Draft of: Merikle, P. M., & Reingold, E. M. (1992). Measuring unconscious perceptual processes. In R. F. Bornstein & T. S. Pittman (Eds.), Perception without awareness (pp., 55-80). New York: Guilford Press.
Measuring Unconscious Perceptual Processes
Philip M. Merikle and Eyal M. Reingold
MEASURING UNCONSCIOUS PERCEPTUAL PROCESSES
Nearly 300 years ago, Leibniz in his New Essays on Human Understanding reached the following conclusions.
There are hundreds of indications leading us to conclude that at every moment there is in us an infinity of perceptions, unaccompanied by awareness or reflection; that is, of alterations in the soul itself, of which we are unaware because the impressions are either too minute or too numerous, or else too unvarying, so that they are not sufficiently distinctive on their own. But when they are combined with others they do nevertheless have their effect and make themselves felt, at least confusedly, within the whole. (Leibniz, 1981, p. 53)
That is why we are never indifferent, even when we appear to be most so, as for instance over whether to turn left or right at the end of a lane. For the choice that we make arises from these insensible stimuli, which, mingled with the actions of objects and our bodily interiors, make us find one direction of movement more comfortable than the other. (Leibniz, 1981, p. 166)
Although Leibniz was not the first person to recognize the importance of unconscious cognitive processes, he was one of the first to give a clear statement of the possible role that unconscious perceptual processes may play in directing behavior. However, even though the conceptual distinction between conscious and unconscious cognitive processes has a relatively long history (see Whyte, 1960 for an extended discussion), there is still considerable controversy over whether unconscious perceptual processes can be shown to play an important role in human experience. Despite many experimental reports during the past 100 years suggesting that behavior is influenced by unconscious perceptual processes, interpretation of these empirical findings remains controversial. For example, Holender (1986), after a review of recent evidence suggesting that stimuli may be perceived without conscious awareness, concluded that "on the basis of current evidence it is most likely that these stimuli were indeed consciously identified" (p. 1). Dixon (1986), on the other hand, after reviewing Holender's article, concludes "that the most interesting phenomenon to which Holender's paper draws attention is the extraordinary antipathy some people still have toward the idea that we might be influenced by things of which we are unaware. Would it be putting it too strongly to say it reminds one of the scepticism of 'flat earth theorists' when confronted with the alarming theory that the world is round?" (p. 30).
The controversy surrounding the concept of unconscious perception revolves almost entirely around questions concerning what constitutes an adequate behavioral measure of conscious experience. The most frequently followed experimental approaches are based on the dissociation paradigm. Ironically, with the dissociation paradigm, it is necessary to have an adequate measure of conscious perceptual processes before it is possible to demonstrate that a dissociation between two measures of perception implicates unconscious processes. Given that there is no consensus as to what constitutes an adequate measure of conscious perceptual experience, it is perhaps not surprising that there is also no consensus as to the role of unconscious perception in directing human behavior.
Fortunately, there are ways to resolve this long-standing controversy over the status of unconscious perception as an empirical phenomenon. In this paper, we first review the more traditional approaches to the study of unconscious perception. We indicate why the research based on these approaches is problematic and controversial. In light of the inherent problems with these traditional approaches, we discuss two alternative approaches. We believe that these alternative approaches have the potential to resolve some of the long-standing controversies that have continually plagued experimental investigations directed at the study of unconscious perceptual processes.
The Dissociation Paradigm: Traditional Implementations
The vast majority of studies directed at demonstrating unconscious perception have relied on a version of the dissociation paradigm (see Erdelyi, 1985, 1986). The basic logic underlying this paradigm requires comparisons between two different measures of perception. One measure is assumed to index the stimulus information available to awareness or consciousness, and the second measure is assumed to index the perception of stimulus information, independent of whether the information is available to consciousness. According to the underlying logic, perception without awareness is demonstrated whenever it can be shown that stimulus information which is completely unavailable to consciousness is nevertheless perceived and capable of influencing higher-level decision processes.
This frequently used version of the dissociation paradigm has three requirements or criteria that must be satisfied. First, an adequate measure of the perceptual information available to consciousness must be selected. Second, the selected measure of conscious perceptual experience must be shown to indicate null sensitivity. Finally, given that the measure of conscious awareness indicates null sensitivity, the second measure of perceptual processing must be shown to have significant sensitivity to the stimulus information.
