, a particular stimulus will be deertied a representative cause if the stimulus and response are linked via a rule, an implicit theory, a presumed empirical covariation, or overlapping connotative networks" (p. 249). They go on in the same article to reflect a tenuous position previously expressed by Snyder (1976). "Several sets of observations seem to suggest that perceptions about the causes of behavior may be independent of the actual causes of behavior" (p. 54, emphasis ours). They treat their theory of cause as the source of accurate causal statements and then try to account for the subjects' failures to report "accurate" causes of their own behavior. Is the subject chained inside Plato's cave, while the experimenter can roam about at will to observe the true self as well as the shadow self?.
SITUATIONISM, INTERACTIONISM, AND THE CONTINUING SELF When Mischel (1968) masterfully attacked the then-standard trait theories he advocated a stimulus-response reinforcement version of situationism. Bowers (1973) notes that if one adopted the mechanistic features of this variant of situationism, he would perceive "individual differences in behavior within the same situation as something of an embarrassment-to be conceptualized as a result of past experience or simply as error variance" (p. 318). From a mechanist perspective one expects behavioral continuity. Bowers found this current position wanting, and then advocated a position he called interactionism--a position that resembles the position
4
Mancuso and Ceely
variously described as interdependence (Sampson, 1977) or contextual/sm (Jenkins, 1974; Mancuso, 1977; Mancuso & Sarbin, 1976; Sarbin, 1977). The interactionist, or contextualist, position observes a tenet that "situations are as much a function of the person as the person's behavior is a function of the situation" (Bowers, 1973, p. 327, emphasis ours). But the behavior scientist risks the possibility that this "person" will metamorphize into an uncontrolled homunculus who sits on the shoulder of the "unconscious," directing the behavior of the self.
INFORMAL EXPLORATIONS OF SELFHOOD If one were to "pretend" that he is a lion, he would have a difficult time obtaining social confirmation of the reality of his role enactment. Thus he might conclude that his "pretense" does not qualify him to assume the self-status of a lion. If another person introduces himself as Vincenzo Carrato and then "pretends" that he is un conoscit&re della cucina ltaliana, his audience could confirm his role enactment, and he would conclude that he, as his true self, may assume the status of a sophisticated Latin food lover. There might, however, be a spectator who would refuse to validate his presentation, noting that Carrato is a surname originating in Campania, and that real Italian cooking begins north of Firenze. And so, his status of a conoscitbre would be no more "real" than was the first actor's status as a lion. His audience failed to verify his acts, in either case, and if he takes his audience seriously, he needs to change his act. He would need to lay out a scenario that would allow his audience to accept the authenticity of the self he had presented. Consider these five domains of behavioral analysis--(a) self-regulations, (b) role playing as a technique in social psychological study, (c) the study of self-perception from an attribution perspective, (d) the problem of situationism and continuity, and (e) playing the conoscitbre to an audience that withholds validation. Each example represents an instance of a situation in which there is disagreement about conceptualizing selves, particularly the implied or inferred causal aspects of selves. The social psychologist who hesitates to accept the basic utility of data collected in a roleplaying situation questions the comparability of behavior that evolves in vivo and the behavior generated when a self has been granted the status of a cause of behavior. Investigators are advised to create a veridical stimulus-as-causer Similarly, the attributionist who compares self-perceived causes to actual causes of behavior appears to assume that the investigator creates real causes. To grant epistemological parity to the experimenter's perceived
Self as Memory Processing
5
causes and the subject's perceived causes is to a p p r o a c h dangerously close to a position that gives a person the p o w e r to control, as well as to perceive, his own motives! Obviously, to claim that "situations are as m u c h as function o f the p e r s o n " seems to encourage a conceptualization o f self as cause. Such a conceptualization has been a n a t h e m a in some strongholds o f m o d e r n psychology, and, in the absence o f a m o r e acceptable position o f causality, psychologists are persuaded to hold to the kinds o f efficient causality statements that underlie the mechanist paradigms that have guided the most respected psychology o f the 20th century (Rychlak, 1968, 1976). Finally, the hostile character w h o denies the authenticity o f the a c t o r ' s conoscitbre status leads us to infer his old-fashioned belief that a self is fully established by o n e ' s genetically determined biological structure. C a n psyc h o l o g y a b a n d o n mechanistic conceptions, develop useful ways o f discussing behavior regulation and a self, and yet remain scientifically creditable2
PARADIGMS AND THE SEARCH FOR THE SELF The continuing s e l f n r e a l , imagined, hypothetical, causal, or o t h e r w i s e - - f o r c e s itself into psychologists' central processors. But contrasting and apparently irreconcilable basic assumptions have been the source o f often acrimonious debate. Consider the following statement: Role taking [in a social situation] is required for whatever perspective [a person] assumes, whether natural self, specific other, or generalized other. Such role taking, in conjunction with situtional determinants, results in some type of role enactment. Given a constant external situation, overt role enactment may be the same or it may differ for self-role taking as opposed to the taking of some other role. But why should one type of role enactment be considered as being in some sense prior, more basic, ormore real than another role enactment? My answer is that it should not. This answer gives epistemological parity to role behavior stemming from all role perspectives. One self-role performance is granted no more or no less importance than is any other self-role performance. (Hendrick, 1977, 470, material in brackets ours)
In contrast, Bowers 0 9 7 3 ) cites Skinner (1955-1956) in the following context. • . . the situationist model, taken to its extreme, ultimately discredits m a n as a n a u t o n o m o u s active agent. Situationism's foremost philosopher--Skinner (19551956) put it this way: " E v e r y discovery o f an event which h a s a part in shaping a m a n ' s behavior seems to leave so m u c h the less to be credited to the m a n himself; and as such explanations become m o r e a n d more comprehensive, the contribution which m a y be claimed by the individual himself appears to approach z e r o " ~ . 52). (Bowers, 1973, p. 331)
F r o m Skinner's perspective there could be no m e a n i n g in H e n d r i c k ' s claim that " o n e self-role p e r f o r m a n c e is g r a n t e d n o m o r e or n o less ira-
6
Mancuso and Ceely
portance than is any other self-role p e r f o r m a n c e . " Selves do not write the scripts for performance! Selves are caused! They are not causes! Skinner, working within a paradigm that has the root metaphor o f the machine, would locate the "actual cause" o f a self-defining script outside o f " t h e p e r s o n . " Scripts, it is assumed, would not be available for enactment if they had not been "built i n " by the applications of some kind of force, such as reinforcement, anxiety reduction, or social approval. Further, from a mechanist's perspective it must be assumed that scripts are stable and repeatable. Their repetition would be regenerated, in some way, by the stimuli to which they had been attached. While noting Bowers's (1973) comment on the mechanists' potential embarrassment upon observing individual differences in behavior within assumedly similar situations, one may see that users o f the mechanistic postions have tried to explain both repeatability and intrapersonal variation without reverting to the trait conceptma basically formist concept that has been found to be wanting as a means o f explaining continuous, ongoing persons (Mischel, 1968). Repeatability would be contingent upon the distal presence of classes of stimuli to which reinforced responses may generalize. At the same time, apparently "inconsistent" (from a trait perspective) responses will be observed in one person's behavior, because the reinforcement has bonded "inconsistent" responses to a variety o f specific stimuli. The same person might be " k i n d " and yet " c a l l o u s , " since specific stimuli are associated with assumedly kind responses, whereas other specific stimuli are bonded to assumedly callous-type responses. By this formulation one, at one time, removes the concept of the trait as the inner guide to consistent behavior while avoiding the temptation to accept a concept !ike Rycfilak's (1976) identity, which looks dangerously like the notorious inner helmsman who " d e c i d e s " that the person may vary his behavior. Here, however, is where the battle line is drawn. Any theory can be written from either an introspective("behind the mask") or an extrospective("in front of the mask") slant (Rychlak, 1968, pp. 27-34). I think that personality theorists always gravitate to the introspectiveperspectiveand assumethereby that there/s an identity from that slant on events. (Rychlak, 1976, p. 210) Rychlak and Hendrick see something like an identity that "takes roles." Skinner and others who use mechanistic formulations conclude that an individual does not " t a k e roles." The individual's contribution approaches zero.