مدل های تطبیق جهانی غیرتجردی : چارچوبی برای بررسی اثر تمایز در حافظه آشکار و ضمنی

One very well-known memory phenomenon is the observation that if a specific item seems to be different or rare in any way from other items present in the same encoding context, this item is likely to be remembered more accurately. This phenomenon, named the von Restorff effect or the isolation effect, has been known since 1933 and been considered today as one of the possible ways to create distinctiveness. The aim of this article is to report that the current debate concerning its non-emergence in implicit memory tasks is directly due to a theoretical disagreement about the nature of human memory. In this paper, we conclude that non-abstractive global-matching models can provide an effective theoretical framework for the study of the distinctiveness effect with isolation both in implicit and explicit memory tasks.

One of the most remarkable feats of human memory lies in the ability to remember an object that has been identified as being different from other objects in any given environment. In 1933, Hedwig von Restorff conducted the first systematic series of studies designed to examine this deviant stimulus effect in memory. von Restorff presented participants with sequences containing one item that was different from all the other items presented in the same context and reported that this isolated item was recalled better than the others that were not isolated. von Restorff's research focused exclusively on recall and recognition (von Restorff, 1933, for a review of von Restorff's work, see Hunt, 1995). This phenomenon, referred to as the distinctiveness effect, has traditionally been considered to be characteristic of episodic memory. Fortunately, our understanding of human memory and the definition of the distinctiveness effect have evolved since 1933. Like every other research topic in cognitive psychology, it seems that independent theoretical debates about the nature and the organization of human memory have had an influence on our comprehension of the distinctiveness effect obtained with an isolation paradigm. In this paper: • we aim to show that the current debate about the existence of a distinctiveness effect (or lack thereof) on implicit memory is the result of a theoretical disagreement about the nature of human memory. Indeed, this disagreement mirrors the juxtaposition of the structural and functional approaches to the distinctiveness effect which conceive of the topography of human memory in different ways; • finally, we believe that there is a unified theoretical framework which can explain all the manifestations of the isolation effect in both implicit and explicit memory tasks. This approach, consisting of the so-called “non-abstractive global-matching models” (for example; MINERVA II described by Hintzman in 1986 and CHARM developed by Metcalfe Eich in 1982), seems to provide an effective framework to study the various manifestations of the distinctiveness effect. After outlining the problem of how our understanding of the distinctiveness effect has become fragmented across different studies, we will address the question of whether and why some authors believe that the isolation effect cannot be found in implicit memory tasks. Then, we will present the central principal of matching process and what are two constraints that are needed to be resolved in order to understand the mechanism behind the distinctiveness effect within an implicit memory task. Finally, we will explain how this mechanism works and why it can provide an accurate framework for the study of the distinctiveness effect. 1.1. How the distinctiveness or the isolation effects have been studied in the literature? According to Nairne, 1997 and Nairne, 2006, there is little agreement about the boundary conditions that create or underlie distinctiveness because distinctiveness has been defined in various ways over the years, sometimes being thought of as a property of a stored trace, sometimes as a retrieval cue and, in yet other cases, as a particular type of processing. Indeed, the growing number of experimental paradigms, coupled with the use of different materials, has led to the emergence of several different conceptions of the distinctiveness effect and its theoretical implications. In 1991, Schmidt reported four classes of research into distinctiveness covering the most representative experiments conducted up to that time. According to Schmidt, perceptual (McLaughlin, 1968) or categorical distinctiveness (Hunt & Mitchell, 1982) are operationalized only in within-subjects designs. Effects found first in recall and then, as the degree of incongruity increases, in recognition are therefore classed as primary distinctiveness effects. To define secondary-distinctiveness, he gives the examples of orthographical distinctiveness (Hunt & Elliot, 1980) and bizarre imagery (McDaniel & Einstein, 1986), as well as the generation effect (Slamecka & Graf, 1978), because of the fact that these types of manipulations can be obtained only when using within-subjects designs and because these effects are a function of the type of memory test used. He also reports studies conducted with nude pictures (Ellis, Detterman, Runcie, McCarver, & Craig, 1971) and pictures of traumatic events (Loftus & Burns, 1982) which provoke an emotional response that leads to the observation of distinctive memory performances. Finally, any manipulation in which study-test congruence is important and which can be obtained using both within and between-subjects designs is classified as ‘processing distinctiveness’. This class includes, for example, depth of processing (Craik & Tulving, 1975). In fact, beyond the need to classify the different ways in which distinctiveness is manipulated in memory, Schmidt (1991) clearly demonstrated (p. 525) that distinctiveness is associated with a range of different phenomena. He highlights a very importance distinction between these different kinds of distinctiveness effects, with some of them being operationalized in within or between-subjects designs, some of them necessitating study-test congruency and some emerging as a function of the type of memory test used. In addition, studies conducted since 1991 have shown that the distinctiveness effect can be obtained by isolating a much larger range of properties. For example, let us look in more detail at some examples of what Schmidt would term ‘primary distinctiveness’ studies that have been conducted since 1991. Distinctiveness