Draft of: Toth, J. P. & Reingold, E. M. (1996). Beyond perception: Conceptual contributions to unconscious influences of memory. In G. Underwood (Ed.), Implicit cognition (pp. 41-84). Oxford: Oxford University Press.
Beyond perception: Conceptual contributions to
unconscious influences of memory
Jeffrey P. Toth
Eyal M. Reingold
Table of contents
B. Perceptions of current research on unconscious influences of memory.
C. A selective review of early research on unconscious influences.
D. Contemporary evidence for conceptual unconscious influences of memory.
1. Conceptual influences on "perceptual" indirect tests.
i. Levels of processing and self generation.
ii. Cross-form priming.
2. Conceptual influences on "conceptual" indirect tests.
i. Semantic cues.
ii. Unconscious influences from new and pre-existing associations.
3. Conceptual unconscious influences in context.
i. List-wide context and the priming of interpretation.
ii. Text as context.
iii. Orientation to the past as a context for the present.
E. Summary of contemporary research.
F. Conclusions: Getting "set" for future research on unconscious influences.
"Whenever knowledge of the possible interpretation or conceptualization of something helps in perceiving that thing, we say the processing is conceptually-driven. That is, the process starts with conceptualization of what might be present and then looks for confirming evidence, biasing the processing mechanisms to give the expected result. ... Conceptually driven processing and data-driven processing almost always occur together, with each direction of processing contributing something to the total analysis" (Lindsay & Norman, 1977, p.13).
In what ways can a person be unconsciously influenced by the past? In the last quarter century, there has been a great deal of research directed at answering that question. However, the majority of that research has focused on the transfer of prior perceptual or data-driven processing. Much less emphasis has been placed on conceptual factors in the production of unconscious influences even though events are never processed in conceptual isolation--without a meaningful or interpretive context. The purpose of this chapter is to address the question of whether memory for prior meaning-based processing can unconsciously influence subsequent thought and behavior--that is, the question of "conceptual priming." More generally, we describe how conceptually-driven or "top-down" processes may play a role in all forms of priming, both perceptual and conceptual.
We had three specific goals in writing this chapter. First, we wanted to point out the inadequacies of indirect tests as measures of unconscious influences, especially as they relate to the question of conceptual priming. This goal seems important because, if we are to truly understand the way prior conceptual processing affects subsequent thought and behavior, we need to begin with unequivocal demonstrations of such effects. Our second goal was to identify some of the conditions necessary for the transfer of prior meaning-based processing. To anticipate, we argue that conceptual transfer depends on reinstating the context surrounding the prior event in which conceptual processing took place. Our third goal was to point out what we see as an important but missing factor in contemporary theories of unconscious influences--the subject's set or "level of attunement" (Bransford, McCarrell, Franks, & Nitsch, 1977) for task processing. We relate cognitive set to current notions of "executive" processing (e.g., Norman & Shallice, 1986), but argue that, rather then being absolute, unconscious (automatic) influences are relative to the context set by intentions (Jacoby, Ste-Marie, & Toth, 1993; Neumann, 1984).
The chapter is organized as follows. We begin by giving a brief overview of research on unconscious influences of memory ("implicit memory"). We then take a selective look at early research on unconscious influences. Aside form general historical interest, the purpose of this section is to provide a background for evaluating contemporary research on unconscious influences of memory. Early investigations provide a nice contrast for contemporary studies because, whereas the issues of set and context were viewed as critical by early investigators, these issues have been largely ignored in modern studies. The largest portion of the chapter is devoted to reviewing contemporary evidence for unconscious conceptual influences of memory. Here we focus mainly on indirect tests of memory and on evidence from the process dissociation procedure. We conclude by assessing how our review informs theorizing about the nature of memory; and, what it implies for the nature of conceptual unconscious influences.
B. Perceptions of current research on unconscious influences of memory
An influential study in the modern characterization of unconscious influences was published by Winnick and Daniel in 1970. In the first phase of their Experiment 2, subjects were shown words, pictures of objects, and definitions, and were required to pronounce the words depicted by the different stimuli. Subjects were then asked to recall the words and, as expected, words generated from pictures and definitions were better recalled than those evoked from the printed words themselves. However, a very different pattern emerged in the following test of tachistoscopic identification. For this test, words that had been pronounced at study were flashed very briefly for identification although "no indication was given that the same words would be used." Results showed that performance in tachistoscopic identification was almost the opposite of that in free-recall: words originally presented in printed form showed the greatest facilitation (i.e., the lowest identification thresholds), whereas words previously elicited by definitions or pictures had thresholds nearly indistinguishable from nonpresented control words. Winnick and Daniel (1970) described their findings as indicative of "a kind of perceptual sensitization" (p.80).
Winnick and Daniel's (1970) results were replicated in a number of later studies using word identification as well as other indirect tests such as stem and fragment completion (e.g., Jacoby, 1983b; Jacoby & Dallas, 1981; Graf & Mandler, 1984; Weldon & Roediger, 1987; for a review see Roediger & McDermott, 1993). In addition to words generated from pictures or definitions, these experiments showed that other conceptual manipulations such as levels of processing (Craik & Lockhart, 1972), which are known to have large effects on recognition and recall, had little or no effect on indirect tests using physically degraded ("data-limited") retrieval cues. Similar to the account given by Winnick and Daniel (1970), one of the most straightforward explanations of these effects appealed to "transfer-appropriate processing" the notion that memory reflects the overlap in processing requirements at encoding and retrieval (Jacoby, 1983b; Kolers & Roediger, 1984; Morris, Bransford, & Franks, 1977; Roediger & Blaxton, 1987). Unlike generating, reading isolated words requires that specific visual patterns--the printed words--be perceptually analyzed. Similarly, tests such as word identification and stem completion require the subject to perform a visual or "data-driven" analysis of the degraded test cues in order to garner evidence as to their identity. Thus, because of their prior visual processing, read words show a larger benefit than generated words in subsequent identification or completion tests. Also, because semantic elaboration occurs subsequent to identification, levels of processing has little effect on these tests.
At around the same time the above findings were being reported in normals, curious findings were emerging in research with memory-impaired populations such as those with Korsakoff amnesia (for a review see Moscovitch, Vriezen, & Gottstein, 1993). In a classic series of studies, Warrington and Weiskrantz (1968, 1970, 1974) found that amnesics, who showed little recollection of prior experience on traditional tests such as recognition and recall, showed nearly normal retention when provided with fragments of the target items (e.g., word stems). Warrington and Weiskrantz (1974) used these results to argue that amnesia reflected a deficit in retrieval marked by an extreme susceptibility to interference; and that this deficit could be ameliorated by providing sufficient environmental support at retrieval. Further evidence, however, suggested a more complex account of the memory impairment in amnesia. Most important, amnesics given fragmented retrieval cues did not report that they were remembering previous events but rather claimed to simply be guessing. Additional research validated the amnesic's subjective reports by showing that their memory performance was sensitive to the instructions given at test. When provided with word stems, for example, and told to complete them only with items they remembered being presented earlier, they performed very poorly. However, performance with these same cues improved dramatically when the patients were told to complete the stems with "the first word that came to mind" (e.g., Graf, Squire, & Mandler, 1984).
Taken together, the experiments with normals and amnesics provide two critical findings that will have to be accounted for by any general theory of memory. First, contrary to the notion that elaborating the meaning of an event always leads to better memory (e.g., Craik & Lockhart, 1972), performance on certain tests can be relatively immune to the effects of prior meaningful elaboration. More generally, this finding suggests that there is no complete or absolute memory for prior events. As captured by the notions of encoding specificity (Tulving, 1983) and transfer appropriate processing (Morris et al., 1977), memory performance always reflects the interaction between encoded information and information available in the retrieval environment. The second finding, perhaps related to the first, is that an intentional orientation to remember is not a prerequisite for demonstrating the effects of prior experience. Memory can be expressed incidentally or unconsciously as a facilitation in the identification (Winnick & Daniel, 1970), production (Gardiner, Boller, Moreines, & Butters, 1973), or reprocessing of stimuli (Cohen & Squire, 1980), or as a bias in the interpretation of ambiguous stimuli (Jacoby & Witherspoon, 1982). It is important to emphasize the impact these results have had on our conceptualization of memory. Prior to these findings, memory was a conscious activity and meaningful elaboration was the best way to enhance it. After these findings, memory could occur unconsciously and when it did, prior perceptual operations seemed to best mediate the effects of the past.
Could it be that conscious uses of memory are inherently conceptual and unconscious influences inherently perceptual? At least in broad strokes, this is the impression one gets from surveying both "structural" and "processing" theories of memory. For example, Tulving and colleagues (Tulving, 1983, 1985; Tulving & Schacter, 1990) have argued that explicit memory is subserved by an episodic memory system which records the context and meaning of prior events, whereas implicit memory or "priming" is based on the operation of a "presemantic, perceptual representational system." Similarly, Squire (1992) states that repetition priming "occurs as changes in early-stage perceptual processing systems in posterior cortex, before conceptual or semantic analysis is carried out and before the involvement of the hippocampal formation and the development of declarative memory" (p.212). Finally, Roediger (1990) states that "most explicit tests draw on the encoded meaning of concepts, or on semantic processing, elaborative encoding, and the like... [M]ost implicit tests (and all those tests in which impoverished perceptual information is presented) rely heavily on the match between perceptual operations between study and test" (p.1049).
It seems likely that at least part of the current emphasis on perception stems from the fact that, as was the case in Winnick and Daniel's (1970) Experiment 2, many of the studies investigating dissociations between memory and awareness confound task demands with test instructions. That is, tests using indirect memory instructions have also tended to have perceptual (or lexical) task demands, whereas tests using direct memory instructions have tended to ask for information about the meaningful aspects of prior experience, although such conceptual demands are often tacit. Recently, investigators have begun to correct this confound and are beginning to investigate the possibility that memory for prior conceptual processing may produce unconscious influences (e.g., Blaxton, 1989). Surprisingly, one of the earliest modern demonstrations of conceptual priming can be found in the same paper that had such a large role in launching the current focus on perceptual unconscious influences. Even more interesting, that demonstration was obtained using what today is considered by many to be the definitive data-driven test: word ("perceptual") identification.
Although Winnick and Daniel's (1970) second experiment had a large impact on theoretical approaches to memory, their first experiment produced a finding that is not as easily incorporated. In their Experiment 1, two groups of subjects generated a limited number of the United States (e.g., Arizona, Kentucky, etc.). One group performed this task visually by writing the states on a sheet of paper while another group performed an auditory version of the task by naming the states aloud. Following the generation task, the subjects were given an indirect test of tachistoscopic identification for five of the states they had named (or written), five unnamed states, and five control words. Results showed that previously generated states were identified at a threshold significantly lower than that for unnamed states and matched control words. Interestingly, the reduction in threshold was the same whether the subjects had originally written the states or simply named them aloud. Even more important, the unnamed states, although not physically presented at encoding, were identified more readily than control words. Winnick and Daniel cast their results in terms of a subject's "response set" and concluded that "in order for prior experience with words to facilitate their subsequent [identification], it is not necessary that the words be seen in the first stage" (p.76).
What can we make of these results? By not informing their subjects that test items included previously encountered words, Winnick and Daniel's experimental methodology conforms to current standards for assessing implicit memory. Moreover, they used a test that is generally considered to be perceptual and data-driven, yet showed priming for words that had not been presented in the experimental context. The finding of enhanced identification of nonpresented words strongly conflicts with current theoretical approaches which uniformly identify priming effects in word identification with perceptual processes or presemantic memory systems (Roediger, 1990; Tulving & Schacter, 1990). Of course, a proponent of those theories might argue that Winnick and Daniel's results offer only weak support for unconscious conceptual effects because conscious response strategies would have produced effects in the same direction. Subjects were surely aware that the majority of test words were from the same conceptual category and this knowledge, this awareness, may have altered the nature of the test. We agree that the pattern of data obtained by Winnick and Daniel may have been influenced by consciously-controlled processes but question whether more recent demonstrations of priming--both conceptual and perceptual--do not reflect similar influences. But conscious contamination notwithstanding, we believe that what distinguishes Winnick and Daniel's Experiment 1 from more recent investigations of conceptual priming is their use of a retrieval environment that contained a coherent set of related words. By having the majority of their test words be US states, Winnick and Daniel may have created a retrieval context--a "response set"--specific enough to recruit the conceptual operations used in the original generation task.
Why are we framing contemporary memory research with a study conducted over twenty years ago? We believe the Winnick and Daniel (1970) paper encapsulates an important change in the way memory is studied. Indeed, their paper can be viewed as signifying a "paradigm shift" in memory research--not in the Kuhnian sense, but rather methodologically--from an emphasis on processing meaningful, interrelated material (Experiment 1) to processing unrelated items presented "out of context" (Experiment 2). Modern studies of unconscious influences are more similar to Winnick and Daniel's Experiment 2 than their Experiment 1. However, if we are correct that the retrieval context and the subject's set play a critical role in priming, then there may be something to learn from looking more closely at the theoretical precursors of Winnick and Daniel's first experiment.