* He "emits roles," which vary according to reinforcement contingencies! It would be audacious to o f f e r a resolution to the basic argument: Shall one conceptualize the perceived self as cause, or shall one conceptualize the self as caused? To make any headway toward a resolution o f the argument, one must consider his assumptions regarding two very important issues that shape one's conclusions about how to fit the self into behavior
Self as Memory Processing
7
theory. First, a theorist's assumption about the scientific legitimacy of one or another concept of causality will shape that theorist's view on the self as cause. Second, a theorist's view on the "storage and retrieval processess" in memory operations will influence his view on the self as cause. It would be helpful to explore alternative assumptions relative to these two basic issues, and to test the utility of these alternative assumptions in a theory of the self. What Is Caused?
The canons of a science that is guided by a mechanistic paradigm require a particular kind of causality statement. Rychlak (1976) aptly refers to the resulting perspective as a "tracking approach" to science. A discrete event is taken to be the unit of study, and then the "laws" that emerge from such study must meet the requirement of "tracking" the event. The peak experiences for mechanistic scientists derive from successes in laying out the tracks so that the position of the event will be extrapolated successfully, even at those times when the event is not directly observable. In top-rated mechanistic psychology, the event is muscle-induced movement, which is discussed as behavior. When speaking of cause, then, a mechanistically inclined psychologist would be speaking of the cause of a behavior. In such a system the behavior must, without equivocation, be related to variables that determine the "orbit" of the response. In that this concept of cause is central to a mechanistic paradigm, the user of this system cannot treat lightly the observation that a behavior is never a replica of another behavior. To allow the possibility that self, as a behavior-changing variable, may intervene between external causal variables and the caused responses is to allow a wholly unacceptable violation of the paradigm's basic notion of causality. We will make the observation, perhaps trivial, that there are other ways to look at causality and that the paradigms that support the alternative approaches do require that one think in terms of basic units other than behavior or responses. From a contextualist position (see Jenkins, 1974; Mancuso, 1977; Mancuso & Sarbin, 1976; Sarbin, 1977, for further discussion of contextualism in behavioral science) one takes the total context as the basic unit of study. For the present discussion, we need to consider that change, in a contextualist paradigm, is epigenetic. Contexts flow; they are not to be conceptualized as discrete, unique events. This point is well argued by Neisser (1976), who, to present his current position, first abrogated an earlier perspective (Neisser, 1967) in which perception had been treated as a static, temporally discrete affair. Neisser then advocated the perspective embodied in the following quotations: "Not only reading but also listening, feeling, and looking are skillful activities that occur over time. All of them depend
8
Mancuso and Ceely
on preexisting structures, here called schemata, which direct perceptual activity and are modified as it occurs" (Neisser, 1976, p. 14). " A t each moment the perceiver is constructing anticipations of certain kinds of information, that enable him to accept it as it becomes available" (p. 20). " I n most other respects, however--in its continuity, its cyclic nature, its dependence on continuously modifed schemata--perceiving is a kind of doing" (p. 52). A person's perceptions, then, are to be treated as but one strand of a flowing psychologically significant context. "Behaving," that is, positioning overtly observable muscle groups, represents another strand in the context. Distal events that interrelate to muscle involvement, sensory ending alteration, and neural activity represent other strands of the context. Thinking about behavior within this contextualist paradigm discourages the view that change is to be represented as the sudden appearance of a discrete event--a new strand. A contextualist describes change (see Pepper, 1942, pp. 242-246) in terms o f the shifting interconnections of fused strands within the context. To a contextuaiist, nothing will come from looking at a change in an isolated strand. In fact, the very idea of isolating strands is incomprehensible. "All supposed elements a r e . . , f u s i o n s . . , so far as anyone claims immediate experience with them" (p. 245). If one isolates "the response" out of an ever-evolving context, which acts as the ever-changing backdrop for whatever one calls the response, the isolation is artificial and, at best, is done as a heuristic convenience. In such cases, the expectation that one observes a response that is "similar" to a prior response also is taken as an artifice. In later discussion we will return to the matter of "repetition as reconstruction," and in that discussion we will look at the evidence that response repetition is best represented as a reconstructive retrieval process. The point of this section is: A theory of persons need not include a set of principles that tracks responses, since there is no requirement that one explain a response as the basic unit of the system. One's theory could consider fused contexts and changes in the interrelationships in those contexts. Thus a theorist is not required to follow the presumption that "responses" are caused. Changes in contextualist configurations might require explanation. Causal explanations would be framed to describe changes in the contextual relationships, to explain the epigenetic evolution of the context. The shift from efforts to set out "straight-line," efficient causality statements toward setting out statements that describe intracontextuai relational changes represents more than an isolated act of faith. The shift follows from a declaration that one is accepting a new paradigm. The paradigm shift has been encouraged by the failure of the mechanistic paradigm, relative to explaining the constructivist features of human behavior. From within the contextualist paradigm one's focus is not on the re-
Self as Memory Processing
9
sponse but on the context that frames the event that the behavior analyst might call a response, for one presumes to explain response reconstruction, or retrieval, in terms of the embedding context. Foreview. From the foregoing, one recognizes the logical route to the following precursory statements: A self need not be treated as a caused response. A self--as a perception o f a person--may, indeed, be treated as a reconstruction. In this way, one treats selves as objects. Self as a reconstruction might be extracted as a " r e s p o n s e . " Should one follow this heuristic convenience, however, he will profit from seeing the reconstructed self as one strand in the context within which that self is embedded. The reconstructed self, then, must be accounted for in terms of the context, which includes a person, which may be regarded as a self-as-process. Thus changes in the reconstructed self-as-object can be understood in terms of variations in the self-as-process--another major strand of the ever-flowing context. Changes in context, following from the presumption developed earlier, would be taken to be changes o f relationships within the context. It follows, therefore, that one would think of a change in self-as-object as an effect o f a change in the self-as-process by which self-as-object is reconstructed. In short, the self-as-object would change, as would any reconstructed (retrieved, remembered) event, as a result of variations in the construction system. The self-as-process, in that it changes, "causes" changes in the self-as-object.