can be created through the intrinsic visual context of the task: participants may have to learn a list of items, all of which are words written in a specific color apart from a single word that is written in another color (e.g.; Bireta, Surprenant, & Neath, 2008) or, alternatively, a list of words in which the size of one word is changed (Kelley & Nairne, 2001). Distinctiveness can also be generated by means of the extrinsic visual context: participants may have to categorize words presented in frames, of which a quarter is red and three-quarters are yellow (Oker & Versace, 2010). Other possibilities include temporal or serial distinctiveness in which a retention interval is generally inserted between photographic items (Neath, 1993), verbal stimuli (Neath & Crowder, 1990) or the induction of a temporal manipulation in the secondary-distinctiveness domain (Gounden & Nicolas, 2012). Other authors have extended the temporal distinctiveness hypothesis (Nairme et al., 1997 and Brown et al., 2009) to form a spatial distinctiveness hypothesis (Oker et al., 2009 and Guérard et al., 2010). Others studied the distinctiveness effects as an uniqueness of contextual cues with words (Park, Arndt, & Reder, 2006) and to examine the interaction between distinctiveness and fluency effects (Lloyd & Miller, 2011) and between familiarity and novelty among proverbs (Poppenk, Köhler, & Moscovitch, 2010). Some authors interested to use distinctiveness effect as a method for studying age-related differences (Smith, 2011, Bireta et al., 2008, Geraci et al., 2009 and Nicolas and Gounden, 2010). 1.2. How the distinctiveness or the isolation effects have been explained? While the distinctiveness effect been manipulated in different ways during the development of this field of research, we have also witnessed changes in the theoretical framework within which distinctiveness is discussed. For instance, Schmidt (1991) reports that in the early days of research into distinctiveness, Rundus (1971) tested a rehearsal-based explanation of the effect identified by von Restorff within the context of Atkinson and Shiffrin's (1968) model. Rundus (1971) claimed that distinctive events may be rehearsed more frequently than common events. Other authors (Jenkins and Postman, 1948 and Green, 1956) have pointed out that one intuitive explanation is that a distinctive item is conspicuous, perceived as salient and arouses surprise. This draws attention to it and results in enhanced memory. However, this idea was rapidly shown to be inaccurate (Dunlosky et al., 2000, Hunt, 1995, Pillsbury and Raush, 1943 and Kelley and Nairne, 2001). However, the emergence of models capable of explaining the cause of the distinctiveness (or isolation) effect has been dependent not only on “what” was being tested but also on “how”. Probably the most significant example in this regard is Eysenck's model (1979). This model, which specifically focuses on recognition, highlights the need for information that is concordant both with the study-encoding phase and the test phase. This unique information is remembered well because it is processed more thoroughly than non-distinctive stimuli. However, Schmidt (1991) rightly observes that while this explication would be consistent with the findings of semantic orienting tasks and explains the depth of processing effects, the model does not explain the effects of distinctiveness on free recall. Once again this is a model, which appears to explain the cause of a distinctiveness effect in one condition (concordance of information) and one task (recognition) only. In an attempt to develop a more thorough explanation, Reed R. Hunt, 1995 and Hunt, 2003 worked for several years on constructing an extremely comprehensive and complete distinctiveness hypothesis. In his view, distinctive processing is defined as the unique processing of an item at encoding that enhances the discriminability of this item at retrieval (e.g., Jacoby and Craik, 1979 and Lockhart et al., 1976). Alternatively, it can be defined as the impact on memory of a deviant stimulus occurring in a context of similarity. Hunt and McDaniel (1993) further argue that distinctiveness can be conceived of as a process of discrimination during retrieval following the detection of an incongruent event at encoding as a result of the combined processing of similarity and difference. According to these authors, if the similarity and difference between a stimulus and other stimuli in its vicinity cannot be compared then there will be no effect to be observed (also Hunt & Smith, 1996; for a review see Hunt & Lamb, 2001). Hunt (2003) was later to acknowledge that “these ideas are taken directly from theory and research on similarity and difference judgment (e.g., Medin, Goldstone, & Gentner, 1993) and applied to the encoding and retrieval processes in memory” (p. 812). Again, this model would appear to provide a perfect explanation of the effects observed in free recall. Nevertheless, such an explanation would suggest that this effect can be observed only in explicit memory tasks in which the use of intentional memory instructions or the mobilization of conscious recollection (Weldon & Coyote, 1996) are required in order to obtain it. Smith and Hunt (2000) assumed that distinctive processes would be ineffective in retrieval unless the original encoding context is reinstated. This implies that intentional memory instructions must be administered at test in order to reinstate the original context. Indeed, the importance of contextual information in the distinctiveness effect has been frequently alluded to in many accounts of serial position effects in free recall. For example, according to the contextual retrieval hypothesis (Glenberg et al., 1980), the efficiency of recall depends on the efficiency of the reactivation of the encoding context of the items by the test context, and therefore on the specificity of the encoding context. 