Obviously, the success of any studies based on this experimental strategy depends critically upon the adequacy of the behavioral measure used to index conscious experience. In addition, the experimental data must establish that this measure exhibits null sensitivity. If either the measure of awareness is inadequate or null sensitivity is not established in a convincing manner, then any dissociation between the assumed measure of awareness and the second measure of perception provides no definitive evidence either for or against unconscious perception. These two issues, adequacy of the measure and adequacy of the procedures used to establish null sensitivity, are central to the continuing controversy over whether or not perception without awareness can be demonstrated convincingly. In fact, whenever studies are criticized, the criticisms inevitably focus on either the adequacy of the measures of awareness or the adequacy of the procedures used to establish that the measures exhibit null sensitivity.
Even though many different measures have been used to index conscious perceptual experience, almost all measures fall within one of two possible general classes. One class of measures is characterized by the fact that conscious awareness is indexed primarily by a subject's experiential self reports. In contrast, with the second class of measures, conscious perceptual experience is assumed to be indexed by some measure of a subject's discriminative abilities. Both types of measures have serious weaknesses. For this reason, approaches based on either class of measures have failed to provide noncontroversial evidence for perception without awareness. In the following sections, approaches based on each class of measures are briefly reviewed to highlight the critical weaknesses.
Self Reports of Conscious Experience. Historically, in the earliest experimental studies of unconscious perception, the behavioral measure used to index conscious awareness was simply subjects' self reports indicating whether or not the perceived stimulus information was useful for the experimental task. Many of these early studies were reviewed by Adams (1957), and the experiments reported by Sidis (1898) provide an excellent illustration of this general approach. Sidis showed subjects cards containing a single printed digit or letter. The interesting aspect of these experiments is that "the subject was placed at such a distance from the card that the character shown was far out of his range of vision. He saw nothing but a dim, blurred spot or dot" (p. 170). In fact, "the subjects often complained that they could not see anything at all; that even the black, blurred, dim spot often disappeared from their field of vision . . ." (p. 171). However, when Sidis asked the subjects to name the character on a card, their responses were correct considerably more often than would be expected on the basis of pure random guessing, even though many subjects expressed the belief ". . . that they might as well shut their eyes and guess" (p. 171). On the basis of these and similar findings, Sidis concluded that his experiments indicated ". . . the presence within us of a secondary subwaking self that perceives things which the primary waking self is unable to get at" (p. 171).
Many other investigators (e.g., Miller, 1939; Stroh, Shaw, & Washburn, 1908; Williams, 1938) have reported findings quite similar to those originally reported by Sidis (see Kihlstrom, this volume, for other examples). In fact, the phenomenon is so robust that Adams (1957) suggested its use as a classroom demonstration. Thus, the experimental evidence clearly indicates that normal subjects can make accurate perceptual discriminations even when they believe, as indicated by their self reports, that their perceptual experiences are inadequate to guide their choices. If one accepts the assumption that a adequate definition of conscious perceptual experience can be based solely on subjects' self reports, then the results from these studies provide definitive evidence for perception without awareness.
Many researchers, however, feel uncomfortable measuring conscious awareness solely in terms of self reports. A major reason for caution is that it is difficult to know what criteria individuals use when reporting their conscious experiences (Merikle, 1984). Statements indicating an absence of relevant conscious experience may simply reflect biases introduced by either the experimental instructions or an individual's preconceived ideas concerning the value of particular types of perceptual experiences for making decisions. For example, from the subjects' statements in the Sidis experiments, it is clear that, on at least some occasions, they saw both the cards and also "dim, blurred spots or dots" on these cards. Thus, self reports may only reflect an individual's own theories of how perceptual experience guides behavior rather than a true absence of conscious perceptual experience. As noted by Merikle (1984), the fundamental problem with using self reports as a measure of awareness is that it transfers the responsibility for defining awareness from the investigator to the observer.
Despite the problems in using self reports to distinguish between the presence and absence of relevant conscious experience, there are instances when self reports do seem to provide a completely adequate measure. For example, blindsight patients report no relevant conscious visual experiences when objects are presented to their blind fields. However, when these patients are required to guess concerning the size, shape, or orientation of these objects, their guesses demonstrate that they do in fact have considerable knowledge of these objects (see Weiskrantz, 1986, for a review). Likewise, amnesic patients demonstrate learning of skills such as reading mirror-inverted script and puzzle solving despite their claims indicating that they cannot remember ever performing these tasks before (see Shimamura, 1986, for review). Paradoxically, for reasons that are not entirely obvious, many researchers, ourselves included, seem more willing to accept the self reports of brain damaged patients than the self reports of normal subjects when inferring the presence or absence of conscious experiences.