What Is Remembered? The second issue whose resolution is crucial to shaping theorist's views o f the self-as-cause is the issue o f how one represents " s t o r a g e " and the reconstruction process relative to memory function. In an earlier section the analysis led to the conclusion that a mechanist's perspective encourages conceptualization o f the self as a caused event, but the paradigm does not allow a self to be regarded as a cause. The self-as-object--the perceived self--one infers, is represented in some way within the boundaries o f the person as organism. At the least, the representation involves some connective bond that attaches sensory input to whatever shall be called the response, that is, the perceived self. One may apply, perhaps, the concept o f stimulusresponse bond to talk about the perceived or caused self. One would also need to invoke a concept o f a stored representation o f those externally executed responses, which are taken as self-defining responses; otherwise there would be no way to explain repetition in a person's self-presenting behavior. One could not avoid an assumption o f a stored " s o m e t h i n g " to account for the observation that a person gives similar self-defining responses in similar situations. H o w does one describe the stored repre-
10
Mancuso and Ceely
sensation? What is retrieved from the memory operation when self-defining responses are reenacted? Shepard (1975) has written about the standard mechanistic solution to representation in memory that "it is the purely verbal mediators and, especially, the implicit naming responses that have been most extensively acknowledged, formalized, and empirically studied" (p. 87). The internal naming response could be assumed to follow the same laws as those that were applicable to the external, corresponding overt responses. Questions about transformation of stored representations, which are quite troublesome to the behavioristic models, were deflected. First, since it is clear that generally the word that is learned for a particular object is associated with it by arbitrary convention, there has never been any temptation to puzzle about the formal or structural relation between a word (or its internal representation) and the external object, say, for which that word stands. Second, since a word is not only arbitrary in this sense but also is a fixed, categorical unit, it is not in general the case that the kinds of transformations to which objects are subject in the external world are in any way paralleled by analogous transformations in the corresponding words. (Shepard, 1975, p. 87)
To conceptualize storage units as words, as still pictures associated with words, or as a program of sequenced motor actions depends on an assumption of first order isomorphism. That is, such conceptualizations reflect the belief "that there is some structural resemblance between an individual, internal neurophysiological event and the individual external object that it represents" (p. 88). In light of the quantity of cogent argument and empirical validation of other conceptions, however, one no longer can work comfortably from a position of "first order isomorphism." Shepard, for example, presents a case for a perspective that incorporates a principle of second order
isomorphism. In second order isomorphism, our knowledge of the world is embodied--not in the structure of each individual internal representation considered by itself--but rather, in the extent to which different representations are functionally equivalent or have something functional in common with each other at the neurophysiological level. (p.
89) Weimer (1977), taking the same considerations into a discussion of motor actions, says that "the keyboard in the cortex controls abstract, functionally specified actions, rather than particular muscle movements" (p. 274). No eventmnot words, not objects, not visual scenes, not specific motor acts, nothingmmay be regarded as having a stored representation that maintains a structural resemblance to that event. The salient point, for our discussion of self, relates to the transformability of the internal representation and the psychological operations involved in establishing
Self as Memory Processing
11
functional equivalences. To Shepard, as a student of cognition, transformability provides a key to determining one's conception of what is stored. "Any clarification of the way in which the mind carries out such mental transformations should, at the same time, help to clarify the nature of the representations that are thus being transformed" (p. 91). The problems interlock. An investigator may demonstrate the validity of a concept about internal representation transformability. On the other hand, one's concept o f what is stored takes its strength from demonstrations of the kinds of transformations that are made in internal representations. To establish a useful position from which to carry on further exploration and discussion Shepard (1975) has laid out a cogent construction. The simple fact that we can differentiallyattend to component parts of an analyzablestimulus(Shepard,Hovland& Jenkins, 1961)--that we can, for example, focus or vividlyimaginethe upper right hand corner of the square, almostto the exclusion from awareness of the rest of the square--strongly indicates that our internal representationis differentiated,does have parts. (p. 91) By accepting the proposition that internal representation somehow relates to "parts," Shepard identifieswhat has become a major focus in the study of the organization of memory structures, i.e., the role of attributes or properties. The idea that encoding and retrieval can be discussed in terms of attributes or features has been accepted by a number of memory theorists (Anderson & Bower, 1972; Kintsch, 1970; Rumelhart & Ortony, 1977; Tulving & Thomson, 1973). The position may be summarized, as has been done by Murdock and Anderson (1975): " N o representation of an item per se is stored; rather, certain features are selected or abstracted, and these features, or attributes, represent the stored information" (p. 153). To treat the to-be-remembered event as a multidimensional amalgam of features, attributes, or properties binds a theorist's propositions regarding the nature of retrieval. As Shepard (above) has noted, one's discussion of retrieval processes follows logically from his assumptions relative to what it is that is being stored. And if one assumes that squares are retrieved by parts, then one could also assume that selves are retrieved by parts. Features as Stored Units. The feature storage assumption, though it appeared and has been summarily utilized by some personality theorists, particularly those who have worked from a constructivist person perception approach (Bannister & Fransclla, 1971; Kelly, 1955; Mancuso, 1970; Sarbin, Taft, & Bailey, 1960), can be more fully elaborated by drawing on the support being given to the model by psychologists who study memory and cognitive processes. Collins and Quillian (1972a,b), for example, have used a semantic network model, more recently elaborated by Collins and Loftus (1975), by which to discuss storage and retrieval. In their model, as in the personality
12
Mancuso and Ceely