C. A selective review of early research on unconscious influences
Research and theorizing on unconscious influences has a long history in psychology, perhaps beginning with Helmholtz's unconscious inference. However, it is probably in Freud's psychoanalytic approach that non-perceptual unconscious influences were given their strongest statement (see Erdelyi, 1985). Freud was interested in how hidden anxieties, impulses, and expectancies drive thought and behavior. In this sense, much of psychoanalysis can be viewed as an attempt to understand unconscious influences mediated by the meaning of prior events. Indeed, phenomena such as action slips and slips of the tongue, as well as repression phenomena such as memory distortions, all suggest unconscious conceptual influences. As noted by others, (e.g., Khilstrom, 1987), contemporary views of unconscious influences bear more relation to automaticity as described in the attention literature (e.g., Shiffrin & Schneider, 1977) and to habit as described by James (1890) than to the psychoanalytic unconscious. Nevertheless, there are interesting similarities between the psychoanalytic approach to unconscious influences and the current cognitive approach. For example, the study of action slips continues to inform theories of attention, memory, and "executive processes" (Norman, 1981; Norman & Shallice, 1986; Reasons, 1984, 1990) and unconscious influences from emotional (Isen & Diamond, 1989; Niedenthal, 1992) and motivational sources (Bargh, 1990; Jacoby & Kelley, 1990)--all of which are related to a subject's conceptualization of prior events--are central topics in social cognition.
Another similarity concerns the methodological strategies used to infer unconscious influences. For both the cognitive and psychoanalytic approaches, unconscious influences are often inferred from the structure that subjects impose on ambiguous stimuli under nondirected ("implicit") test conditions. In this sense, an indirect test is similar to a Rorschach and can be as potentially difficult to interpret (see Jacoby, Toth, Lindsay, and Debner, 1992). There are, however, important differences in the way psychoanalytic and cognitive approaches treat performance on tests such as the Rorschach. One that is particularly interesting is that, like many indirect test cues, a Rorschach ink blot is more of a perceptual, than a conceptual, stimulus; yet, responses to a Rorschach have rarely been used to infer the nature of perceptual processing or the status of a subject's perceptual memory systems. Rather, they are used to interpret a subject's needs, fears, and motivations; in short, their conceptual state. What this difference in emphasis suggests is that it is not the nature of the stimulus (perceptual or conceptual) that determines the subjects' response to it, but rather the question being asked about that stimulus. We return to this issue below. For present purposes, we simply want to note that the first systematic experiments using what are now called perceptual indirect tests were not directed at perception per se, but rather in how perception is influenced by nonperceptual factors such as drives, emotions, expectancies, and a somewhat vague entity known as "set". A nice example is provided by the work of Sanford.
Sanford (1936) was interested in the effects of motivational factors on "imaginal" processes. To investigate the issue, he gave children word-association and ambiguous-pictures tests immediately before or after a regular meal and found significantly more food-related responses (e.g., "meat") on both tests from the hungry group. A follow-up study using a more rigorous design (Sanford, 1937) compared the performance of fasting and nonfasting college students on five tests: word association, interpretation of ambiguous pictures, chained associations, completion of fragmented pictures, and stem completion. Interestingly, Sanford went to great lengths to disguise the nature of his tests: "[T]o be sure that the giving of food responses was not influenced by some conscious and deliberate determination... [a]ll subjects were told that the experiments had to do with speed of response under different conditions" (p.147). Thus, in the word association tests, subjects were told to respond "as quickly as possible with the first word which came to mind" and for the chained associations tests they were told to "write as rapidly as they could" (p.146). In addition, the critical tests followed a number of "irrelevant 'speed tests'" (p.148).
As in his previous study with children, Sanford found that the fasting college students gave the highest number of food-related responses and, for the non-fasting students, food-related responses increased as a function of time since their previous meal. How would we explain Sanford's findings in current parlance? Can hunger be viewed as a priming manipulation? If so, one might expect priming on conceptual tests (e.g., word association), but Sanford also found facilitation on data-driven tests (e.g., stem completion) suggesting some perceptual mediation. Are motivations and expectancies perceptual or conceptual? Perhaps Gibson (1941), in his extensive, critical review of the concept of set, summed it up best:
"At about the turn of the century it began to be realized that the events in a psychological experiment--reactions, associations, judgments, or thoughts--were determined by something other than the reportable events themselves and that this was itself a psychological problem. ... [T]his phenomenon of set established by the Wurzburg psychologists was conceived as something quite distinct from association, from reaction, and from ordinary conscious content. It was not any of these things, but was a determiner of them all." (Gibson, 1941, p.783).
Sanfords' work represents one of many experiments directed at the issue of set and it is in this context that a variety of indirect tests were invented. For example, Rees and Israel (1935) used anagrams and Siipola (1935) used "skeleton words" (word fragments) to study the effects of set on the organizational nature of perception. In one study, Rees and Israel (1935) gave subjects a series of 40 anagrams, the first 20 of which could be completed only with words from a specific category ("eating solutions" or "nature solutions"). By use of this "training series" they hoped to influence the solutions given to the 20 anagrams in the second half of the test; these anagrams could also be completed with categorically related words but allowed for unrelated completions as well. Not surprisingly, they found that the categorical set established by the first 20 anagrams had a large influence on subsequent performance; in comparison with control subjects, those given the training series were more likely to solve ambiguous (multiple completion) anagrams using words congruent with the categorical set established in the first half of the test. Interestingly, Rees and Israel used subjective reports to determine "the degree to which the sets involved a conscious attitude of search for the related words. Of the 34 subjects, only 11 reported awareness of such an attitude accompanying realization of the relatedness of the words. For the others, the reports indicated that the sets had operated automatically" (p.11). Of course, methodological issues limit the conclusions that one can draw from this research. Nevertheless, it is of interest in showing apparent automatic influences of conceptual information on an ostensibly data-driven test, in its use of self-report techniques to assess awareness, and in the questions it raises for current methodological approaches to unconscious influences.
The investigation of set reached its zenith in the research movement known as the "newlook" in perception. The central idea in the newlook movement was that perception was an act of categorization. As stated by Erdelyi (1992): "Perception, according to the New Look, was an interaction phenomenon: Organismic factors such as needs, expectancies, and defenses interact with the input, and perception is the end result." Newlook researchers pushed set to its logical extension by investigating the effects of a variety of non-perceptual factors, such as emotional state and economic value, on perception. One classic example was the finding that coins were judged to be larger than metal or cardboard discs that were objectively equivalent in size; also, identical coins were judged as larger by children from families with low, as compared to high, socio-economic status (Bruner & Goodman, 1947).
Although it has been claimed that the newlook movement died for lack of empirical support (Greenwald, 1993; but see Erdelyi, 1992), its influence nevertheless spilled over into more cognitive research. For example, in a study conceptually similar to Winnick and Daniel's "response-set" experiment, Corteen and Wood (1972; Corteen & Dunn, 1974) presented subjects with a list of words, some of which were city names that were paired with a mild electric shock thereby causing a increase in autonomic nervous activity as measured by galvanic skin response. In a subsequent dichotic listen task, shock-associated city names, city names that had not been previously presented, and control words were presented on the non-attended channel. Despite the fact that both shadowing performance and direct memory testing revealed no awareness of their presentation, shock-associated words produced an increase in autonomic activity in comparison to control words. More importantly, words (i.e., city names) neither previously presented nor associated with shock, but semantically related to shock-associated words, also increased autonomic activity over baseline. Corteen and Wood's (1972) findings, like those of Winnick and Daniel (1970) and the early newlook researchers, suggest that unconscious retrieval processes can be affected by the conceptual information available in the retrieval environment as well as by the motivational concerns subjects bring to the experimental setting.
What does this brief historical survey tell us about the nature of unconscious influences? We see four facets that are worth emphasizing. First, in addition to prior perceptual and conceptual processing, unconscious influences may originate from a variety of sources including long standing motivational and emotional concerns. Although this point is recognized today--most clearly by social psychologists--it has yet to be incorporated into contemporary theories of memory. We believe that these various sources of influences are contextually embedded in the events that later serve as the basis for unconscious influences of memory. The second point is that unconscious influences are most likely to occur in retrieval environments that reinstate the prior context in which that earlier processing took place. This is especially true for unconscious influences of prior conceptual information. Again, although this point may seem obvious, one would not get that impression by looking at the retrieval environments used in modern investigations of priming. Third, tests and test stimuli by themselves do not determine the nature of the psychological processes that are used in response to those tests and stimuli. Thus, not only is it questionable whether indirect tests provide process-pure measures of unconscious influences (Toth, Reingold, & Jacoby, 1994), it also seems unlikely that "perceptual" or "conceptual" indirect tests uniformly provide process-pure measures of perception or conception (Hunt & Toth, 1990).
The final, most important, point is that a subject's cognitive set is an integral constituent of all psychological acts including those that reveal unconscious influences of memory. Although set has been interpreted in a variety of ways (see Gibson, 1941), its most basic meaning is a state of anticipation or a readiness to respond to stimuli in a particular way (e.g., Woodworth & Schlosberg, 1958). As a form of expectancy, set is similar to descriptions of conceptually-driven processing (see quote on p.3 by Lindsay & Norman, 1977). Also, by incorporating the notions of task-orientation and goal-state, cognitive set is akin to modern characterizations of executive processes. Models of executive processes attempt to characterize how a large number of distinct neurocognitive processes are organized and co-ordinated so as to produce coherent, goal-oriented behavior (e.g., Moscovitch, 1992; Norman & Shallice, 1986; Stuss, 1991). However, most contemporary models draw a sharp distinction between executive control processes and the more automatic processes (or "special purpose processing systems") through which the "mechanics" of perception, thought, and behavior are actually carried out. That is, they assume that automatic processes are highly modular and thus independent of the context in which processing occurs. In contrast, we assume that automatic (and unconscious) processes are sensitive both to the processing context and to the intentional state of the system as a whole. In our view, memory for prior processing (perceptual, conceptual, emotional, or motivational) can influence thought and behavior, but only in the context of an overriding intention, goal, or set.
A nice example of the context specificity of automatic influences is provided in a personal anecdote related by Shallice (1982). He describes walking into a well known room and finding himself making an unusual "pulling movement" with his arm. At first this movement was unintelligible to him, but he soon realized that the light in that room was controlled by a cord which had become hooked up in a cupboard door. Shallice (1988) looks upon this incident as showing how the "initiation and execution of this action [is] not normally controlled by a conscious intention to execute it" (p.328). True enough, but this analysis fails to acknowledge the context in which the action occurred. It seems likely that professor Shallice does not make this unusual pulling movement at random, or even upon walking into most rooms. Rather, we expect that this movement would occur only in the context of that, or a highly similar, room. We would further argue that the movement was triggered not only by the specific room, but also by a specific goal; to turn on the light.
We propose that cognitive set--the executive, "top-down" processes involved in maintaining a goal or task orientation--forms an integral aspect of the operations involved in most, if not all, memory experiments. Some of the implications of this perspective are explored near the end of the chapter. First, we examine contemporary evidence for unconscious conceptual influences of memory. Few modern studies have investigated the relationship between set and memory (but see Bransford et al., 1977), and so we say little about that relationship for much of our review of contemporary empirical findings. However, it is possible to cast current results in terms of context, which can be viewed as one of the major cognates of set. As our review progresses, we increasing find contextual factors to be critical for the production of unconscious influences of memory.
D. Contemporary evidence for conceptual unconscious influences of memory
Studies of unconscious influences of memory are often separated into two classes based on the type of cue provided at retrieval. The first, most popular, class of tests are those that measure responses to single, often physically degraded, stimuli presented in isolation; examples of such perceptual indirect tests include word identification and fragment completion. The second major class of tests are those that measure responses to semantic cues in the absence of perceptual information that could directly elicit the target response; tests of this type are referred to as conceptual indirect tests and include word association and category-examplar generation. A final set, often treated as a variety of conceptual indirect tests, are those that measure responses to a mixture of perceptual and conceptual cues, such as the time to read connected discourse (i.e., text), or responses to degraded cues (e.g., word fragments) presented in the context of intact words that are associatively related to the correct response.
Although apparently straight-forward and certainly useful for descriptive purposes (e.g., the organization of chapters), classifying tests as perceptual or conceptual has a number of drawbacks that often go unacknowledged. One of the main problems with this classification scheme is that it leads to terminological confusions in the description of results, and to circularity in their interpretation. For example, if a conceptual study manipulation such as self generation affects performance on a perceptual indirect test, is that conceptual priming? Or should "conceptual priming" be reserved for performance on conceptual indirect tests? More importantly, how should such an effect be interpreted. A quick survey of the literature reveals little consensus on the matter with some reserchers interpreting such effects as perceptual (e.g., Gardiner, 1988; Roediger & McDermott, 1993), others as conceptual (Hirshman, Snodgrass, Mindes, & Feenan, 1990). Echoing our review of early research on unconscious influences, we question whether it is theoretically sound to define underlying processes in terms of the cues provided at retrieval.