Even though there are instances when all but the most skeptical person will accept self reports as a valid measure of conscious perceptual experience, most researchers reject any experimental approach for distinguishing conscious from unconscious perceptual processes that is based solely on self reports. If self reports were generally considered to provide an adequate measure of conscious experience, then there would be no controversy whatsoever over the empirical status of perception without awareness; the results reported by Sidis (1898) and others would be definitive. However, since the perception-without-awareness phenomenon still generates considerable controversy, it is obvious that it is not possible to provide compelling demonstrations of unconscious perception when conscious and unconscious experiences are distinguished solely on the basis of self reports.
Perceptual Discriminations As Measures of Conscious Experience. Given the problems associated with using self reports to measure of conscious experience, many researchers have followed Eriksen's (1960) earlier suggestion and defined awareness on the basis of tasks that measure perceptual discriminative capacities. Two typical measures used in recent studies are forced-choice, present-absent decisions (e.g., Balota, 1983; Dagenbach, Carr, & Wilhelmsen, 1989; Fowler, Wolford, Slade, & Tassinary, 1981; Groeger, 1984, 1986; Kemp-Wheeler & Hill, 1988; Marcel, 1980, 1983) and forced-choice discriminations among a small, known, set of stimulus alternatives (Avant & Thieman, 1985; Cheesman & Merikle, 1984, 1986; Purcell, Stewart, & Stanovich, 1983). Methodologically, measures of discriminative capacities have an important advantage relative to self reports; perceptual sensitivity can be assessed independent of the influence of preconceived biases. It is primarily because of this methodological advantage that measures of perceptual discrimination have been the preferred measures of consciousness in many recent studies.
Approaches to the study of perception without awareness based on measures of discriminative capacity are best illustrated by the classic Marcel (1983) studies. These experiments created considerable excitement when Marcel first presented his findings in 1974, as the results seemed to demonstrate that visual stimuli are perceived even when observers cannot discriminate between their presence or absence. Although Marcel explored several different methodologies, the experiments that had the greatest impact were the ones that appeared to demonstrate that decision times to target stimuli were influenced by nondetectable stimuli presented immediately prior to the onset of the targets. In addition, Marcel also demonstrated that the magnitude of the influence of nondetectable stimuli was virtually equivalent to the magnitude of the influence of clearly visible stimuli. Taken together, these findings seemed to provide convincing evidence for perception without awareness, especially when other investigators using similar methodologies found comparable results (e.g., Balota, 1983; Fowler et al., 1981; McCauley, Parmelee, Sperber, & Carr, 1980). In fact, these findings, taken as a whole, convinced many former skeptics that perception without awareness was indeed a valid phenomenon.
Given the potential significance of the findings reported by Marcel and subsequent investigators, these studies were carefully scrutinized by a number of critics (e.g., Cheesman & Merikle, 1985; Holender, 1986; Merikle, 1982; Nolan & Caramazza, 1982; Purcell et al., 1983). The major issue raised by these critics concerned whether the methodology used in these studies was adequate to establish that the measure of stimulus detection indicated a true null sensitivity. On the basis of these critiques, it now appears that the methodology used in these experiments was probably inadequate. When similar experiments but with better methodology for establishing null performance were conducted, no evidence for perception in the absence of discriminative responding was found (e.g., Cheesman & Merikle, 1984; Nolan & Caramazza, 1982; Purcell et al., 1983). Thus, what initially appeared to be convincing, relatively noncontroversial, evidence for perception without awareness became inconclusive evidence once the methodology for establishing null sensitivity was carefully scrutinized.
The controversy over whether it is possible to demonstrate perception in the absence of stimulus detection is certainly not settled. On one side of this on-going debate, there are investigators who claim to have demonstrated perception in the absence of stimulus detection with methodologies that they are immune to many of the earlier criticisms (e.g., Greenwald, Klinger, & Liu, 1989; Groeger, 1988; Kemp-Wheeler & Hill, 1988). On the other side of this debate, there are investigators, such as ourselves, who believe that any current evidence for perception in the absence of stimulus detection comes from studies in which the experimental methodology was totally inadequate to establish a true absence of an ability to respond discriminatively. For example, the studies reported by Kemp-Wheeler and Hill (1988) can be criticized because too few trials (i.e., 40) were used to establish an accurate threshold for null stimulus detection, and the studies reported by Greenwald et al., 1988) have been criticized because the apparent null sensitivity of their spatial discrimination task may have been completely artifactual (Doyle, 1990). Thus, the debate goes on, and there is no reason to expect that the issue will be resolved in the near future. In fact, given how difficult it is to demonstrate null sensitivity for any measure of perception (see Macmillan, 1986), it may be impossible to ever obtain compelling, noncontroversial evidence for perception without awareness, as long as the experimental logic requires a convincing demonstration that some perceptual discriminative capacity is nonexistent.