theory of Kelly (1955), one creates a construction, or a semantic structure, by integrating sensory input with the multidimensional construct space that constitutes the person's existing knowledge system. One may then treat the memory process as a reconstruction that entails locating stimulus input from the event within the construct system. Concretely, Collins and Quillian (1972b) report studies that use the true-false reaction time technique. How does one re-cognize the validity or invalidity of the statement " A canary has skin" or " A canary can sing?" In Collins and Quillian's (1972b) model, it is assumed, that evaluation of a sentence is a matter of establishing a path through the semantic network, interrelating the different concept nodes referred to in the sentence. Node interrelationships are defined by a set of pointers (e.g., superordinatesubordinate or property relations) that organize the concept structure along certain major dimensions. Having traced a path through various nodes, the person's evaluation of the sentence is based on any found discrepancies between the preestablished nodal interrelationships and the relationships that are developed by the pointers drawn up in the sentence. Thus the sentence "The canary can sing" is easily recognized as valid in that only one pointer need be followed within the node marked by the term canary, and that pointer leads directly to the property "can sing." Recognition of the validity of the sentence "The canary has skin" requires a more complex process, in that the pointers must pass through two hierarchical levels--from the node marked by the term canary to the node marked by bird, and forward to the node marked by animal. At that node there is a marker to the property "has skin." One may also assume that reaction times measure the speed of processing through property and hierarchical level relationships. Using this model, Collins and Quillian were able to produce the predicted curves that plot the reaction times required to process the variety of sentences that they devised. In their extension of the semantic network model used by Collins and Quillian (1972a), Collins and Loftus (1975) clarify aspects of the model that were obscure in earlier work and that are particularly germane to the discussion here. First, in regard to the property links that relate the concept nodes, Collins and Loftus indicate that not all links are of equal importance. Rather, associated with the different links relating to a concept are "criterialities, which are numbers indicating how essential each link is to the meaning of the concept" (19. 408). This assumption makes explicit the hierarchical nature, not only of the superordinate-subordinate relations that hold among nominal concepts (e.g., canary is a bird, bird is an animal) but also of the property relations that are linked to concepts. Collins and Loftus also elaborate the concept of storage of properties at different points in the concept hierarchies. Earlier descriptions of the
Self as Memory Processing
13
model had allowed the inference that certain properties are stored only once in memory (e.g., the property " f l y " is stored at the superordinate concept node "bird," but not at subordinate nodes like "canary" or "hawk"). This is not a necessary presumption. "People can reasonably be supposed to store certain properties at more than one level in a hierarchy" (p. 409). Furthermore, the utility of the model may be extended by-accepting the proposition that the properties or attributes associated with a concept will be those that allow the concept to be unambiguously identified within the given context in which it is encountered (Bobrow & Norman, 1975, p. 127). By this proposition one begins to consider the salient issues involved in accessing different concepts during reconstruction. Rumelhart, Lindsay, and Norman (1972), in their formulation of a "directed search" model of long-term memory processes, have introduced additional relevant considerations. Using conceptualizations that parallel those of Collins and his associates, they have described the basic structural elements of the "data base" to be sets of nodes interconnected by relations. However, in regard to nodal representation, two distinctions are introduced: the first is between primary and secondary nodes and pertains to the degree of abstraction in the information represented; the second is between concepts, events, and episodes and pertains to the organization of information represented in memory structures. Primary nodes involve abstracted concept definitions in the relational format and usually correspond to a natural language word; that is, primary nodes represent generalized information. Secondary nodes, on the other hand, "represent the concepts as they are used in specific contexts--a token use of a primary type node" (Rumelhart et al., 1972, p. 203). In addition, secondary nodes need not correspond in any simple or direct way to natural language equivalents. In their specification of the three kinds of information to be represented in long-term memory, Rumelhart et al. (1972) focus on the types of relations (i.e., processes) used to organize them. The functional relations associated with concepts that are of primary importance are hierarchical class relations (dog is an animal) and property or characteristic relations (a dog has feet and a dog is dangerous, respectively). While these relations are neither necessary nor sufficient to define concepts, they provide crucial information in the manipulation (transformation) of the conceptual information in the memory system (Rumelhart et al., 1972, p. 204). Events, involving actors and objects, are organized in terms of single actions that are treated as act relations. Episodes, on the other hand, involve clusters of events, or action sequences, and are organized by special relations called
propositional conjunctions. Rumelhart et al. (1972) are careful to point out that the encoding of information into a memory structure form of representation is done in
14
Mancuso and Ceely
terms of secondary nodes. That is, initial encoding of a given concept, event, or episode would be specific and contextually bound. A process of generalization or abstraction is presumed to operate whereby similarly encoded instances of a relation or concept are retrieved, and features or characteristics common to each are identified. In this way the person experiences psychological equivalence between current input and the many events that contributed to establishing the relevant nodes in long-term memory. These features then enter the respective primary nodes as part of the abstract definition used to specify the given entry. Once established, primary concept nodes are linked to secondary nodes through an isa relation while, within event and episode nodes, secondary to primary node links are established by acts relations. Considerations concerning acts relations, while not emphasized in this paper, are not to be overlooked. Weimer (1977) successfully warns cognitive psychologists to be wary of the conceptual traps that are laid by adopting sensory analogues as the root metaphor for discussions of perceiving, knowing, and acting. In concord with Neisser (1976) and Powers (1973, 1978), Weimer emphatically reminds behavior scientists that acting and representation of acting are the bedrock of meaning. Even the decoding of verbal messages can be understood by relating the process completely to the listener's encoding operations relative to producing a speech stream like that which he is hearing and processing. Reprise. To develop a suitable description of a self as a memory we have thus far reviewed the following propositions regarding memory function: (a) Internal representations bear no structural resemblance to external events; (b) encoding of memory representations is discussed by reference to feature, property, and relations (isa and acts) storage; (c) properties and concepts are thought to be hierarchically ordered in memory representations; (d) general and specific knowledge are thought to be different in terms of concept, event, and episode nodes and in terms of the relations that link them. The Self as a Schema If one follows the propositions presented thus far, he would not expect that any self-defining response is stored, wholly formed as a discrete memory unit that is caused to "run o f f " to make its appearance in overt behavior. Instead, one thinks of stored processing systems, complete with dimensions, nodes, and various types of relations. Several other concepts that define such systems will help to describe retrieved selves. In their treatment of issues raised by Collins and Loftus (1975) and Rumelhart et al. (1972), Rumelhart and Ortony (1977) focus on the idea o f
Self as Memory Processing
15