A second problem with the distinction between perceptual and conceptual indirect tests is the notion that the former measure processing "out of context" while the latter measure processing "in context." We take issue with this characterization and assume that there is always a context for processing; the important questions concern the way in which one context (e.g., the isolated-word-paradigm context) differs from others, and the implications of those differences for the operation of memory. A related issue concerns the role of context at encoding. Many researchers tacitly assume a form of exclusivity between perception and conception such that the two trade-off in processing. For example, it has been argued that isolated stimuli engage relatively more perceptual than conceptual processing, and vise versa for items presented in larger groups such as word pairs, connected text, etc. (Jacoby, 1983b; MacLeod, 1989; Roediger & Blaxton, 1987). The notion of a processing trade-off is perhaps best formalized in Roedier et al.'s (1989) claim that data-driven and conceptually-driven processing represent "endpoints on a continuum." However, a single continuum is inconsistent with much of the theorizing surrounding the perceptual/conceptual distinction (e.g., Craik, 1991; Jacoby, in press; Weldon, 1991) and cannot explain a number of experimental findings reviewed below.
Despite these problems, we follow the current zeitgeist and organize our review in terms of the perceptual/conceptual indirect test distinction. Indeed, explicitly using this structure helps us to reveal its inadequacies. We should warn readers that, although we are interested in the unconscious effects of prior conceptual processing, much of our review is dedicated to showing that there are few unquestionable examples of such effects. The major issue is whether indirect tests can be treated as process-pure measures of unconscious influences (Dunn & Kirsner, 1989; Jacoby, 1991; Reingold & Merikle, 1990). If they can (e.g., Rajaram & Roediger, 1993), then examples of conceptual unconscious influences abound. If they cannot (e.g., Toth et al., 1994), then it becomes more difficult to unequivocally identify such examples (see Reingold & Toth, this volume). In our review of the literature, we rely on the results from the process dissociation procedure (Jacoby, 1991; Jacoby, Toth, & Yonelinas, 1993), as well as from other methods (e.g., post-test interviews) to evaluate results gained from indirect tests of memory.
Conceptual influences on "perceptual" indirect tests
Levels of processing and self generation. One basis for the claim that conceptual processing can unconsciously influence subsequent performance comes from findings of LoP and generation effects on perceptual indirect tests. In the LoP paradigm, subjects are presented with single words at encoding and asked to focus their attention on either semantic (meaningful) or non-semantic (physical or lexical) aspects of the word (Craik & Lockhart, 1972). LoP effects in memory performance (i.e., semantic > nonsemantic) are thought to provide some of the most straightforward evidence of conceptual influences because, it is assumed, perceptual encoding is equated in the semantic and non-semantic encoding conditions and only the degree of meaningful elaboration is varied (see Challis and Brodbeck, 1992, for alternative characterizations). The generation paradigm also involves a manipulation of conceptual processing, but that manipulation is usually confounded with differences in perceptual processing. The general encoding procedure is for subjects to read some words in isolation, and generate others on the basis of semantic cues. There is usually some limited amount perceptual information provided in the generate condition, but not enough to completely specify the target. Priming from self-generated items (i.e., generated > new) has been taken to indicate conceptual transfer. Even stronger evidence for conceptual transfer is provided by a "generation effect" (Slamecka & Graf, 1978) in which generated items (which lack complete perceptual specification) facilitate performance significantly more than items that are read (and thus are fully specified perceptually).
Despite initial evidence that LoP and self generation do not effect performance on tests of word identification and stem completion (e.g., Jacoby, 1983b; Jacoby & Dallas, 1981; Graf, Mandler, & Hayden, 1982), such effects have now been produced in a number of independent laboratories and on a wide range of "perceptual indirect tests." For example, significant LoP effects in word-fragment completion were observed by Challis and Brodbeck (1992), who also reviewed previous investigations of this variable. They found that in 33 of 35 cases (experiments or experimental conditions) priming was greater in the semantic, as opposed to nonsemantic, condition. In a more extensive review, Brown and Mitchell (1994) surveyed 166 cases from 38 studies and found that 79% yielded positive LoP effects, over half of which were statistically significant. Interestingly, there was no statistical difference between the percentage of LoP effects found on perceptual versus conceptual indirect tests. Also worth noting is that both the elderly (Chiarello & Hoyer, 1988) and memory-impaired patients with various etiologies (Graf, Squire, & Mandler, 1984) have shown reliable LoP effects on ostensibly perceptual indirect tests.
Similar to results using the LoP paradigm, words generated on the basis of semantic cues and incomplete perceptual specification have also been found to produce significant priming (generated > new) on perceptual indirect tests. Tests showing such effects include auditory and visual word-stem completion (Bassili, Smith, & MacLeod, 1989), word-fragment completion (Blaxton, 1989; Weldon, 1991), anagram solution (Srinivas & Roediger, 1990), picture-fragment completion (Hirshman et al., 1990), and word identification (Masson & MacLeod, 1992). Even more remarkable, significant generation effects (generate > read) have been found on word-fragment completion (Gardiner, 1988a; 1989), anagram solution (Gardiner, Dawson, & Sutton, 1989) and word identification (Toth & Hunt, 1990). The generation effects appear to depend on the complete reinstatement of the physical cues used to constarin generation at encoding. Gardiner (1988) had subjects generate or read intact words in the context of short definitions (e.g., It swings in a clock: -E-D-L-M). In a subsequent indirect fragment completion test, subjects showed significantly greater priming for previously generated, as compared to read, words but only when the test fragment was identical to that presented at encoding. Changing the test fragment by only one letter elimnated the generation effect (Gardiner et al., 1989).
Analogous results in a word identification test were found by Toth and Hunt (1990, Experiment 1) who had subjects generate words in absence of semantic guidance in order to better equate the nature of encoding in the generate and read conditions. Subjects read single words and generated others from isolated word-fragments that were missing one interior letter. This study manipulation was crossed with the type of test cue given in word identification; subjects were given masked presentations either of complete words or of word-fragments that were missing the same letter as that at encoding. When complete words were flashed, generated and read word showed an equivalent amount of priming (.21 vs. .19). In contrast, when a fragment was flashed, previously generated words were identified significantly more often than words that had been read (.33 vs. .20). Interestingly, this effect occurred even though subjects could not reliably discriminate between the type of stimulus presented at test (whole words vs. fragments).
The Toth and Hunt (1990) results, as well as those by Gardiner and colleagues (1988a, 1989; Gardiner et al., 1989), suggest that generation effects on perceptual indirect tests only occur when retrieval cues are perceptually similar to the generation cues used at encoding; the effects do not appear to generalize to non-specific test stimuli such as different fragments, different anagrams, or complete words in word identification. Do these effects reflect the recruitment of prior conceptual processing? Gardiner (1988; see also Roediger & McDermott, 1993) has argued that these effects reflect perceptual priming but this interpretation seems circular in that the nature of processing proposed to underlie the effect is defined by the assumed perceptual nature of the test. However, as demonstrated in the studies review above, it is not at all clear that so-called perceptual indirect tests can be viewed as reflecting only prior perceptual processing. Moreover, it seems unusual to label a generation effect perceptual when the facilitation in performance is actually greater for stimuli having fewer perceptual characteristics at encoding. Our view is that the generation effects obtained by Toth and Hunt (1990) and Gardiner (1988) reflect context specific recruitment of prior encoding operations, both perceptual and conceptual (see also Masson & Freedman, 1990; Masson & MacLeod, 1992). Although the particular conceptual operations involved in this task may be relatively low level (e.g., lexical), they still reflect inferential processes that go beyond the perceptual information given. We thus view these experiments as indicative of a more general principle: Reinstating contextual features of a prior event--including specific perceptual features--can be sufficient to recruit prior conceptual (or inferential) processes which occurred in the same episode.
Results from LoP and generation studies show that, contrary to initial reports, perceptual indirect tests can be affected by prior conceptual processing. The important question is whether these effects truly reflect conceptual unconscious influences or are the result of other factors such as undetected conscious uses of memory. Toth et al. (1994) addressed this issue by manipulating LoP and comparing performance on an indirect stem-completion test with estimates of conscious and unconscious influences derived from the process-dissociation procedure. Replicating previous studies, performance following indirect test instructions showed a small but highly significant LoP effect (semantic = .51, nonsemantic = .45). In contrast, estimates of unconscious influences showed no effect of LoP (.45 vs .44). Importantly, the estimate of unconscious influences following non-semantic study was nearly identical to performance on the indirect test for that condition. This convergence is to be expected if the two paradigms are measuring the same construct and the implicit test is uncontaminated by conscious uses of memory. Evidence for the lack of contamination was provided by estimates of conscious uses of memory which were quite high following semantic study (.27) but near zero following nonsemantic study (.03). Bowers and Schacter (1990), using post-test interviews, also found evidence for conscious contamination: "test aware" subjects showed large LoP effects (.20) but "test unaware" subjects had a small nonsignificant effect (.03). Both studies, then, suggest that previous findings of LoP effects on indirect tests are the by-product of conscious uses of memory.
Toth et al. (1994) also examined whether generation from semantic cues produces subsequent unconscious influences. Subjects read isolated words or generated words from sentences and the first letter of the target word (e.g., The witch gave Snow White a poisoned a____). In the test phase, subjects given indirect test instructions showed a typical pattern of results: Read words were given as completions more often than generated words (.54 vs. .44) but both conditions were well above baseline (.30). It is the difference between baseline and the generation condition that some researchers have claimed shows conceptual priming (e.g., Hirshman et al., 1990; Masson & MacLeod, 1992). Application of the process-dissociation procedure, however, suggested a very different conclusion. Whereas generated words afforded significantly more conscious control than read words (.34 vs. .21), they were at a large disadvantage in producing unconscious transfer (.32 vs. .51). Indeed, comparing the unconscious estimate for generated items to baseline (.32) indicated that generated items produced no unconscious influence. Jacoby et al. (1993) found a similar cross-over interaction of processes using anagrams as the cues for generation. In contrast to indirect test performance which showed significant priming for words derived from previously presented anagrams, estimates of unconscious influences did not differ from baseline. Taken together, results from Toth et al. (1994) and Jacoby et al. (1993) suggest that generation effects on indirect tests may often reflect conscious uses of memory.
In summary, a large number of studies have shown that conceptual processing manipulations such as LoP and self-generation can significantly influence performance on ostensibly perceptual indirect tasks. On closer examination, however, many of these effects appear not to reflect conceptual priming but rather are mediated by more conscious uses of memory. With the possible exception of results using perceptually-specific cues (Gardiner, 1988; Toth & Hunt, 1990), it would appear that unconscious influences obtained with isolated stimuli are mediated completely by prior perceptual processing. Such a conclusion would be consistent with Winnick and Daniel's (1970) claim that priming of unrelated words in tests such as word identification reflects "perceptual sensitization;" and with more recent claims that priming on perceptual implicit tests is due to modification of presemantic representation systems (Moscovitch, 1992; Schacter, 1992; Tulving & Schacter, 1990), or the selective transfer of perceptual or data-driven processing (Roediger, 1990).
There is, however, at least one curious anomaly in these theoretical accounts. If item-specific priming is due only to changes in the nature of perception, then it would seem to follow that such changes could be detected as an enhancement of perceptual sensitivity. But experiments explicitly designed to test that idea using signal detection methods have shown that performance in many of the so-called perceptual indirect tests are driven, not by enhancement of perceptual sensitivity, but rather by changes in response bias (Masson & Freedman, 1989; Ratcliff & McKoon, 1988; Ratcliff, McKoon, & Verwoerd, 1989). Response bias is generally thought to reflect a post-perceptual stage of processing although Ratcliff et al. (1989) argued for the notion of a "perceptual bias" to characterize priming in perceptual indirect tests. Nevertheless, the signal detection studies point to the presence of some form of decision processes in these tests. Thus, while we find little evidence for the influence of prior conceptual processing in the experiments reviewed above, there is data suggesting that the priming effects observed involve processing that goes beyond perception per se.
Cross-form priming. Another set of empirical findings that have been linked to conceptual factors is that of cross-form priming. Cross-form priming refers to enhanced performance on a test stimulus that is perceptually dissimilar from the nominal stimulus originally presented at encoding--as, for example, when auditory presentation of a word increases its subsequent production on a visual word-stem completion test. A number of different types of cross-form priming have been investigated, including changes in typography (e.g., Kolers & Roediger, 1984; Jacoby & Hayman, 1987), the size and orientation of objects (e.g., Biederman & Cooper, 1991, 1992; Jolicoeur, 1985), voice characteristics in auditory priming (e.g., Jackson & Morton, 1984; Schacter & Church, 1992), and stimulus clarity (e.g., Jacoby, Baker, & Brooks, 1989; Snodgrass & Feenan, 1990). Here, however, we focus on studies in which the encoded stimuli have no physical overlap with their respective retrieval cues (i.e., cross-modality and picture/word priming) because these cases bear most directly on the question of conceptual priming (for more general reviews of the cross-form priming literature see Kirsner, Dunn, & Standen, 1989; Roediger & McDermott, in press).