To a large extent, the entire controversy over perception without awareness stems from the failure to find a generally agreed upon measure of conscious experience (see also Marcel, this volume). Both classes of measures, self reports and perceptual discriminations, can and have been criticized for not providing completely adequate measures of conscious experience. A primary concern is that these measures, no matter how careful an experimenter is in the conduct of a study, may not provide a true indication of when subjects do not have access to any relevant conscious information. It is for this reason that critics of studies based on self reports are so concerned with possible preconceived biases and that critics of studies based on measures of perceptual discriminations are so concerned that null sensitivity has been established in a convincing manner.
This obsessive concern with the adequacy of the measure used to index conscious experience reflects a need to satisfy a critical underlying assumption, which is rarely stated explicitly. Namely, whenever null sensitivity is equated with null awareness, it is implied that the selected measure provides an exhaustive index of all relevant conscious experience. This is a very strong assumption that is often difficult to justify (see Reingold & Merikle, 1990). However, this assumption of exhaustiveness must be made before any dissociation between two indices of perception can be interpreted as evidence for perception without awareness. If this assumption is not made, then it is always possible that relevant, conscious, stimulus information was available despite either subjects' claims to the contrary or results indicating that the selected measure of perceptual discriminative capacity exhibits null sensitivity. Thus, if the selected measure of consciousness is not potentially sensitive to all relevant conscious experience, then any dissociation between this measure and another measure of perception may simply indicate that the two measures are maximally sensitive to different aspects of consciously perceived information.
Experimenters are often tempted to select the most sensitive measure of perception as their indicator of conscious awareness. Intuitively, sensitive measures such as stimulus detection seem to satisfy the exhaustiveness assumption. However, this research strategy cannot be successful unless the selected measure is sensitive to only conscious perceptual experiences, or in other words, the selected measure indexes conscious perceptual processes exclusively. If the selected measure is not an exclusive measure of conscious experience and is actually sensitive to both conscious and unconscious perceptual processes, then establishing null sensitivity may eliminate all evidence for perception, either conscious or unconscious. As noted by Bowers (1984), this approach may be tantamount to defining perception without awareness out of existence.
The exclusiveness assumption is probably more difficult to justify than the exhaustiveness assumption. In fact, why should any measure of perception be sensitive to consciously available information exclusively? In the absence of any good theory specifying how conscious and unconscious processes are related to particular behavioral measures, a more prudent a priori position may be not to make any assumptions concerning how particular measures are related to either conscious and unconscious processes. However, if no assumptions are made, then correct responses on any particular measure may reflect conscious, unconscious, or both conscious and unconscious influences. Similarly, null sensitivity on any measure may reflect an absence of conscious, unconscious, or both conscious and unconscious processes. Thus, if no measure of perception can be assumed to reflect conscious processes exclusively, then all measures of perception are potentially exhaustive measures of both conscious and unconscious processes.
This discussion of underlying assumptions illustrates why it has been so difficult to demonstrate perception without awareness in a convincing manner. The logic underlying traditional implementations of the dissociation paradigm requires an adequate measure of awareness before any dissociation between measures can be considered as evidence for perception without awareness. At the very least, it must be assumed that the selected measure of consciousness is exhaustive in the sense that it is a sensitive of all relevant conscious experience. Otherwise, any demonstration of perception in the absence of sensitivity on the selected measure cannot provide convincing evidence for perception without awareness. On the other hand, even if the exhaustiveness assumption can be justified, it may still be impossible to obtain convincing evidence for perception without awareness. If the selected measure of consciousness is also an exhaustive measure of unconscious perceptual processes, then establishing null sensitivity may eliminate all evidence for perception whatsoever. Ideally, one would like to an exhaustive measure that exclusively indexes relevant conscious perceptual experiences. Unfortunately, such a measure does not exist.