the embedding of representational units within memory structures and on the distinction between the representation and storage of abstract versus specific information in memory. Adapting the concept used by earlier workers (Bartlett, 1932; Mancuso, 1970; Piaget, 1952/1963), Rumelhart and Ortony (1977) have replaced the term node with the term schema, attempting to emphasize both the overall organizational (i.e., fused gestalt) properties of this knowledge unit and its more generalized nature: "Schemata are data structures for representing the generic concepts stored in memory. They exist for generalized concepts underlying objects, situations, events, sequences of events, actions, and sequences of actions" (Rumelhart & Ortony, 1977, p. 101). The organization of schemata involves specification of both constituent elements and variables and the interrelationships among these elements. A hierarchical dimension to this organization can be derived directly from Collins and Loftus (1975), who proposed that "criterialities" are attached to the properties associated with a given concept node or schema. These criterialities order properties in terms of their importance in identifying a schema. In this way, embedded schemata or subschemata can be taken to represent hierarchically ordered generic, person-describing attributes or properties. For example, in the sdf-defining schema " J o h n /sa 'helpful' person," helpful constitutes a generic subschema, that is, a schema subsumed by a superordinate like socially good. Additional subschemata might include "sympathetic to those less fortunate," " w a r m , " "well informed," "responsible family man," etc., not all of which necessarily have the same criterialities attached. In an action situation the individual creates a self to correspond to the context of that situation. This self, a dominating schema (called a superordinate construction, when one uses Kelly's [1955] terms) is constructed from a stored representation of the superordinate construction and its subsumed constructions. The subschemata involved constitute variableslots that are filled during the (re)construction process. Continuing with the above example, John, driving down the highway and coming upon a woman standing beside a car with a flat tire, would instantiate the subschema "helpful" relative to his self; that is, he would pull over and offer to assist in changing the tire. However, should the individual standing by the car have been a man instead of a woman, it is quite conceivable that John would have continued driving on. In this context (i.e., motorist with a flat tire), the instantiation of the subschema "helpful" is contingent on the properties he has attached to men and to women motorists; that is, John has hypothetically attributed to his conceptualization of male motorists the properties "informed and capable of changing tires," while the properties
16
Mancuso and Ceely
"uninformed" or "indisposed to change flat tires" have been built into the event node relevant to women. The latter retrieves the "helpful" schema. Rumelhart and Ortony (1977) go on to draw a parallel between selfdefining schemata and a play in which "the internal structure of the schemata corresponds to the script of the play" (p. 101). This is, in fact, a theme treated extensively by Abelson (1975), Schank (1975a,b), and Schank and Abelson (1977). Taking issue with what they find to be an undue emphasis on the overly abstract nature of the models offered by a number of semantic memory theorists, Schank and Abelson focus instead on episodic memory, that is, memory organized around one's personal experiences (Tulving, 1972). The unit chosen to represent this specific or contextually bound information in memory is the episode, corresponding to an event or action sequence. However, as Rumelhart et al. (1972) and Rumelhart and Ortony (1977) have done, Abelson and Schank also propose the operation of a generalization process in memory (re)organization. This process is instigated through the grouping of contextually similar episodes. Features or properties common to the different episodes are then identified and organized in terms of property and concept interrelationships at a generic level. The resulting memory structure is a script that, like Rumelhart and Ortony' s (1977) schemata, can be thought of as containing variable slots corresponding to the range of property and concept instances contained in the various episodes subsumed by the respective scripts. Reconsider our example, John, as he conceptualizes his self as a "helpful person." One would look for and expect to find in that conceptual system a variable slot, "person in need of assistance." In that the script or schema acts as a context, the value assigned to the variable slot "person in need of assistance" is constrained by the value instantiated in the slot "problem situation." Specifically, once the situation is identified in terms of the subschema "motorist with a flat tire," the range of "person in need of assistance" values that will instigate John to enact his "helpful" subschema by assisting the motorist in changing the tire is correspondingly focused to include, hypothetically, elderly men and women motorists. Given, however, a different context but the same "problem situation," i.e., a motorist with a flat tire and an approaching storm or a motorist with a flat tire on an infrequently traveled road, John might conceivably expand the range of values of the "person in need of assistance" slot to include male motorists as well.
Remembering a Self Given the generic nature of a schema, or script, it follows that its utilization for the interpretation or enactment of given self-relevant
Self as
Memory Processing
17
behaviors will be a function of the context in which it is enacted. In some instantiations o f a self-schema a variable may be assigned a value that it wouM not take in another instance. Particular self-presenting episodes will show extensive interepisode variation, in that memory search can retrieve any number of encoded values to assign to the variables within the schema for that episode. Or, to state the negative, the model does not require that a self in a context must be a precise replica of the self in a previously occurring, "similar," context. Each created self may be an "original self," in that relational and attributional schemata may have different values within alternative contexts or when subsumed by different superordinate schemata. This feature becomes exceptionally important for explanation of the contradictions of self that are frequently encountered as one goes about reconstruing his own self or the self of others. One further source of "interepisode variation" in the enactment of self-relevant behaviors can also be discussed in very general terms. This source of variation involves the differential retrieval and utilization of episode information from long-term memory structures. Using quasineurological terms, Collins and Loftus (1975) propose that access of information in memory structures is the result of a "diffuse activation" process that begins with the "priming" of a given concept node (schema). Once a node has been primed (i.e., accessed), activation spreads out in parallel, along the links that organize the network, to other nodes (subschemata). In this way, the information accessed will be a function of (1) a "summative" process that operates as activation is diffused to different concept nodes and links in the network and (2) a "source specific activation released from a particular node" (Collins & Loftus, 1975). In the spread of activation through the network, superordinate connections and high criterial properties are particularly important in reaching the "activation threshold" (activation being a variable quantity) necessary for the node to be accessed. (Note that it is being assumed that activation in and of itself is not sufficient for access but must be summed to a given threshold value to ensure access.) Having proposed the concept of activation, one has reentered the behavior science territory where some of the giants of psychology have fallen. Propositions about activation should be useful in describing the motives that affect the instigation and direction of cognitive activity. The motive statements relative to the current concepts in this area have not been bravely put forth (secWeimer, 1977, pp. 294-295). Apparently the fate of concepts like drive and reinforcement has cast a shadow over the field. Acknowledging the justifiable caution about venturing a motive statement, while affirming the necessity for its specification, we note that the concept of spread of activation, as used by Collins and Loftus (1975), is compatible with a cognitive motivational position advocated by Mancuso (1977) and Weimer (1977).