Changing presentation modality from study to test was one of the first manipulations compared on direct and indirect tests of memory. Although early experiments suggested that priming effects on were completely modality-specific (e.g., Ellis, 1982; Jacoby & Dallas, 1981; Kirsner & Smith, 1974) it now seems clear that modality-specificity is the exception, not the rule. Cross-modality priming has been found on word identification (Clarke & Morton, 1983; Kelley, Jacoby, & Hollingshead, 1989; Weldon, 1991), fragment completion (Blaxton, 1989; Hunt & Toth, 1990), auditory and visual stem completion (Bassili et al., 1989; Rajaram & Roediger, 1994), anagram solution (Srivinas & Roediger, 1990), and lexical decision (Kirsner, Milech, & Standen, 1983). Although within-modality priming is generally about twice that observed across modalities, significant cross-modality priming is a common finding on indirect tests. Weldon and Roediger (1987; Roediger et al., 1989; Weldon, 1993) suggested that cross-modality priming may reflect the contribution of conceptual processes to indirect test performance. A related hypothesis is that cross-modality priming reflects the operation of a modality-independent representation or process (Kirsner et al., 1983; Kirsner et al., 1989; see also Weldon, 1991, 1993). Note that both hypotheses would be unnecessary if cross-modality "priming" was due to conscious uses of memory.
In addition to cross-modality effects, picture-to-word priming has also been reported by a number of researchers (e.g., Brown, Neblett, Jones, & Mitchell, 1991; Durso & Johnson, 1979; Kirsner, Milech, & Stumpfel, 1986; Roediger et al., 1992; Weldon & Roediger, 1987). Weldon (1991, 1993) has argued that picture-to-word priming reflects lexical activation due to spontaneous covert naming of the picture at study. However, manipulations designed to influence the probability of covert naming of pictures have little or no effect on word-fragment completion (Weldon & Roediger, 1987; but see Weldon & Jackson-Barrett, 1993) or word identification performance (Kirsner et al., 1986). Also, the covert naming hypothesis does not explain why covert picture naming either fails to occur or fails to produce priming in some situations (Rajaram & Roediger, 1993; Weldon, 1991), but produces complete picture-to-word transfer in others (Alejeno & Carr, 1991; Brown et al., 1991).
Along with enhanced word processing from prior presentation of pictures, word-to-picture priming has also been demonstrated both in the identification of fragmented pictures (Hirshman et al., 1990) and in picture naming (Durso & Johnson, 1979). Using the latter test, Brown et al. (1991) found cross-form (word-to-picture) priming to be as great as within-form (picture-to-picture) priming when the type of prime was manipulated between subjects. In within-subject designs, cross-form priming was reduced but still significant when picture and word primes were presented in separate blocks at study; when primes were presented in a completely mixed fashion, cross-form priming was eliminated (cf. Challis & Brodbeck, 1992). On the surface, Brown et al.'s results would appear to represent an extreme violation of transfer appropriate processing. Mediation from prior lexical activation (Weldon, 1991) may provide some reconciliation but, in addition to the inconsistencies noted above, there is no obvious reason why the lexical contribution from word primes should change as a function of list composition (mixed vs. blocked). These same criticisms apply to explanations based on unconscious conceptual mediation.
Importantly, lexical or conceptual explanations of cross-form priming would only be required if such priming represented true unconscious influences of memory. Recent results question that assessment. Jacoby et al. (1993) had subjects read and hear different lists of words and then used a process-dissociation procedure to estimate conscious and unconscious influences in a visual stem-completion task. In two experiments, prior visual exposure produced significant unconscious influences as estimated by the process-dissociation procedure but prior auditory experience produced little (Experiment 1a) or no (Experiment 1b) effect. The small effect found in Experiment 1a was attributed to subjects pronouncing the stems at test, thus producing an auditory retrieval cue, rather than to automatic cross-modal transfer. In two other experiments, Jacoby et al. (1993) found that anagrams presented visually and solved aloud did not drive estimates of unconscious influences above baseline. These results cast doubt on a general lexical activation hypothesis (Weldon, 1991) and instead suggest that cross-form priming is a by-product of controlled strategies, such as pronouncing the stem or consciously retrieving list items.
Using a fragment-completion tests, Jacoby, Yonelinas, and Jennings (in press) also examined cross-modality priming as assessed by the process-dissociation procedure, indirect test instructions, and a subjective-report ("remember/know") technique (Tulving, 1985; Gardiner, 1988b). Under the assumption that conscious and unconscious influences make independent contributions to performance, estimates of unconscious influences from both the process-dissociation procedure and the remember/know technique showed no cross-modality transfer. In contrast, indirect tests showed significant cross-modality priming. However,, when attention was divided at study, thereby reducing conscious recollection and the possibility of conscious conatmination, the cross-modality effect on the indirect test was totally eliminated. For the other two measures, estimates of conscious influences were significantly reduced by the divided-attention manipulation, but estimates of unconscious influences did not change. These results strongly suggest that cross-modality priming arises from conscious uses of memory. Similar arguments can be made for picture/word priming (see Toth et al., 1994, for further discussion). At the very least, these results recommend divided attention or speeded retrieval (Weldon & Jackson-Barrett, 1993) as a converging operation for establishing the validity of an indirect test as a measure of unconscious influences.
In summary, as was the case for LoP and self-generation, cross-form priming effects appear to be the by-product of conscious uses of memory. What do these findings imply? One specific implication is that there may be no need to postulate modality-independent representations of the form advocated by Kirsner and others (Kirsner et al., 1983, 1989; Rajaram & Roediger, 1993). More generally, because conscious processes enable a person to translate symbolic stimuli into a variety of perceptual forms, contaminated indirect tests may often underestimate the specificity of the effects of prior experience. Cross-form "priming" is a ubiquitous phenomenon in the memory literature and stands as one of the primary sources of support for theories postulating abstract, context-independent mental representations. In contrast to these views, the present analysis suggests that the unconscious influences may be more episodically based than has been appreciated (Jacoby & Brooks, 1984; Logan, 1988; Jacoby & Kelley, 1987, 1990).
Conceptual influences on "conceptual" indirect tests
In all of the studies described above, effects of prior experience were assessed by presenting subjects with isolated stimuli under degraded (data-limited) conditions. The subject's goal in these tasks is to identify or generate members of a particular class (e.g., English words) on the basis of the surface information presented. It is also possible to present subjects with retrieval cues containing no perceptual information that could directly elicit the target response; rather, the test cue consists of semantic or associative information and the subjects' task is to generate a response that is meaningfully related to the cue. One of the earliest modern uses of such a conceptual indirect test is attributable to Storms (1958) who found that presenting words for a subsequent memory test increased the frequency with which these words were produced on an intervening, indirect word-association task. Storms had originally described the phenomenon as a "recency" effect but Segal and Cofer (1960), who replicated Storms' findings, renamed it "priming" to avoid confusion with the recency effect found in list learning experiments. Thus, in what is an interesting historical twist, the term priming, which has been treated as an exclusively perceptual phenomenon (Tulving & Schacter, 1990), was originally used to describe unconscious influences in a conceptual task.
Semantic cues. Perhaps the first systematic review of performance on what are now called conceptual indirect tests was done by Cofer in 1967. Coming from the verbal learning tradition, Cofer was interested in the effects of context on production of responses in a variety of verbal-association tasks. Although this work contains no explicit discussion of unconscious influences, many of the experiments described made use of free-association tests, interpolated between a mnemonic study and test phase, and thus anticipated more recent experiments using conceptual indirect tests. Experiments on "direct priming" showed that previously studied words were often given as responses to associatively related cues (Clifton, 1966; Storms, 1958; Grand & Segal, 1966; Segal & Cofer, 1960). Also, experiments on indirect or "mediated priming," in which presentation of an item or set of items increases subsequent responses with related stimuli, are suggestive of unconscious influences of associative information. For example, Cramer (1964, 1966) presented sets of four words (e.g., dark, black, moon, dream) and had subjects free associate to the last one. Results showed that the first three words--which are known to elicit the target response night--increased the frequency with which "night" was given as a response to "dream" (from .09 to .32 in this example). Mediated priming may also affect performance on direct tests of memory. Words associatively related to target words have been found to produce more false recognitions (Underwood, 1965; see also Juola, Fischler, Wood, Atkinson, 1971; Hermann, McLaughlin, & Nelson, 1975) and are given more often as recall intrusions (Cramer, 1965; Deese, 1959) than comparable unrelated words (Roediger & McDermott, 1994).
Some of the best evidence that semantic cues can automatically elicit previously encountered information comes from work with amnesics (see Moscovitch et al., 1993). Shimamura and Squire (1984) presented memory-impaired Korsakoff patients with a list of single words (e.g., hot) and later presented semantic associates of these words (e.g., stove) with indirect free-association instructions. Despite the amnesics' nearly complete inability to recall list items, their production of target words in the free-association test was equivalent to that of a matched control group. A similar result was found when the groups were presented with highly-related cue-target pairs at study (e.g., table-chair) and later asked to produce the first word that came to mind in response to the cue (table-?). For both groups of subjects, production of the target words was almost three times above baseline. Schacter (1985; Schacter & McGlynn, 1989) has obtained analogous results using idioms such as "sour grapes" and "small potatoes."
In addition to word-association, other conceptual indirect tests include category-exemplar generation (Hamann, 1990; Rappold & Hastroudi, 1991; Srivinas & Roediger, 1990) and answering general knowledge questions (Blaxton, 1989; Hamann, 1990). As with word-association, these tests contain no perceptual information that could directly specify the target response, and thus facilitation as a function of prior presentation is often taken as demonstrative of conceptual priming. However, a potentially important issue is whether these tests are measuring prior meaning-based processing or simply an increased tendency to output the target in response to any cue, conceptual or perceptual. In many of the experiments using conceptual indirect tests, priming is found after simple presentation of the target at study (e.g., Blaxton, 1989; Schacter, 1985) and even following nonsemantic encoding tasks (e.g., Hamann, 1990). Thus, prior presentation can increase the probability that an item will be output independent of any specific meaningful processing at study. Analogous to findings with perceptual indirect tests (e.g., Masson & Freedman, 1989; Ratcliff et al., 1989), these observations raise the question of whether so-called conceptual priming reflects the tranfer of prior meaning-based processing, or simply a bias to respond with previously encountered items irrespective of prior processing. An important goal for future research is to disentangle the effects of prior meaning-based processing versus those arising from simple item registration (cf. Moscovitch, 1992).
As an operational definition of a conceptually driven test, Roediger et al. (1989) proposed that words generated at study should produce superior transfer in comparison to those that were read in "no context" conditions. Presumably, the idea behind this approach is that if you know what kind of test you have, you can infer the nature of the processes which produce priming on that test. On the basis of this logic, it has been argued that category-exemplar generation and tests of general knowledge measure unconscious influences of prior conceptual processing (Blaxton, 1989; Srivinas & Roediger, 1990; see also Smith & Branscomb, 1988). As with perceptual indirect tests, however, the effects found with conceptual cues generally parallel those found on direct tests, raising the question of conscious contamination. In fact, one of the first studies applying the process-dissociation methodology to a conceptual indirect test (category exemplar generation) found evidence that apparent conceptual priming actually reflected conscious uses of memory (Jaciw & McAndrews, 1993).
From our perspective, the problem with Roediger et al.s operational approach is its underlying assumption that indirect tests are process-pure measures of implicit memory. If, in constrast to this assumption, performance on indirect tests is contaminated by conscious uses of memory, then Roediger et al's operational criteria will not validly identify unconscious conceptual processing. Toth et al. (1994) suggested a more stringent criterion for identifying conceptual transfer: Apply Roediger et al.'s (1989) generate/read comparison to estimates of unconscious influences derived from the process dissociation procedure. For this approach, unconscious conceptual influences would be demonstrated if estimates of automatic influences for words generated without perceptual support at study exceeded baseline; or, more impressively, if the generation estimates exceeded those of previously read items. LoP effects on estimates of automatic influences would also suggest unconscious conceptual transfer. Both of these results have been obtain on tests of recognition memory (Jacoby, 1991; Toth, 1992), but not on tests using physically degraded stimuli such as word stems (Jacoby et al., 1993; Toth et al., 1994). One potentially important difference between the two tasks is the goal of retrieval: Stem-completion requires the identification of a perceptual/lexical pattern whereas recognition requires a subject to localize an item in a particular episodic context. An important goal for future research is the application of process dissociation procedures to tests employing only semantic retrieval cues.