1.3. Is the distinctiveness or the isolation effect really task dependent? In order to answer this question, in 1998, Rajaram reported two experiments in which she manipulated distinctive dimensions of encoded stimuli in a variant of the R/K paradigm. In experiment 1, the conceptual dimension of the to be remembered homographs (e.g. bank) was manipulated by requiring subjects to encode the dominant (e.g. money–BANK) or the nondominant (river–BANK) meanings. In experiment 2, the perceptual dimension was manipulated by presenting orthographically distinctive (e.g. subpoena) or orthographically common (sailboat) words. An advantage of conceptually salient (dominant meaning) items and perceptually distinctive (orthographically distinctive) items was observed only in certain remember responses but not in know responses (for a review of the R/K paradigms, see Tulving, 1985 and Gardiner, 1988). Gardiner (1988) hypothesized that “remember judgments are based on conceptual processing arising from the episodic memory system and know judgments are based on perceptual processes carried out by the procedural memory system” (p. 71). “These effects of conceptual salience and perceptual distinctiveness were observed only on remember responses, not know responses” (p. 76). For Rajaram (1998), these results support the hypothesis that the processing of distinctive or salient attributes boosts the recollective component of explicit memory. The participants in these experiments were explicitly asked to identify on a rating scale how closely each word was related to the other. They were also informed that they would subsequently be asked to perform a memory task. Another study (Geraci & Rajaram, 2002) reported similar results revealing the effect of orthographic distinctiveness on an explicit but not on an implicit test. In this study, they did not found the orthographic distinctiveness effect in the indirect test of word fragment completion, but they succeeded to observe this effect within direct tests of word fragment cued recall and free recall. According to authors, “the distinctiveness effect depends on direct reference to study context” (p. 273). Finally, in two experiments conducted by Smith and Hunt (2000), half the participants were asked in a study phase to judge the similarity of highly similar items, while the other half were asked to make difference judgments. They were then tested using explicit (cued recall) and implicit memory tests (word-association and category exemplar production). As expected, performances on the explicit tests were better for the difference than the similarity judgments, whereas the manipulation affected the implicit tests only when the subjects claimed to be aware of the relationship between study and test. For Smith and Hunt (2000), these results demonstrate the critical role of cues provided by intentional instructions at the time of testing. When considering this finding, we have to take account of the importance of the conscious recollection that can occur between the study and test phases. Consequently, according to Weldon and Coyote (1996), because category-production and word-association tests do not require participants to discriminate between items in the same way that item-specific study tasks do, manipulations affecting distinctive processing will not transfer to implicit tests. Here, we are reaching the crux of the problem. If distinctive processing consists of the unique processing of an item at encoding in a way that enhances the discriminability of this item at retrieval, why is it not possible for manipulations affecting this distinctive processing to be transferred to an implicit memory test? What is in the nature of implicit memory which can make it independent to reflate this “combined processing of similarity and difference at encoding” as presented by Hunt and McDaniel (1993)? These issues are in some respects similar to a related debate about whether or not context effects can be found in implicit memory (an issue that has been widely discussed since Godden & Baddeley's experiment with underwater divers in 1975). This question of whether semantic memory is dependent on or independent of encoding context has been a subject of debate ever since Tulving's (1983) “synergistic ecphory” theory. Many studies of human memory conducted in the field of cognitive psychology have questioned the abstractive nature of conceptual information. The classical view, referred to as the multiple system perspective or structural approach (e.g., Anderson, 1983, Cohen and Squire, 1980, Squire, 1987, Tulving, 1985 and Tulving, 1995), claims that human memory consists of multiple systems. The most important of these are episodic memory for autobiographical information and semantic memory for conceptual information. The structural conception of memory typically assumes that our selective attention isolates critical information during perception by filtering out background situations (e.g., Kruschke, 1992, Tulving, 1972 and Trabasso and Bower, 1968; for a review, see Yeh & Barsalou, 2006). For these reasons, conceptual knowledge is reported to be inherently amodal, with the conceptual system and perceptual system fulfilling distinct functions. Within this perspective, Tulving imagines that knowledge results from an abstraction process from episodic to semantic memory. Consequently, conceptual information in semantic memory remains purely amodal since the contextual information provided by our past experiences is lost during this abstraction process. Like Rajaram in 1998, most experimental designs and underlying paradigms studying distinctiveness effect are based on dissociation (episodic versus procedural memory systems). We are therefore justified in asking whether it is appropriate to use tasks employed to demonstrate dissociations based on the preconceived idea that several different memory systems exist when constructing an experiment designed to investigate the distinctiveness effect. In our view, such an approach would, in most cases, provide support for the view that the distinctiveness effect can indeed not be observed in implicit tests.

Our main aim in this paper was to contribute to the literature on the distinctiveness effect by introducing the idea that the claim whether or not the distinctiveness effect (or the isolation effect) can be found with an implicit (or indirect) test is especially based on a topographical point of view regarding the nature of human memory. Moreover, we proposed in this paper why probabilistic models of memory provide us an efficient and accurate framework for the study of distinctiveness effect with isolation paradigm within implicit memory tasks. At the same time, the debate as to whether or not distinctiveness effects can be found in implicit memory tasks is as yet unresolved. Global-matching models provide persuasive arguments indicating why distinctiveness effects might be found in implicit memory tasks. It should therefore now be acknowledged that distinctiveness is not a property of a memory system but rather the result of a global difference between memory traces.