It is doubtful that the long-standing controversy over perception without awareness will ever be resolved solely on the basis of studies that simply demonstrate a dissociation between two measures of perception. No matter what measure of perception is used to index conscious experience, there will always be a considerable number of skeptics who question whether the selected measure actually provides a satisfactory exhaustive index of all relevant conscious experience. For this reason, it may be impossible to ever demonstrate perception without awareness in a convincing, noncontroversial manner through traditional implementations of the dissociation paradigm. Clearly, to resolve this controversy, alternative approaches are required.
In the following sections, we discuss two alternative approaches to the study of perception without awareness (see Erdelyi, this volume, for another alternative). One approach emphasizes how qualitative differences can be used as converging evidence to validate measures of perception as possible exhaustive indices of relevant conscious experience. The second approach illustrates how much of the controversy over the measurement of conscious experience can be bypassed by using a methodology that does not require any assumption that a particular measure necessarily provides either an exhaustive or an exclusive index of awareness. We believe that both of these approaches have considerable promise for resolving some of the more controversial issues that have continually plagued experimental studies of unconscious perceptual processes.
Qualitative Differences. One way to establish that a measure of perception does in fact provide an adequate exhaustive index of relevant conscious experience is to demonstrate that the selected measure predicts qualitative differences across conscious and unconscious perceptual states. Such qualitative differences can provide the converging evidence necessary to validate a measure as an exhaustive measure of conscious awareness. The potential importance of demonstrating qualitative differences between conscious and unconscious perceptual states has been emphasized previously (e.g., Merikle & Cheesman, 1986; Dixon, 1971; Shevrin & Dickman, 1980). However, despite this recognition, studies directed at demonstrating qualitative differences between conscious and unconscious processes (e.g., Chessman & Merikle, 1986; Groeger, 1984, 1988; Jacoby, this volume; Jacoby & Whitehouse, 1989; Marcel, 1980) are the exception rather than the rule. We suggest that unless measures are validated by satisfactory converging evidence, it may be impossible to ever establish satisfactory, non-controversial measures of awareness. Thus, we consider demonstrations of consistent, reliable qualitative differences across presumed conscious and nonconscious perceptual states critical for defining and discovering non-arbitrary measures of conscious experience.
Recently, we completed a series of studies that illustrates the value of an approach to the study of perception without awareness based on establishing qualitative differences (Merikle & Reingold, 1990). The studies were directed at determining whether stimulus detection can be shown to provide an adequate measure of conscious awareness. Even though stimulus detection has considerable intuitive appeal as a measure conscious awareness, demonstrations of perception in the absence of stimulus detection do not necessarily imply perception in the absence of conscious experience. As suggested by our previous discussion of underlying assumptions, correct stimulus detection does not necessarily reflect conscious experience, unless one makes the exclusiveness assumption, and failures to detect a stimulus do not necessarily indicate an absence of relevant conscious experience, unless one makes the exhaustiveness assumption. Our research strategy to escape this impasse over initial assumptions was to demonstrate that detect and nondetect states lead to qualitatively different behavioral consequences. In this way, we obtained converging evidence consistent with the assumption that a failure to detect a stimulus indicates an absence of relevant conscious perceptual experience.
The basic methodology used in these studies involved the concurrent measurement of stimulus detection and stimulus recognition. Both word and nonword stimuli were studied in separate but similar experiments. To illustrate the general methodology consider the experiment involving word stimuli outlined in Figure 1. As indicated by the figure, each trial consisted of three phases: stimulus presentation, stimulus detection, and stimulus recognition. Initially, either a word or a blank field was presented for 50 ms, and each stimulus presentation was both preceded and followed by a masking stimulus. Following each stimulus presentation, two words were presented together with the question, "Was either word presented?" The subjects task was simply to indicate "yes" or "no." The reason the two words were presented during the detection phase of the experiment was to ensure that the retrieval environments for the detection and the recognition tasks were matched. Once a subject made the detection decision on each trial, the same two words were presented a second time but the question was changed to "Which word was presented?"