Unconscious influences from new and pre-existing associations. Another phenomenon that has been related to the issue of conceptual priming is the unconscious use of associative information, either between items having a pre-existing semantic relationship or between unrelated items that have been not been previously associated. The general procedure for this paradigm is to present a cue-target pair at encoding and then to elicit retrieval using a combination of perceptual and conceptual cues. Generally, retrieval cues consist of a part of the target (e.g., a word stem) presented either in the context of the cue with which it was paired at study (intact condition) or in the context of a different cue, often taken from another studied pair (recombined condition). Unconscious influences of associative information are revealed by the difference between stem-completion performance in the intact and recombined conditions. Because both target words were previously presented, the advantage for intact over recombined pairs can only be explained by memory for associative information.
This paradigm was first used in a series of experiments by Graf and Schacter (1985; Schacter & Graf, 1986a, 1986b) to explore the possibility of "implicit memory for new associations." Initial studies suggested that new associations could produce unconscious influences in both amnesics and controls, but only when the unrelated cue-target pairs had been processed semantically and relationally at encoding. Subsequent research, however, has uncovered an inconsistent pattern of positive and negative results (see Bowers & Schacter, 1993, for a review). On the basis of post-test interviews, Bowers and Schacter (1990) classified subjects as either "test aware" or "test unaware" and found that only the former group showed implicit memory for new associations. Also, studies with amnesics have shown that the acquisition of novel associations is negatively correlated with the severity of amnesia (Schacter & Graf, 1986b; Shimamura & Squire, 1989). Both results suggest that the basic effect may often be due to conscious uses of memory.
In an attempt to determine whether "implicit memory" for new associations is actually implicit, Reingold and Goshen-Gottstein (submitted) used both indirect test instuctions and the process-dissociation procedure. Performance in the indirect test following elaborative encoding revealed the usual advantage for stems presented in intact, as compared with recombined, test pairs (.46 vs. .36). Process dissociation estimates of conscious control also showed an advantage for intact as compared to recombined conditions (.30 vs. 10), confirming that reinstatement of a recently encountered associative context can enhance recollection of individual items (Tulving & Thompson, 1973). However, estimates of unconscious influences showed little difference between performance in intact and recombined conditions (.29 vs. .32, ns), suggsting that many results obtained with this "implicit" paradigm may reflect conscious contamination. Interestingly, using a nonsemantic task in which subjects simply copied the unrelated word pairs (Micco & Masson, 1991), Reingold and Goshen-Gottstein did find evidence that associative information can unconsciously influence performance. This finding suggests that meaningful elaboration, in and of itself, may be less important for acquiring new associations than has been claimed (see Schacter & Graf, 1986a). Perhaps the critical factor is that the items be strongly related or integrated so as to produce at least some degree of unitization (Hayes-Roth, 1977). Important questions for future research include the extent and nature of the relational processing required.
Although unconscious influences from new associations has produced equivocal results, Jacoby (1994) has used the process-dissociation procedure to provide evidence that exposure to pre-existing associations can produce subsequent unconscious influences. At encoding, subjects made relatedness judgments to both related (e.g., knee-bone) and unrelated (e.g., apple-shell; turtle-cider) pairs, under conditions of either full or divided attention. Unrelated pairs were formed by randomly pairing members of related pairs that were rejoined at test. At test, subjects were presented with a cue word followed by the initial letter of the associatively-related target word (e.g., knee-b; apple-c). For the inclusion test condition, subjects were told to complete the target with an associatively-related word from the study list or, if they could not remember an appropriate study item, to compete the stem with the first related word that came to mind. For the exclusion test condition, subjects were instructed to complete stems with related words that were not presented at study (e.g., acceptable responses to knee would be bend, brace, band, etc.). As expected, divided, as compared to full, attention to study pairs significantly reduced conscious recollection for the individual words, but had no effect on estimates of unconscious influences (cf. Jacoby et al., 1993). Most important, estimated unconscious influences were significantly greater when the original associative context was reinstated (i.e., intact condition) in comparison to when the prior context was not reinstated (i.e., recombined condition). In subsequent research, Jacoby (in press) has shown that this automatic associative effect survives a divided attention manipulation at study, and a change in modality from study to test (cf. Schacter & Graf, 1989).
Conceptual unconscious influences in context
In all of the studies reviewed above, the focus has been on the type of cues provided at test. However, as indicated in our historical review, there are reasons to question whether one can infer the cognitive processes involved in a task simply on the basis of the cues provided. This conclusion would appear to be supported by the studies reviewed above in that conceptual encoding manipulations were found to influence "perceptual" tests and perceptual (non-semantic) encoding tasks enhanced performance on "conceptual" tests. While an analysis of retrieval cues is certainly important for understanding memory, classification of processes on the basis of test cues alone misses the broader context in which retrieval is taking place. In this section, we review evidence for the influence of context and set in the production of unconscious influences of memory.
List-wide context and the priming of interpretation. One of the first studies to suggest an important role of the retrieval context in the production of unconscious influences was done by Jacoby (1983a) who found that facilitation in word identification was enhanced when the test list contained a high, as opposed low, proportion of previously studied words (see also Kasserman, Yearwood, & Franks, 1987). Allen and Jacoby (1990) replicated this effect (but see Challis & Roediger, 1993) and provided additional data suggesting that the effect was not due to conscious uses of memory. An interesting aspect of these experiments, one that suggests an effect of set on retrieval, is that performance on new items decreased as the number of old items increased. Experiencing a large number of test items originating from a specific prior episode (e.g., the encoding task) may act to establish an anticipatory set for encountering similar items. Such a set may inhibit or otherwise interfere with performance on non-set (i.e., "new") items. As the number of old items increases, so does the anticipatory set; new items suffer in this context because they do not fit the expected pattern.
A further experiment by Jacoby (1983a) also suggested the operation of cognitive set. In an experiment investigating facilitation effects over a five-day period, Jacoby (1983a) found that identification performance was larger if the test list contained only items studied in a single session. When items from multiple sessions were tested in a mixed list, much less facilitation was found. Although not designed to investigate conceptual priming, these studies provide evidence that unconscious influences are sensitive to the context in which retrieval occurs. Moreover, they suggest that perceptual processing alone is not sufficient to mediate priming and, in this regard, anticipate more recent work suggesting an integration of perceptual and conceptual processes (e.g., Jacoby et al., 1992; Levy & Kirsner, 1989; Masson & Freedman, 1990).
The experiments by Jacoby (1983a) suggests that contextual factors can alter the expression of unconscious influences by affecting which particular episodes, and what aspects of those episodes, are recruited to support current performance. It could be argued, however, that these effects, although context-dependent, are essentially taking place with the perceptual domain. Is there any evidence that prior conceptual operations can be evoked through contextual reinstatement? A suggestive example is provided by Jacoby and Witherspoon (1982) who found evidence for what could be called interpretive priming. In the first phase of their experiment, amnesics and controls heard questions that biased the low-frequency meaning a homophone embedded in the questions (e.g., "Name a musical instrument that employs a reed"). In a later indirect spelling task, both groups showed a large increase in the number of words spelled in accordance with their previously biased meaning, an effect on spelling that was independent of recognition memory for the previously presented words.
Is homophone biasing a perceptual or data-driven effect? Given that the test word was presented in isolation, spelling would seem to fit popular descriptions of data-driven tests (e.g., Bainbridge, Lewandowsky, & Kirsner, 1993; Moscovitch, 1993). In contrast, we believe that the critical aspect of any task is not whether the cue is isolated or presented in the context of other items--whether it is "perceptual" or "conceptual"--but what the subject must do with that cue given the overall task demands. Spelling of a homophone requires that a particular meaning be instantiated. In line with this goal, results showed that it was the episodic interpretation of the initial event that was used to guide the interpretation of the subsequent event (see Masson & Freedman, 1990).
Another study which shows repetition priming for episodic interpretations was done by McAndrews, Glisky, and Schacter (1987) who presented amnesics with sentences that were difficult to interpret without the knowledge of a key word or phrase. For example, the sentence "The haystack was important because the cloth ripped" makes little sense in isolation, but is readily understood when contextualized with the key word "parachute." Amnesics with severe memory deficits were initially shown a set of these puzzle sentences and then given the key word. Facilitation was measured by their ability to produce the key word or phrase when later presented with the sentences alone. Despite showing substantial impairment on tests of free recall and recognition memory, the amnesic's ability to solve the puzzle sentences was nearly three times above baseline solution rates, a level of comprehension that did not show signs of decay after a delay of one week. The puzzle-sentences used by McAndrews et al. (1987) can be viewed as distinctive context cues that helped to retrieve specific (i.e., episodic) interpretations encountered earlier.
The homophone-bias and puzzle-sentence studies show that when an ambiguous stimulus pattern is presented, prior conceptual processing surrounding its initial interpretation may be recruited to influence its subsequent interpretation. Other experiments, using more traditional data-driven tests, have also suggest the importance of episodic interpretations for subsequent priming effects. For example, Masson and Freedman (1990) found that lexical decisions to repeated homogrpahs may produce no priming when a context word precedes them and therefore biases different meanings on thier first and second occurrence. Vriezen, Moscovitch, & Bellos (in press) have also looked at the effects of repeated interpretations by presenting subjects with isolated words and manipulating the type of judgment made on the first and second presentation. They found that making a semantic judgment to a visually presented word resulted in a small but significant priming effect on a subsequent naming or lexical decision task. However, no facilitation was found when the order of tasks was reversed (i.e., from naming or lexical decision to semantic judgments) even though the same perceptual/lexical unit was presented on both occasions (see also Ratcliff, Hockley, & McKoon, 1985). Perhaps more surprising, although performance on a specific word was facilitated when a semantic judgment was repeated (e.g., "Is it bigger than a breadbox?"), no facilitation was found when the second judgment made reference to a different semantic domain (e.g., "Is it manmade?"). Overall, these results pose problems for views that treat data-driven processing as independent from the more general context in which that processing is carried out, and instead suggest that repetition priming effects are specific to the match between prior and current task goals.
Text as context. Taken as a whole, the studies reviewed so far suggest an important role for context in the production of unconscious influences. Additional evidence for contextual factors is provided by research using connected discourse. Oliphant (1983) investigated the relationship between context and repetition priming in a single-word lexical decision task. Repeated words had first been encountered either in isolation (as part of a study list or a previous lexical decision task) or in meaningful passages (preexperimental questionnaires or task instructions). Oliphant found that, unlike words presented in isolation, words presented in text did not facilitate performance on the lexical decision task. A similar pattern of results was reported by MacLeod (1989) using fragment completion: Target words embedded in meaningful passages produced much less priming than did words presented as part of a to-be-learned list. Both the Oliphant and MacLeod results have been interpreted as showing that context reduces the amount of perceptual analysis given to specific word forms, thereby producing a memory trace unable to support subsequent perceptual priming (Jacoby, 1983b; Roediger & Blaxton, 1987). For example, MacLeod (1989; see also Masson & MacLeod, 1992) has stated that "...context plays a critical role in priming: As a word moves from being contextually bound in meaningful discourse to being isolated in a list, its probability of priming increases" (p. 398).
That context plays a critical role in mediating the effects of prior experience is in accord with the present framework. However, MacLeod's conclusion focuses on contextual factors at encoding without considering the importance of contextual factors at retrieval. That unrelated words presented in isolation at study show the highest level of priming would be expected on an indirect retention measure which itself contains little meaningful structure. The present analysis suggests that if the context provided by meaningful discourse were available at retrieval, even "contextually bound" words would show performance facilitation or priming.
Our interpretation receives strong support from a study by Levy and Kirsner (1989) who manipulated the surface characteristics of both meaningful passages and word-lists in order to study word- and text-level transfer on indirect measures of memory. Replicating Oliphant (1983), they found that words embedded in natural text did not facilitate performance on a subsequent measure of word identification. Words presented in isolation, however, did facilitate identification performance, the magnitude of which varied with the similarity of surface characteristics (e.g., modality, type-font) from study to test. More importantly, when the earlier-read passages were re-presented at test, performance (reading time) was significantly facilitated and varied with surface-level similarity. The latter finding rules out the possibility that the failure to find text-to-word transfer on isolated-word tests (e.g., word identification, lexical decision, or completion) was due to context reducing the role of data-driven processing at encoding (cf. Jacoby, 1983b; MacLeod, 1989; Roediger et al., 1989). Contrary to this account, Levy and Kirsner's results show that memory for perceptual characteristics was integrated with other, more conceptual, aspects of the prior reading experience (see also Jacoby et al., 1992; Kolers & Roediger, 1984). Apparently, whether a particular aspect of prior processing will influence subsequent performance depends on the context in which that performance is assessed. Transfer is most comprehensive when the test reinstates the context surrounding the original encoding episode.