18
Mancuso and Ceely
How do these considerations apply to advancing the discussion in systems in which theorists have proposed a distinction between episodic and generic memory structures (Abelson, 1975; Rumelhart et al., 1972; Rumelhart & Ortony, 1977; Schank, 1975a,b)? Assuming that a generic memory structure is to be utilized (i.e., for the interpretation of input stimuli or for the enactment of self-relevant behavior), a dominating or superordinate schema (a primary node) would be "primed," that is, accessed from long-term memory. However, given the principles for memory access proposed by Collins and Loftus (1975), the search process would not necessarily stop there. Rather, activation would spread via the relational system linking embedded subschemata and the episodes that can be subsumed by them (i.e., through the hierarchical class relations established by/sa links and the property relations established b y / s and has links). Because of the "summative" process involved in activation spread, not all the subschemata associated with a given superordinate schema or construct, nor the episodic information subsumed by them, would necessarily be directly or immediately accessed. Accessing would depend, in part, on the property and concept information stored in a given episodic memory structure and on the criterialities associated with the given property information. As mentioned previously, the properties stored will be a function. of the context in which an event was originally encoded (see Collins & Loftus, 1975, for a discussion of this as well as other factors related to the spread of activation). The important point in regard to the accessing process is that specific or episodic information may also be made available that will effect the instantiation of variable values within schemata. The episodic information accessed will vary as a function of the activation process and the context in which this process is initiated. In this way, the availability of episodic information elaborates the psychological context within which a given schema or generic knowledge structure is utilized. Schank and Abelson (1977) approach these issues by speaking of "scripts" as the unit of generic knowledge structure, and "isa" and "props" links as the means' for interrelating script and episode information. In regard to our previous example, John's instantiation of his self-defining schema "helpful," the enactment of self-relevant behaviors in relation to this schema would also be function of available episodic experience that could be accessed. For instance, should John recently have had a conversation with a friend who reported having difficulty with his car on the road and had been unable to get anyone to stop and help him, his (John's) access to and instatiation of the "helpful" subschema (in the context o f "male motorist with a flat tire or some other automobile malfunction") would become more probable. Alternatively, should John
Self as Memory Processing
19
recently have had an animated conversation with a female associate about the archaic practices of male chauvinism in modern society, his instantiation of the "helpful" subschema (in the context of "female motirist with a flat tire") would become less probable. Both of the above instances reflect the assumption that initially John has "primed" his "helpful" subschema and that, subsequently, activation has spread out via property and attribute relations to the episodic memory structures that can be linked to this subschema. In the first instance, the link is established with respect to the variable slot "problem situation," whereas, with respect to the second, it is established in relation to the variable slot "person in need of assistance."
A CAUSING SELF AND A CAUSED SELF A continuous, ongoing, re-cognizable self is re-cognizable in terms of the superordinate schemata available to that self-as-process to apply to those episodes circumscribed by those superordinate schemata. The superordinate schemata, as scripts that underlie the immediate retrieval process, are regarded as stable, while the relationships and variable attributes within the schemata may shift within separate instantiations of the superordinate schema. Schemata that represent the attributes and relational features of superordinate schemata are taken to be elements of storage, and in a reconstruction process these schemata become the retrieved elements that constitute the memory of self. Overt action also follows from these schematizations, in that the values of muscle-positional variables also are assigned as the flow of the psychological context occurs. The particular self-defining scripts, themselves storable schemata, can alter only in relation to a person's overall construct system. In that massive system changes are uneconomical, one postulates a conserving quality relative to superordinate schemata. Thus a person's self is recognizable in terms of his repeated instantiations of his superordinate schemata. Additionally, a quantitative feature induces recognizability of selfhood. A person's construct system would incorporate a finite number of schemata, or constructions. In the most basic case, for example, the most atomic schemata in a person's system, called constructs by Kelly (1955), or primitives by Norman, Rumelhart, and LNR (1975), would be limited to a number depending on the basic sensorimotor actions that a person can or has performed. Thus different individuals would employ different constructs, depending on the experiential opportunities for developing those concepts. In this way, self-defining enactments would be limited by the available, quantifiable generic schemata that are stored for retrieval, and to which episodic reconstructions may be referred. An aspiring
20
Mancuso and Ceely
gourmet who had not encountered Dalmatian wines could not assign them as a value relative to his "collect" schema (an act schema) as he enacts the "gourmet" schema (an/sa schema). His self-enactment, thereby, is limited by his not having available a schema for Dalmatian wines. Another feature contributes to stability of a self that is cognized by psychologists and other people. Although the values of variables within a superordinate schema are interchangeable, the dimensions or attributes, in that they also are schemata, are resistant to change. They can change only in terms of changes in their interior relations and their relations to other schemata. Thus, for example, when a person enacts a schema into which he incorporates his construct aggressive, we will expect that construct to remain stable in terms of its internal relational structure ("aggressive" is when one is "self-assertive"), and in terms of its relations to other schemata ("aggressive" behavior isa " b a d " behavior). Selves, then, are re-cognizable--to the person himself, to others, and to psychologists--because persons are cognizing units, and their selfdefining enactments reflect stable features of their cognizing system. A self-enactment has meaning, to a contextualist psychologist, in terms of that enactment's relationship to a convenient context. That context contains (a) the source of sensory input; (b) the person's cognitive system, complete with superordinate and subordinate schemata, property, and relational features; and (c) the values assigned to sensorimotor-related schemata by the person's "plans," or superordinate constructions.
The Causing Self Nevertheless, if one uses a model of self built on the conceptions of memory sketched above, he could look for a person's recognizable, continuous self. One would seek to define a self as the construct system that one "carries about" as he continuously meets the memory-evoking events that flow by his sensory apparatus to become integrated into that very system. A self-theorist would be particularly interested in the schemata that define whatever the person characterizes as hi s self; that is, one would be particulady interested in the individual's self theory (Bannister & Agnew, 1977; Epstein, 1973), just as he is interested in a psychologist's theory of selfhood. Behavior change intercessions could be seen as efforts to tie in to expectations, that is, the schemata imposed on sensory input. In some instances a behavior change specialist might be able to explicate the "grammar systems" (Mancuso, 1979) of a person's self and other person cognizing processes. In other instances, he can arrange means by which the person can experimentally apply his constructions in order to obtain confirmation or disconf'n'mation of their utility.
Self as Memory Processing
21
To treat the role-playing issues, one would regard the self created in a role-playing event as a self "caused b y " the person's schematic processing of input. In this sense, a role-played self has "epistemological parity" (Hendrick, 1977, p. 470) with a self that would be enacted in the context of a controlled social psychological study. They are not "the same 'caused' selves," but the "causing self"--the person's construct system--is the "same self." This stored, continuous, causing self would account for the retrieval of selves that vary relative to varied contexts. Similarity of self-presentation would be a matter of the person's cognizing of the stimulus input and is, in this way, not under an experimenter's control. External stimuli, from this perspective, determine the similarity of an event to the extent that "similar" constructions are retrieved to assimilate those stimuli. In short, the "causing" self creates situation similarity.