Reading tasks offer one of the most flexible paradigms for investigating the nature of contextual reinstatement because, in addition to the repetition of specific words, other forms of contextual information can be manipulated including syntactic, semantic, and thematic structures. Space limitations restrict us from reviewing that literature here (see Levy, 1993), but we believe it confirms a powerful role for contextual (as well as episodic) factors in the production of unconscious influences. As one brief example that provides some symmetry with the studies reviewed above, Levy, Masson, and Zoubek (1991) found that reading particular words can result in little or no transfer to the reading of texts composed of those words, unless higher-order textual characteristics (i.e., syntax, thematics, etc.) were also encoded as part of the original event. This was true even when surface characteristics (i.e., type fonts) were perfectly matched.
Orientation to the past as a context for the present. We have argued that conceptual priming is best revealed when the context surrounding prior processing is reinstated at retrieval. Unfortunately, using indirect tests to investigate context and set effects is difficult, if not impossible, because of the possibility that subjects will become aware of some critical past event. This is a serious limitation for the indirect test approach because there are reasons to believe that unconscious influences may be sensitive to the context set by awareness and intent (Jacoby, Ste-Marie, & Toth, 1993; Wegner, 1993). Direct tests of memory focus a subject's awareness on the past, thus encouraging the intentional reinstatement of an earlier episodic context. Jacoby et al. (1993) have shown that performance on direct tests is often supported by unconscious influences of memory. Thus, if we are correct about the importance of context in producing unconscious conceptual influences, we should find conceptual influences on the unconscious processes that support performance on direct tests such as recognition memory.
Toth (1992) found such evidence using an exclusion test in combination with a response deadline technique. In Phase 1 of his second experiment, subjects made semantic and nonsemantic judgments to different sets of words presented under incidental study instructions. In the second phase, subjects were presented with an auditory word-list which they were told to remember. In the recognition memory test that followed, subjects were told to accept only the words they had been asked to remember (i.e., those presented aurally); both new words and those from Phase 1 were to be rejected. Results showed that the ability to reject words from Phase 1 depended on both retrieval time and prior processing. At the long deadline (1500 ms), where subjects had ample time for recollection, incorrect acceptance of Phase 1 words did not differ as a function of prior processing (semantic=.20; nonsemantic=.21). In contrast, at the short deadline (500 ms), where recollection was curtailed and responses more familiarity-based, false acceptance was significantly higher following a prior semantic (.56), as opposed to nonsemantic (.30), orienting task. This result suggests that, contrary to previous claims (e.g., Jacoby & Dallas, 1981; Mandler, 1980) automatic familiarity is not based only on perceptual factors (i.e., "perceptual fluency") but can also reflect prior conceptual processing (see also Jacoby, 1991). This conclusion was supported in a follow up experiment that used the process-dissociation procedure to estimate the magnitude of conscious recollection and automatic familiarity. As expected, estimated recollection was significantly affected by both LoP and response time at retrieval; that is, estimates were higher following semantic processing and at the long response deadline. In contrast, estimates of familiarity did not change across the response deadline, but were significantly higher following semantic, in comparison to nonsemantic, study at both points in retrieval.
As a final example of how orientation to the past can set the context for unconscious influences, consider two experiments by Ste-Marie and Jacoby (1994). They were interested in factors which result in spontaneous recollection of a prior event, but wanted to measure such factors indirectly. In order to do this, they used a variant of the flanker paradigm (Eriksen & Eriksen, 1974). Subjects studied a list of visually presented words and were then given a speeded, visual test of recognition memory. Words presented for recognition memory judgments (henceforth "targets") were flanked above and below with another word that could have been either previously presented in the experiment ("old") or not previously presented ("new"), but subjects were told to attend and respond only to the center (target) word. Results showed that when the flanking and target words were both of one type (i.e., old or new) subjects were facilitated in their reaction times to make correct recognition decisions. This result indicates both that selective attention to the middle word was not perfect and that the nature of the flanking word (i.e., its status as old or new in the experiment) may influence responses to the target.
What factors might be important for mediating this flanker (spontaneous recognition) effect? If one thought that automatic influences were completely data-driven or based on perceptual fluency, then decreasing the perceptual similarity between a word's prior presentation and its use as a flanker--for example, having the word be presented auditorily at study--should reduce its influence. However, in experiments designed to test this hypothesis, Ste-Marie and Jacoby (1994) found something much more interesting and, we believe, important. They found that whether a flanking word had an influence on responses to the target depended, not on the perceptual similarity of the flanker from study to test, but rather on the relationship between the target and the flanker. That is, flanking words previously seen at study only influenced responding when the target itself had previously been experienced visually. When the target had previously been presented auditorily, only words that were earlier heard influenced reaction times to targets; study words that had been presented visually had no influence on previously heard targets. Apparently, attention (intention?) to the target word created a set which allowed events of similar kind (i.e., previously heard or previously read words) to gain access to the response system and influence responding. Words from a different "category," although potentially having all the characteristics for perceptual priming, had no influence. Similar results have emerged in studies of attentional orientation (Folk, Remington, & Johnston, 1992). These findings suggest that automatic influences, rather than being absolute or the by-product of cognitively-impenetrable perceptual modules, are senstive both to the context of processing and to the goals set by intentions (Jacoby et al., 1993).
E. Summary of contemporary research.
One of our main goals in this chapter was to critically review the evidence for unconscious uses of prior conceptual processing in order to provide guidelines for future research and theory on the topic. The results of our review were both positive and negative. On the positive side, we find clear evidence that prior conceptual processing can unconsciously influence subsequent behavior. Conceptual priming appears to occur whenever there is a high overlap between the processing context at study and test. Such contextual reinstatement can be either in the form of associative or semantic cues (e.g., Jacoby, 1994, in press) or, possibly, from a high degree of perceptual overlap (Gardiner, 1988; Toth & Hunt, 1990). Perhaps more important, unconscious conceptual influences appear very sensitive to the match between the processing goals operating at encoding and retrieval. Congruent task goals were found to recruit prior conceptual operations in both isolated word paradigms (e.g., the spelling of homographs; lexical decision) and in more contextually rich environments (e.g., reading text; solving puzzle sentences). Also, an intentional orientation to the past may allow prior conceptual processing to unconscious influence performance, possibly through reinstatement of the earlier context, a recreation of the prior set for processing, or perhaps both.
On the negative side of our review, the majority of claims for the existence of conceptual priming are based on research using indirect test paradigms which, although suggestive of unconscious influences, do not rule out the possibility of conscious contamination. Research using the process-dissociation procedure, post-test interviews, and inconsistencies in the indirect test literature, all suggest that many apparent examples of conceptual priming are due to conscious uses of memory. Although some researchers might question the validity of estimates derived from the process-dissociation procedure, we want to emphasize that the results from this procedure are very consistent with those obtained using indirect tests. For example, estimates derived from the procedure show that manipulations of LoP, attention, retrieval time, presentation duration, and aging can have large influences on consciously-controlled uses of memory but leave automatic, unconscious influences unchanged (for reviews see Jacoby et al., 1992; Jacoby et al., in press; Jacoby & Begg, submitted; Toth, Reingold & Jacoby, in press). Moreover, when uncontaminated by conscious uses of memory, performance on indirect tests closely matches estimates of unconscious influences gained from the procedure (Jacoby et al., 1993; Reingold & Goshen-Gottstein, submitted; Toth et al., 1994).
Of course, findings obtained with the process-dissociation procedure do not rule out the possibility that conceptual priming could occur given the cues provided in many of the indirect test paradigms discussed above. Indeed, to the extent that these tests reinstate the prior encoding context, we believe that many of these tests have the appropriate characteristics for eliciting conceptual transfer. But this is part of the problem with indirect test paradigms: The very conditions that are conducive to unconscious use of prior conceptual processing--the reinstatement of context and set--are the exact conditions that often result in spontaneous recollection of the study episode. Conceptual priming may often occur in a variety of tasks and situations but is overlooked because of conscious contamination. For this reason, we believe that further specification of unconscious uses of prior meaning-based processing will require researchers to adopt methods other that those offered by the indirect test approach (see Reingold & Toth, this volume). Obtaining uncontaminated results from indirect tests is not impossible, but it requires that one present items so "out of context" that little direct contact is made with prior processing episodes; hence, little can be learned about conceptual unconscious influences.
As discussed by Jacoby (in press), there are always two potential effects of reinstating context, one automatic, the other controlled. The process-dissociation procedure allows one to reinstate context and still measure unconscious influences despite the potential increase in conscious recollection. Even if one is reluctant to adopt the process dissociation procedure, the data reviewed here support our claim that conceptual priming is most likely to be found in environments with high contextual overlap and thus demand methodologies that go beyond indirect tests given "out of context." One conservative option, shown in the experiment by Toth (1992), is to use an opposition ("exclusion") condition which is often sensitive enough to demonstrate the automatic effects or prior conceptual processing.
F. Conclusions: Getting "set" for future research on unconscious influences
Imagine an experiment in which an entire past event--including the perceptual stimuli, the meaning of the stimuli, the context in which they were encountered, the distribution of attention, and the subject's processing goals or set--was repeated. We imagine that in this situation all measured aspects of performance, including perceptual and conceptual processing, would be facilitated. Of course, entire events cannot be repeated, only aspects of prior events. In this sense, tasks can be viewed as filters that allow some aspects of prior processing to be used in current performance, but not others. Current views of unconscious influences of memory are biased toward perception because perceptual factors are relatively easy to manipulate and measure; that is, we have ready access to available filters. However, other factors--such as set (executive processes) and context--may be equally, if not more, important to understanding the nature of unconscious influences, but they have not been studied by contemporary researchers because of their propensity to evoke awareness of the past.
A central theme in this chapter has been that it is the neglect of context, specifically the context for retrieval, that has limited the scope of conceptual priming observed in the laboratory. This same neglect has limited theoretical approaches to memory because of the tacit assumption that perception and conception can be treated as separate, independent domains of inquiry. Presumably, conceptual processes can be ignored because they arise in a separate system (Tulving & Schacter, 1990; Squire, 1992) or involve a qualitatively different set of processes (Roediger, 1990) than those mediating memory for prior perceptual operations. In contrast to these views, we believe that understanding effects of the past requires that one take into account how perception and conception, either separately or together, are integrated with contextually-defined goals. As stated by Bargh (1990): "Automatic, preconscious influences on behavioral decisions certainly exist... but the decisions themselves are made intentionally and in the service of the current goal" (p.98).
The majority of laboratory demonstrations of unconsious influences of memory has been interpreted as showing almost complete dependence on prior perceptual or data-driven processing; few researchers have worried about whether these demonstrations would be invariant across changes in context or the subjects processing goals. The idea that data-driven processing can be used to account for unconscious influences is similar to the claim that automaticity is stimulus-driven. Indeed, the data-driven/conceptually-driven distinction is very similar to the contrast between stimulus-driven and intention-driven processing that has been popular in theorizing about attention (e.g., Posner & Snyder, 1975). Is performance ever controlled entirely by external stimuli (i.e., data) or entirely by conscious intent? After extensive learning, automatic processes are said to become as encapsulated and uncontrolled as reflexes. However, even reflexes can be modified by attention (Anthony, 1985). Neumann (1984) argued that automatic processes (and, we would add, unconscious influences) are not a characteristic of stimulus-driven processing, but rather are an emergent property of the exercise of specific skills in specific environments. That is, automatic or unconscious uses of the past arise from the integration of stimulus parameters and memory for skills in the context of consciously-controlled goals and intentions. Priming effects from a prior encounter with a word, for example, may arise only in the context of task goals such as completing stems or judging some property of a word or its referent; in other contexts, with other task demands, prior encounters with a word may produce different automatic effects (e.g., Vriezen et al., 1994)--or no effects at all (e.g., Ste-Marie & Jacoby, 1994).
The notion of transfer appropriate processing provides a very appealing and intuitive framework for thinking about memory. Memory is the repetition of a set of operations that occurred together in the past. A more exact theory of memory, however, will require that this "set of operations" is better specified. As a first step toward specification, researchers have focused on the broad distinction between perception and conception. However, while this distinction provides some help in describing the overlap in processing from study to test, it neglects the goals of that processing and the context in which it occurs. When these factors are taken into account, the unit of analysis for a theory of memory becomes the stimulus-goal configuration, rather than isolated perceptual or conceptual processes. Entire events cannot be repeated but similar contexts and sets (goal orientations) often recur, and it in these instances that memory for prior conceptualizations can influence current thought and behavior. Often, such prior conceptualizations are apprehended consciously and thus we can intentionally choose to exploit, or to avoid, the prior influence. More often, we believe, our views of the world are colored covertly and we simply perceive the meaning of the event.