RESEARCHING THE SELF The future of self theory, if one accepts the argument that personality theory can be re-cognized in terms of the work of the more general cognitive theorists, can proceed through the study of personal construct systems. The work of George Kelly (1955) has already provided a basic framework for this venture. Kelly's work has been extensively elaborated by other personal construct theorists (Adams-Webber, 1979; Bannister & Fransella, 1971; Landfield, 1971, 1977; Mancuso, 1970). The use of Kelly's theory, understandably, has been impeded by Kelly's use of a paradigm that ran counter to the prevailing paradigms of mid-20th-century psychology (see Mancuso, 1977 Sarbin, 1977). It was the rare theorist (Sarbin et al., 1960) who could see the potential for describing personalities in terms of how individuals processed person-relevant information. With the rise of models that can deviate from the straightline causality concepts of mid-century mechanism, the person's "implicit personality theories" (Rosenberg, 1977; Wegner & Vallacher, 1977) become the core of descriptions of personalities. The general constructivist foundations of Kelly's personal construct psychology can be acceptable to current personality theoriests (Landfield, 1977). Several lines of research have shown that person- and self-defining schemata may be treated as the basic organization that is imposed over incoming person-relevant information. Kuiper and Rogers (1979; Rogers, Kuiper, & Kirker, 1977) have conducted a series of studies to show that the self-defining cognitive systems act as a generic knowledge structure or schema that is organized in terms of an amalgamation of self-descriptive dimensions. Similarly, Markus's (1977) work adds to the understanding of
22
Mancuso and Ceely
how self-schemata relate to the processing and organization of personrelevant behavior. Hastie and Kumar (1979) reported a study based on an extensive theoretical statement that relates principles o f cognitive organization and person perception. The model that they use to explain person-cognizing, developed from the general models proposed by Anderson and Bower (1972), would be directly analogous to the model proposed above to explain self. One would need, however, more work like that o f Kuiper and Rogers (1979) to determine if individuals interchangably use their self-cognizing organizations and their general person-cognizing organizations. With progress in studies of cognitive function, personality theorists will be better positioned to explicate implicit personality theories o f personality, which are assumed to underlie one's theory of self. The contextualist paradigm invites use. Contextuaiist theories of motivation (Mancuso, 1977; Weiner, 1972) require further development. The utility o f Piaget's (1932/1966, 1952/1963) explorations o f early schemata development has been amply demonstrated (Flavell, 1977; Wadsworth, 1978). One can now, it is hoped, look to further study of cognitive functioning to show us the parallels between a person's "discovery" o f a number system (Piaget, 1952/1965), his discovery o f persons (Wegner & Vallacher, 1977), and his discovery o f his self (Bannister & Agnew, 1977; Epstein, 1973). With this kind o f information we can develop a more meaningful view o f how the self-as-process is implicated in the regulation o f the self-as-object.
REFERENCES
Abelson, R. P. Concepts for representingmundane reality in plans. In D. G. Bobrow & A. Collins (Eds.), Representation and understanding: Studies in cognitive science. New York: Academic Press, 1975. Adams-Webber, J. Personal construct theory. New York: Wiley, 1979. Anderson, J. R., & Bower, G. H. Recognitionand retrieval processesin free recall. Psychological Review, 1972, 79, 97-123. Bandura, A. The self system in reciprocal determinism. American Psychologist, 1978, 33, 344-358. Bannister, D., & Agnew, J. The child's construingof self. In A. W. Landfield (Ed.), Nebraska symposium on motivation: Personal construct psychology. Lincoln: University of Nebraska Press, 1977. Pp. 99-125. Bannister, D., & Fransella, F. Inquiring man. Baltimore: Penguin, 1971. Bartlett, F. C. Remembering. Cambridge: Cambridge University Press, 1932. Bobrow, D. G., & Norman, D. A. Some principlesof memory schemata. In D. G. Bobrow & A. Collins (Eds.), Representation and understanding: Studies in cognitive science. New York: Academic Press, 1975. Bowers, K. S. Situationalismin psychology:An analysis and a critique. PsychologicalReview, 1973, 80, 307-336. Collins, A. M., & Loftus, E. F. A spreading activation theory of semantic processing. Psychological Review, 1975, 82, 407-428.
Self as Memory Processing
23
Collins, A. M., & Quillian, M. R. Experiments on semantic memory and language compreheusion. In L. W. Gregg (Ed.), Cognition and learning in memory. New York: Wiley, 1972. Pp. 117-137. (a) Collins, A. M., & Quillian, M. R. How to make a language user. In E. Tulving & W. Donaldson (Eds.), Organization of memory. New York: Academic Press, 1972. (b) Epstein, S. The self concept revisited. American Psychologist, 1973, 28, 404-416. Flavell, J. Cognitive development. Englewood Cliffs, New Jersey: Prentice-Hall, 1977. Harvey, J. H. Attribution of freedom. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New directions in attribution research. Hillsdale, New Jersey: Erlbaum, 1976. Pp. 73-96. Hastie, R., & Kumar, P. A. Person memory: Personality traits as organizing principles in memory for behaviors. Journal of Personality and Social Psychology, 1979, 37, 25-38. Hendrick, C. Role-taking, role-playing, and the laboratory experiment. Personality and Social Psychology Bulletin, 1977, 3, 467-478. Jenkins, J. J. Remember that old theory of memory? Well, forget it. American Psychologist, 1974, 29, 785-795. Kazdin, A., & Wilcoxin, L. Systematic desensitization and nonspeeific treatment effects: A methodological evalution. Psychological Bulletin, 1976, 83, 729-758. Kelly, G. A. The psychology of personal constructs. New York: W. W. Norton, 1955. Kintsch, W. Models for free recall and recognition. In D. A. Norman (Ed.), Models o f human memory. New York: Academic Press, 1970. Kuiper, N. A., & Rogers, T. B. Encoding of personal information: Self-other differences. Journal of Personality and Social Psychology, 1979, 37, 499-514. Landfield, A. W. Personal construct systems in psychotherapy. Chicago: Rand McNally, 1971. Landfield, A. W. (Ed.). Nebraska symposium on motivation: Personal construct psychology. Lincoln: University of Nebraska Press, 1977. Ma~honey, M., & Arnkoff, O. Cognitive and self control therapies. In S. Garfield & A. Bergin (Eds.), Handbook of psychotherapy and behavior change (2nd ed.). New York: Wiley, 1979. Maltzman, I. Orienting in classical conditioning and generalization of the galvanic skin response to words: An overview. Journal of Experimental Psychology: General, 1977, 106, 111-119. Mancuso, J. C. (Ed.). Readings for a cognitive theory of personality. New York: Holt, Rinehart & Winston, 1970. Mancuso, J. C. Current motivational models in the elaboration of personal construct theory. In A. W. Landfield (Ed.), Nebraska symposium on motivation: Personal construct psychology. Lincoln: University of Nebraska Press, 1977, Pp. 43-97. Mancuso, J. C. Counselling from a personal construct perspective. International Journal for the Advancement of Counseling, 1979; 1, 303-313. Mancuso, J. C., & Sarbin, T. R. A paradigmatic analysis of psychological issues at the interface of jurisprudence and moral conduct. In T. Lickona (Ed.), Moral development and behavior. New York: Holt, Rinehart & Winston, 1976. Pp. 326-341. Markus, H. Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 1977, 35, 63-78. Milgram, S. Behavioral study of obedience. Journal o f Abnormal and Social Psychology, 1963, 67, 371-378. Mischel, W. Personality and assessment. New York: Wiley, 1968. Mixon, D. Instead of deception. Journal for the Theory of Social Behavior, 1972, 2, 145-177. Mixon, D. Why pretend to deceive? Personality and Social Psychology Bulletin, 1977, 3, 647-653. Murdock, B. B., Jr., & Anderson, J. R. Encoding, storage and retrieval of item information. In R. L. Solso (Ed.), Theories in cognitive psychology: The Loyola symposium. HiUsdale, New Jersey: Erlbaum, 1975. Neisser, U. Cognitive psychology. New York: Appleton-Century-Crofts, 1967. Neisser, U. Cognition and reality. San Francisco: Freeman, 1976.
24
Maneuso and Ceely
Nisbett, R. E., & Wilson, T. D. Telling more than we can know: Verbal reports on mental processes. Psychological Review, 1977, 84, 613-624. Norman, D. A., Rumelhart, D. A., & LNR Research Group. Explorations in cognition. San Francisco: Freeman, 1975. Pepper, S. World hypotheses. Berkeley: University of California Press, 1942. Piaget, J. The moral judgment of the child. London: Kegan Paul, 1932. (In paperback: New York: Free Press, 1966.) Piaget, J. The origins o f intelligence in children. New York: International Universities Press, 1952. (In paperback: New York: Norton, 1963.) Piaget, J. The child's conception of number. New York: Humanities Press, 1952. (In paperback: New York: Norton, 1965.) Powers, W. T. Behavior: The control of perception. Chicago: Aldine, 1973. Powers, W. T. Quantitative analysis of purposive systems: Some spadework at the foundations of scientific psychology. Psychological Review, 1978, 85, 417-435. Rogers, T. B., Kuiper, J. A., & Kirker, W. S. Self-reference and the encoding of personal information. Journal of Personality and Social Psychology, 1977, 35, 677-689. Rosenberg, S. New approaches to the analysis of personal constructs in person perception. In A. W. Landfietd (Ed.), Nebraska symposium on motivation: Personal construct psychology. Lincoln: University of Nebraska Press, 1977. Pp. 179-242. Ross, M. The self-perception of intrinsic motivation. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New directions in attribution research. Hillsdale, New Jersey: Erlbaum, 1976. Pp. 121-141. Rumelhart, D. E., Lindsay, P. H., & Norman, D. A. A process model for long-term memory. In E. Tulving & W. Donaldson (Eds.), Organization o f memory. New York: Academic Press, 1972. Pp. 197-246. Rumelhart, D. W., & Ortony, A. The representation of knowledge in memory. In R. Anderson & W. Montague (Eds.), Schooling and the acquisition of knowledge. Hillsdale, New Jersey: Erlbaum, 1977. Pp. 99-135. Rychalak, J. F. A philosophy o f science for personality theory. Boston: Houghton Mifflin, 1968. Rychlak, J. F. Personality theory: Its nature, past, present and--future. Personality and Social Psychology Bulletin, 1976, 2, 209-224. Sampson, E. E. Psychology and the American ideal. Journal of Personality and Social Psychology, 1977, 35, 767-782. Sarbin, T. R. Contextualism: The wordview for modern psychology. In A. W. Landfield (Ed.), Nebraska symposium on motivation: Personal construct psychology. Lincoln: University of Nebraska Press, 1977. Pp. 1-41. Sarbin, T. R., Taft, R., & Bailey, D. E. Clinical inference and cognitive theory. New York: Holt, Rinehart & Winston, 1960. Schank, R. C. Conceptual information processing. Amsterdam: North Holland, 1975. (a) Schank, R. C. The structure of episodes in memory. In D. G. Bobrow & A. Collins (Eds.), Representation and understanding: Studies in cognitive science. New York: Academic Press, 1975. (b) Schank, R. C., & Abelson, R. P. Scripts, plans, goals and understanding. Hillsdale, New Jersey: Erlbaum, 1977. Shepard, R. Form, formation, and transformation of internal representations. In R. Solso (Ed.), Information processing and cognition. Hillsdale, New Jersey: Erlbaum, 1975. Skinner, B. F. Freedom and the control of men. American Scholar, 1955-1956, 25, 47-65. Snyder, M. Attribution and behavior: Social perception and social causation. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New directions in attribution research. Hillsdale, New Jersey: Erlbaum, 1976. Pp. 53-72. Stuart, R. B. Behavioral self-management: Strategies, techniques, and outcomes. New York: Brunner/Mazel, 1977. Tulving, E. Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of memory. New York: Academic Press, 1972. Pp. 76-101.
Self as Memory Processing
25
Tulving, E., & Thomson, D. M. Encoding specificity and retrieval processes in episodic memory. Psychological Review, 1973, 80, 352-373. Wadsworth, B. J. Piagetfor the classroom teacher. New York: Longman, 1978. Wegner, D. M., & Vallacher, R. R. lmplicitpsychology. New York: Oxford University Press, 1977. Weimer, W. B. A conceptual framework for cognitive psychology: Motor theories of the mind. In R. Shaw & J. Bransford (Eds.), Perceiving, acting, and knowing. Hillsdale, New Jersey: Erlbaum, 1977. Pp. 267-311. Weiner, B. Theorires of motivation. Chicago: Markham, 1972. Wolpe, J. Cognition and causation in human behavior and its therapy. American Psychologist, 1978, 33, 437-446. Wortman, C. B. Causal attributions and personal control. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New directions in attribution research. Hillsdale, New Jersey: Erlbaum, 1976. Pp. 23-52.