Preparation of this chapter was supported by a fellowship awarded to Jeffrey P. Toth by the Rotman Research Institute of Baycrest Centre and the Clarke Institute of Psychiatry, Toronto, Canada; and by a Natural Science and Engineering Research Council (NSERC) of Canada operating grant to Eyal M. Reingold. We thank R. Hunt, L. Jacoby, B. Levine, M. Masson, N. Meiran, H. Roediger, and M. Wheeler for their comments on an earlier draft of this chapter.
Alejano, A.R. & Carr, T.H. (1991). The development of abstract processing operations in reading: Repetition benfits as a diagnostic tool. Paper presented at the Midwestern Psychological Association, Chicago, IL.
Allen, S.W. & Jacoby, L.L. (1990). Reinstating study context produces unconscious influences of memory. Memory and Cognition, 18, 270-278.
Anthony, B.J. (1985). In the blink of an eye: Implication of reflex modification for information processing. In P.K. Ackles, J.R. Jennings & M.G.H. Coles (Eds.), Advances in psychophysiology (Vol.1, pp.167-218). Greenwich, CT: JAI Press.
Bainbridge, J.V., Lewandowsky, S., & Kirsner, K. (1993). Context effects in repetition priming are sense specific. Memory & Cognition, 21, 619-626.
Bargh, J.A. (1990). Auto-motives: Preconscious determinants of social interaction. In E.T. Higgins & R.M. Sorrentino (eds.), Handbook of motivation and cognition, Vol. 2 (pp.93-130). New York: Guilford Press.
Bransford, J.D., McCarrell, N.S., Franks, J.J., & Nitsch, K.E. (1977). Toward unexplaining memory. In R. Shaw & J. Bransford (eds.), Perceiving, acting, and knowing (pp. 431-466). Hillsdale, NJ: Erlbaum.
Bruner, J.S. & Goodman, C.C. (1947). Value and need as organizing factors in perception. Journal of Abnormal Social Psychology, 42, 33-44.
Bassili, J.N., Smith, M.C., & MacLeod, C.M. (1989). Auditory and visual word-stem completion: Separating data-driven and conceptually-driven processes. The Quarterly Journal of Experimental Psychology, 41A(3), 439-453.
Biederman, I. & Cooper, E.E. (1991). Evidence for complete translational and reflectional invariance in visual object priming. Perception, 20, 585-593.
Biederman, I. & Cooper, E.E. (1992). Size invariance in viual object priming. Journal of Experimental Psychology: Human Perception and Performance, 18, 121-133.
Blaxton, T.A. (1989). Investigating dissociations among memory measures: Support for a transfer appropriate processing framework. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 657-668.
Bowers, J.S., & Schacter, D.L. (1990). Implicit memory and test awareness. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 404-416.
Bowers, J.S., & Schacter, D.L. (1993). Priming of new information in amnesic patients: Issues and data. In Implicit memory: New directions in cognition, development, and neuropsychology, (ed. P. Graf and M. Masson), pp.303-326. Erlbaum.
Brown, A.S. & Mitchell, D.B. (1994). Levels of processing in implicit memory: A reevaluation. Memory & Cognition, 22, 533-541.
Brown, A.S., Neblett, D.R., Jones, T.C., & Mitchell, D.B. (1991). Transfer of processing in repetition priming: Some inappropriate findings. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 514-525.
Challis, B.H. & Brodbeck, D.R. (1992). Level of processing affects priming in word-fragment completion. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 595-607.
Challis, B.H. & Roediger, H.L. (1993). The effect of proportion overlap and repeated testing on primed word fragment completion. Canadian Journal of Psychology, 47, 113-123.
Chiarello, C. & Hoyer, W.J. (1988). Adult age differences in implicit and explicit memory: Time course and encoding effects. Psychology and Aging, 3, 358-366.
Clark, R. & Morton, J. (1983). Cross-modality facilitation in tachistoscopic word recognition. Quarterly Journal of Experimental Psychology, 35A, 79-96.
Clifton, C. (1966). Some determinants of the effectiveness of priming word associates. Journal of Verbal Learning and Verbal Behavior, 5, 167-171.
Cofer, C.N. (1967). Conditions for the use of verbal associations. Psychological Bulletin, 68, 1-12.
Cohen, N.J. & Squire. L.R. (1980). Preserved learning and retention of pattern analyzing skill in amnesia: Dissociations of knowing how and knowing that. Science, 210, 544-554.
Corteen, R.S. & Dunn, D. (1974). Shock-associated words in a nonattended message: A test for momentary awareness. Journal of Experimental Psychology, 102, 1143-1144.
Corteen, R.S. & Wood, B. (1972). Autonomic responses to shock-associated words in an unattended channel. Journal of Experimental Psychology, 94, 308-313.
Craik, F.I.M (1991). On the specificity of procedural memory. In W. Kessen, A. Ortony, & F.I.M. Craik (Eds.), Memories, thoughts, and emotions (pp.183-197). Hillsdale, NJ: Erlbaum
Craik, F.I.M. & Lockhart, R.S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671-684.
Cramer, P. (1964). Successful mediated priming via associative bonds. Psychological Reports, 15, 235-238.
Cramer, P. (1965). Recovery of a discrete memory. Journal of Personality and Social Psychology, 1, 326-332.
Cramer, P. (1966). Mediated priming of associative responses: The effect of time lapse and interpolated activity. Journal of Verbal Learning and Verbal Behavior, 5, 163-166.
Deese, J. (1959). On the prediction of occurrence of particular intrusions in immediate recall. Journal of Experimental Psychology, 58, 17-22.
Durso, F.T. & Johnson, M.K. (1979). Facilitation in naming and categorizing repeated words and pictures. Journal of Experimental Psychology: Human Learning and Memory, 5, 449-459.
Dunn, J.C. & Kirsner, K. (1989). Implicit memory: Task or process. In S. Lewandosky, J.C. Dunn, & K. Kirsner (Eds.), Implicit memory: Theoretical issues (pp.17-31). Hillsdale, NJ: Erlbaum.
Ellis, A. (1982). Modality-specific repetition priming of auditory word recognition. Current Psychological Research, 2, 123-128.
Erdelyi, M.H. (1974). A new look at the new look: Perceptual defense and vigilance. Psychological Review, 81, 1-25.
Erdelyi, M.H. (1985). Psychoanalysis: Freud's cognitive psychology. New York: W. H. Freeman & Co.
Erdelyi, M.H. (1992) Psychodynamics and the unconscious. American Psychologist, 47, 784-787
Eriksen B.A. & Eriksen, C.W. (1974). Effects of noise letters upon the identification of a target letter in a nonsearch task. Perception and Psychophysics, 16, 143-149.
Folk, C.L., Remington, R.W., & Johnston, J.C. (1992). Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception and Performanmce, 18, 1030-1044.
Gardiner, H., Boller, F., Moreines, J., & Butters, N. (1973). Retrieving information from Korsakoff patients: Effects of categorical cues and reference to the task. Cortex, 9, 165-175.
Gardiner, J.M. (1988a). Generation and priming effects in word-fragment completion. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 495-501.
Gardiner, J.M. (1988b). Functional aspects of recollective experience. Memory and Cognition, 16, 309-313.
Gardiner, J.M. (1989). A generation effect in memory without awareness. British Journal of Psychology, 80, 163-168.
Gardiner, J.M., Dawson, A.J., & Sutton, E.A. (1989). Specificity and generality of enhanced priming effects for self-generated study items. American Journal of Psychology, 102, 295-305.
Gibson, J.J. (1941). A critical review of the concept of set in contemporary experimental psychology. Psychological Bulletin, 38, 781-817.
Graf, P. & Mandler, G. (1984). Activation makes words more accessible, but not necessarily more retrieval. Journal of Verbal Learning and Verbal Behavior, 23, 553-568.
Graf, P. & Mandler, G., & Hayden, P.E. (1982). Simulating amnesic symptoms in normal subjects. Science, 218, 1243-1244.
Graf, P. & Schacter, D.L. (1985). Implicit and explicit memory for new associations in normal and amnesic subjects. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 501-518.
Graf, P., Squire, L.R., & Mandler, G. (1984). The information that amnesic patients do not forget. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 164-178.
Grand, S. & Segal, S.J. (1966). Recovery in the absence of recall: An investigation of color-word interference. Journal of Experimental Psychology, 72, 138-144.
Greenwald, A.G. (1993). New Look 3: Unconscious cognition reclaimed. American Psychologist, 47, 766-779.
Hayes-Roth, B. (1977). Evolution of cognitive structures and processes. Psychological Review, 84, 260-278.
Hamann, S.B. (1990). Level-of-processing effects in conceptually driven implicit tasks. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 970-977.
Hermann, D.J., McLaughlin, J.P., & Nelson, B.C. (1975). Visual and semantic factors in recognition from long-term memory. Memory & Cognition, 3, 381-384.
Hirshman, E., Snodgrass, J.G., Mindes, J., & Feenan, K. (1990). Conceptual priming in fragment competion. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 634-647.
Hunt, R.R. & Toth, J.P. (1990). Perceptual identification, fragment completion, and free recall: Concepts and data. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 282-290.
Isen, A.M. & Diamond, G.A. (1989). Affect and automaticity. In J.S. Uleman & J.A. Bargh (Eds.), Unintended thought. New York: Guilford Press.
Jaciw, M. & McAndrews, M.P. (1993, November). Elaborative processing effects on conceptual priming: A process dissociation analysis. Poter presented at the 34th Annual Meeting of The Psychonomic Society. Washington, DC.
Jackson, A. & Morton, J. (1984). Facilitation of auditory word recognition. Memory & Cognition, 12, 568-574.
Jacoby, L.L. (1983a). Perceptual enhancement: Persistent effects of an experience. Journal of Experimental Psychology: Learning, Memory, and Cognition, 9, 21-38.
Jacoby, L.L. (1983b). Remembering the data: Analyzing interactive processes in reading. Journal of Verbal Learning and Verbal Behavior, 22, 485-508.
Jacoby, L.L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30, 513-541.
Jacoby, L.L. (1994). Measuring recollection: Strategic versus automatic influences of associative context. In Attention and performance XV, (ed. C. Umilta & M. Moscovitch). Cambridge, MA: Bradford.
Jacoby, L.L. (in press). Dissociating automatic and controlled effects of study/test compatibility. Journal of Memory and Language.
Jacoby, L.L., Baker, J., & Brooks, L.R. (1989). Episodic effects on picture identification: Implications for theories of concept learning and theories of memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 275-281.
Jacoby, L.L. & Brooks, L.R. (1984). Nonanalytic cognition: Memory, perception and concept learning. In G.H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol.18, pp.1-47). NY: Academic Press.
Jacoby, L.L. & Dallas, M. (1981). On the relationship between autobiographical memory and perceptual learning. Journal of Experimental Psychology: General, 3, 306-340.
Jacoby, L.L. & Hayman, G.A. (1987). Specific visual transfer in word identification. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 456-463.
Jacoby L.L. & Kelley, C.M. (1987). Unconscious influences of memory for a prior event. Personality and Social Psychology Bulletin, 13, 314-336.
Jacoby L.L. & Kelley, C.M. (1990). An episodic view of motivation: Unconscious influences of memory. In E.T. Higgins & R.M. Sorrentino (eds.), Handbook of motivation and cognition, Vol. 2 (pp.93-130). New York: Guilford Press.
Jacoby, L.L., Levy, B.A., & Steinbach, K. (1992). Episodic transfer and automaticity: The integration of data-driven and conceptually-driven processing in rereading. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 15-24.
Jacoby, L.L., Ste-Marie, & Toth, J.P. (1993). Redefining automaticity: Unconscious influences, awareness and control. In Attention: Selection, awareness and control. A tribute to Donald Broadbent, (ed. A.D. Baddeley & L. Weiskrantz), pp.261-282. Oxford, England: Oxford University Press.
Jacoby, L.L., Toth, J.P., Yonelinas, A.P., & Debner, J.A. (1992). Lectures for a layperson: Methods for revealing unconscious processes. In R.F. Bornstein & T.S. Pittman (Eds.) Perception without awareness, pp.81-120. New York: Guilford.
Jacoby, L.L., Toth, J.P., & Yonelinas, A.P. (1993). Separating conscious and unconscious influences of memory: Measuring recollection. Journal of Experimental Psychology: General, 122, 139-154.
Jacoby, L.L. & Witherspoon, D. (1982). Remembering without awareness. Canadian Journal of Psychology, 36, 300-324.
Jacoby, L.L., Yonelinas, A.P., & Jennings, J.M. (in press). The relationship between conscious and unconscious (automatic) influences of memory: Toward showing independence. In J. Cohen & J,W. Schooler (Eds.), Scientific approaches to the question of consciousness, Hiilsdale, NJ: Erlbaum.
James (1890). Principles of psychology. New York: Holt.
Jolicoeur, P. (1985). The time to name disoriented natural objects. Memory & Cognition, 13, 289-303.
Juola, J.F., Fischler, I., Wood, C.T., & Atkinson, R.C. (1971). Recognition time for information stored in long-term memory. Perception and Psychophysics, 10, 8-14.
Kasserman, J.E., Yearwood, A.A., & Franks, J.J. (1987). Contextual priming effects in perceptual identification. Bulletin of the Psychonomic Society, 25, 233-235.
Kelley, C. M., Jacoby, L. L., & Holingshead, A. (1989). Direct versus indirect tests of memory for source: Judgments of modality. Jounral of Experimental Psychology: Learning, Memory, and Cognition, 15, 1101-1108.
Kihlstrom, J,F. (1987). The cognitive unconscious. Science, 237, 1445-1452.
Kirsner, K., Dunn, J.C., & Standen, P. (1989). Domain specific resources in word recognition. In Implicit memory: Theoretical issues, (ed. S. Lewandowsky, J.C. Dunn, & K. Kirsner), pp.99-122). Hillsdale, NJ: Erlbaum.
Kirsner, K., Milech, D., & Standen, P. (1983). Common and modality-specific processes in the mental lexicon. Memory & Cognition, 11, 621-630.
Kirsner, K., Milech, D., & Stumpfel, V. (1986). Word and picture identification: Is representational parsimony possible? Memory & Cognition, 14, 398-408.
Kirsner, K. & Smith, M.C. (1974). Modality effects in word identification. Memory & Cognition, 2, 637-640.
Kolers, P.A. & Roediger, H.L. (1984). Procedures of mind. Journal of Verbal Learning and Verbal Behavior, 23, 425-449.
Levy, B.A. & Kirsner, K. (1989). Reprocessing text: Indirect measures of word and message level processes. Jounral of Experimental Psychology: Learning, Memory, and Cognition, 15, 407-417.
Levy, B.A., Masson, M.E.J., & Zoubek, M.A. (1991). Rereading text: Words and their context. Canadian Journal of Psychology, 45, 492-506.
Lindsay, P.H. & Norman, D.A. (1977). Human Information processing. New York: Academic Press.
Logan, G.D. (1988). Toward an instance theory of automatization. Psychological Review, 95, 492-527.
MacLeod, C.M. (1989). Word context during intial exposure influences degree of priming in word fragment completion. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 398-406.
Mandler, G. (1980). Recognizing: The judgment of previous occurrence. Psychological Review, 87, 252-271.
Masson, M.E.J. & Freedman, L. (1990). Fluent identification of repeated words. Jounral of Experimental Psychology: Learning, Memory, and Cognition, 16, 355-373.
Masson, M.E.J. & MacLeod, C.M. (1992). Reenacting the route to interpretation: Enhanced perceptual identification without prior perception. Journal of Experimental Psychology: General, 121, 145-176.
McAndrews, M.P., Glisky, E.L., & Schacter, D.L. (1987). When priming persists: Long-lasting implicit memory for a single episode in amnesic patients. Neuropsychologia, 25, 497-506.
Micco, A. & Masson, M.E.J. (1991). Implicit memory for new associations: An interactive process approach. Jounral of Experimental Psychology: Learning, Memory, and Cognition, 17, 1105-1123.
Morris, C.D., Bransford, J.D., & Franks, J.J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16, 519-533.
Moscovitch, M. (1992). Memory and working-with-memory: A component process model based on modules and central systems. Journal of Cognitive Neuroscience, 4, 257-267.
Moscovitch, M., Vriezen, E.R., & Gottstein, J. (1993). Implicit tests of memory in patients with focal lesions or degenerative brain disorders. In H. Spinnler & F. Boller (Eds.), Handbook of neuropsychology (Vol. 8). Amsterdam: Elsevier.
Neumann, O. (1984). Automatic processing: A review of recent findings and a plea for an old theory. In W. Prinz & A.F. Sanders (Eds.), Cognition and motor processes (pp.255-293). Berlin: Springer-Verlag.
Niedenthal, P.M. (1992). Affect and social perception: On the psychological validity of rose-colored glasses. In R.F. Bornstein & T.S. Pittman (Eds.) Perception without awareness (pp.211-235). New York: Guilford.
Norman, D.A. (1981). Categorization of action slips. Psychological Review, 88, 1-15.
Norman, D.A. & Shallice, T. (1986). Attention to action: Willed and automatic control of behavior. In R.J. Davidson, G.E. Schwartz, & D. Shapiro (Eds.), Consciousness and self-regulation: Advances in research (vol. 4). New York: Plenum Press.
Oliphant, G.W. (1983). Repetition and recency effects in word recognition. Australian Journal of Psychology, 35, 393-403.
Posner, M.I. & Snyder, C.R.R. (1975). Attention and cognitive control. In R.L. Solso (Ed.), Information processing in cognition: The Loyola Symposium (pp.55-85). Hillsdale, NJ: Erlbaum.
Rajaram, S. & Roediger, H.L. (1993). Direct comparison of four implicit memory tests. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 765-776.
Rappold, V.A., & Hashtroudi, S. (1991). Does organization improve priming? Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 103-114.
Ratcliff, R. Hockley, W.E., & McKoon, G. (1985). Components of activation: Repetition and priming effects in lexical decision and recognition. Journal of Experimental Psychology: General, 114, 435-450.
Reason, J.T. (1984). Lapses of attention in everyday life. In R. Parasuraman & D.R. Davies (Eds.), Varieties of attention (pp. 515-549). New York: Academic Press.
Reason, J.T. (1990). Human error. Cambridge: Cambridge University Press.
Rees, H.J. & Israel, H.E. (1935). An investigation of the establishment and operation of mental sets. Psychological Monographs, 46, No. 210, 1-26.
Reingold, E.M & Goshen-Gottstein, Y. (submitted). Separating consciously controlled and automatic retrieval processes in memory for new associations. Journal of Experimental Psychology: Learning, Memory, and Cognition.
Reingold, E.M. & Merikle, P.M. (1990). On the inter-relatedness of theory and measurement in the study of unconscious processes. Mind and Language, 5, 9-28.
Roediger, H.L. (1990). Implicit memory: retention with remembering. American Psychologist, 45, 1043-1056.
Roediger, H.L. & Blaxton, T.A. (1987). Effects of varying modality, surface features, and retention interval on priming in word fragment completion. Memory & Cognition, 15, 379-388.
Roediger, H.L. & McDermott, K.B. (1993). Implicit memory in normal human subjects. In H. Spinnler & F. Boller (Eds.), Handbook of neuropsychology, Vol. 8, (pp.63-131). Amsterdam: Elsevier.
Roediger, H.L., Weldon, M.S., & Challis, B.H. (1989). Explaining dissociations between implicit and explicit measures of retention: A processing account. In H.L. Roediger & F.I.M. Craik (Eds.), Varieties of memory and consciousness: Essays in honour of Endel Tulving (pp. 3-41). Hillsdale, NJ: Erlbaum.
Roediger, H.L., Weldon, M.S., Stadler, M.A., & Riegler, G.H (1992). Direct comparison of word stems and word fragments in implicit and explicit retention tests. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 1251-1269.
Schacter, D.L. (1985). Priming of old and new knowledge in amnesic patients and normal subjects. Annals of the New York Academy of Sciences, 608, 543-571.
Schacter, D.L. & Church, B. (1992). Auditory priming: Implicit and explicit memory for words and voices. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 915-930.
Schacter, D.L., & Graf, P. (1986a). Effects of elaborative processing on implicit and explicit memory for new associations. Journal of Experimental Psychology: Learning, Memory, & Cognition, 12, 432-444.
Schacter, D.L., & Graf, P. (1986b). Preserved learning in amnesic patients: Perspectives on research from direct priming. Journal of Clinical Experimental Neuropsychology, 8, 727-743.
Schacter, D.L., & Graf, P. (1989). Modality Specificity of implicit memory for new associations. Journal of Experimental Psychology: Learning, Memory, & Cognition, 15, 3-12.
Schacter, D.L. & McGlynn, S.M. (1989). Implicit memory: Effects of elaboration depend on unitization. Americal Journal of Psychology, 102, 151-181.
Segal, S.J. & Cofer, C.N. (1960). The effects of recency and recall on free association. American Psychologist, 15, 451 (Abstract).
Shallice, T. (1982). Specific impairments in planning. Philosophical Transactions of the Royal Society of London B, 298, 199-209.
Shallice, T. (1988). From neurpsychology to mental structure. Cambridge: Cambridge University Press.
Shiffrin, R.M. & Schneider, W. (1977). Controlled and automatic information processing II: Perceptual learning, automatic attending and a general theory. Psychological Review, 84, 127-190.
Shimamura, A.P. & Squire, L.R. (1984). Paired-associate learning and priming in amnesia: A neuropsychological approach. Journal of Experimental Psychology: General, 113, 556-570.
Shimamura, A.P. & Squire, L.R. (1989). Impaired priming of new associations in amnesia. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 721-728.
Siipola, E. (1935). A group-study of some effects of preparatory set. Psychological Monographs, 46, No. 210, 27-38.
Slamecka, N.J. & Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental Psychology: Human Learning and Memory, 4, 592-604.
Snodgrass, J.G. & Feenan, K. (1990). Priming effects in picture fragment completion: Support for the perceptual closure hypothesis. Journal of Experimental Psychology: General, 119, 276-296.
Srinivas, K. & Roediger, H.L. (1990). Classifying implicit memory tests: Category association and anagram solution. Journal of Memory and Language, 29, 389-412.
Ste-Marie, D.M. & Jacoby, L.L. (1993). Spontaneous versus directed recognition: The relativity of automaticity. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 777-788.
Storms, L.H. (1958). Apparent backward association: A situational effect. Journal of Experimental Psychology, 55, 390-395.
Stuss, D.T. (1991). Self, awareness, and the frontal lobes: A neuropsychological persepctive. In J. Strauss & G.R. Goethals (Eds.), The self: Interdisciplinary perspectives (pp255-278). Ney York: Springer-Verlag.
Squire, L.R. (1992). Memory and the hippocampus: Synthesis of findings with rats, monkeys, and humans. Psychological Review, 99, 195-231.
Toth, J.P. (1992). Familiarity is affected by prior conceptual processing: Differential effects of elaborative study and response-time at test on separable processes in recognition memory. Unpublished data and manuscript.
Toth, J.P., & Hunt, R.R. (1990). Effect of generation on a word-identification task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 993-1003.
Toth, J.P., Lindsay, D.S. & Jacoby, L.L. (1992). Awareness, automaticity, and memory dissociations. In L.R. Squire & N. Butters (Eds.), Neuropsychology of memory (2nd Ed.). pp.46-57.
Toth, J.P., Reingold, E.M, & Jacoby, L.L. (1994). Toward a redefinition of implicit memory: Process dissociations following elaborative processing and self-generation. Journal of Experimental Psychology: Learning, Memory, and Cognition, in press.
Tulving, E. (1983). Elements of episodic memory. New York: Oxford University Press.
Tulving, E. (1985). How many memory systems are there? American Psychologist, 40, 385-398.
Tulving E. & Schacter, D.L. (1990). Priming and human memory systems. Science, 247, 301-305.
Tulving, E., Schacter, D.L. & Stark, H.A. (1982). Priming effects in word-fragment completion are independent of recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 336-342.
Tulving, E. & Thompson, D.M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80, 352-373.
Underwood, B.J. (1965). False recognition produced by implicit verbal responses. Journal of Experimental Psychology, 70, 122-129.
Vriezen, E.R., Moscovitch, M., & Bellos, S. (in press). Repetition priming effects for words accompanied by repeated and non-repeated orienting questions. Journal of Experimental Psychology: Learning, Memory, and Cognition.
Warrington, E.K. & Weiskrantz, L. (1968). New method of testing long-term retention with special reference to amnesic patients. Nature, 217, 972-974.
Warrington, E.K. & Weiskrantz, L. (1970). Amnesic syndrome: Consolidation or retrieval? Nature, 228, 629-630.
Warrington, E.K. & Weiskrantz, L. (1974). The effect of prior learning on subsequent retention in amnesic patients. Neuropsychologia, 12, 419-428.
Wegner, D.M. (1993). Ironic processes of mental control. Psychological Review, 101, 34-52
Weldon, M.S. (1991). Mechanisms underlying priming on perceptual tasks. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 526-541.
Weldon, M.S. (1993). The time course of perceptual and conceptual contributions to word fragment completion priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 1010-1023.
Weldon, M.S. & Jackson-Barrett, J.L. (1993). Why do pictures produce priming on the word-fragment completion test? A study of encoding and retrieval factors. Memory & Cognition, 21, 519-528.
Weldon, M.S. & Roediger, H.L. (1987). Altering retrieval demands reverses the picture superiority effect. Memory & Cognition, 15, 269-280.
Winnick, W.A. & Daniel, S.A. (1970). Two kinds of response priming in tachistoscopic word recognition. Journal of Experimental Psychology, 84, 74-81.
Woodworth, R.S. & Schlosberg, H.E. (1954). Experimental Psychology. New York: Holt.