In draft, Figure 1 would appear here
With concurrent measures of stimulus detection and stimulus recognition, the critical observations involve an analysis of stimulus recognition conditional on stimulus detection. The basic question concerns whether stimulus recognition occurs even when subjects fail to detect a stimulus. In other words, to use the terminology of signal detection theory, can subjects recognize "missed" stimuli? Data relevant to this question are shown in Figure 2. What this figure shows are the proportions correct recognition both when subjects correctly detected a stimulus (i.e., hits) and when subjects failed to detect a stimulus (i.e., misses). Considering only the data obtained when words were used as stimuli, it is clear that stimulus recognition exceeded a chance level of performance following both hits and misses. While no one will be very surprised to learn that correct recognition can occur following correct stimulus detection, it is interesting that correct recognition can also occur following a failure to detect a stimulus. What these data demonstrate is that even when the subjects claimed not to have seen a stimulus, as indicated by their "no" responses on the detection task, they nevertheless perceived sufficient information to recognize the correct stimulus, at least on some proportion of the trials. If it is assumed that stimulus detection is an exhaustive measure of consciousness, then a failure to detect a stimulus must indicate an absence of relevant conscious experience. Given this assumption, stimulus recognition in the absence of stimulus detection is evidence for unconscious perception.
In draft, Figure 2 would appear here
However, without converging evidence to support the adequacy of stimulus detection as a measure of consciousness, there is really no good reason to assume that stimulus detection satisfies the exhaustiveness assumption. In fact, an equally plausible interpretation of the dissociation between stimulus detection and stimulus recognition is that each measure is maximally sensitive to a different aspect of consciously perceived information. One way to resolve this ambiguity over the status of stimulus detection as a measure of consciousness is to obtain evidence indicating that detect and nondetect states are qualitatively different. It is for this reason that we also studied the relation between stimulus detection and stimulus recognition when the stimuli consisted of nonwords made by randomly rearranging the letters in the words. Our reasoning was that if recognition in the absence of detection reflects perception without awareness, then it should only occur with familiar stimuli that have preexisting memory representations. Thus, in contrast to the results found for words, it may be impossible to recognize nondetected, unfamiliar stimuli, such as nonwords.
The results for the nonword stimuli are also shown in Figure 2. Although stimulus recognition was possible following hits, the nonwords were not recognized at all following misses. This overall pattern of results for the nonword stimuli is quite dissimilar to the one found with words. As can be seen in Figure 2, when stimuli are detected (i.e., hits), both words and nonwords can be recognized. However, when stimuli are not detected (i.e., misses), only words can be recognized at a better than chance level of performance. To express these findings in somewhat different terminology, stimulus detection and stimulus recognition are dissociable measures when words are used as stimuli and nondissociable measures when nonwords are used as stimuli.
In draft, Figure 3 would appear here
Given the potential significance of this qualitative difference between detect and nondetect states, we conducted a similar experiment with a somewhat different methodology to confirm that words and nonwords do in fact lead to different patterns of results. The general methodology is illustrated in Figure 3. The major changes from the preceding experiments were that words, nonwords, and blank fields were intermixed and the recognition task required a lexical decision. Thus, on each trial, a word, a nonword or a blank field was presented. Following each stimulus presentation, the subjects were shown a blank field and were asked, "Was either a word or a nonword presented?" Once the subjects indicated their "yes" or "no" answer to the detection question, they were then asked "Was the stimulus a word or a nonword?"
The results of this experiment are shown in Figure 4, which shows the mean A' scores indicating sensitivity to the word and nonword stimuli following both hits and misses on the detection task. The most striking aspect of these data is that the overall pattern is indistinguishable from the findings found in the initial experiments. Following correct stimulus detection, the subjects could make correct lexical decisions to both words and nonwords, on at least some proportion of the trials. However, when the subjects failed to detect a stimulus, they were completely insensitive to the nonwords, even though they could occasionally decide correctly that a stimulus was a word. Taken together, these contrasting patterns of results for the word and nonword stimuli provide another demonstration that the detect and nondetect states are qualitatively different.
In draft, Figure 4 would appear here
These experiments involving the concurrent measurement of stimulus detection and stimulus recognition illustrate the importance of using qualitative differences to validate a measure of awareness. Not only do qualitative differences provide converging evidence to support assumptions concerning measures of awareness, but they also make alternative explanations more complex and less plausible. The differences that we observed between the detect and nondetect states are completely consistent with the assumption that stimulus detection provides an adequate measure of awareness. Presumably, the subjects perceived only familiar stimuli with preexisting memory representations in the nondetect or nonaware state, whereas in the detect or aware state, both familiar and unfamiliar stimuli were perceived. These results thus provide empirical evidence consistent the widespread intuitive notion that a failure to detect a stimulus indicates a complete absence of relevant conscious perceptual experience. Furthermore, the observed qualitative difference between the detect and nondetect states also constrains plausible alternative explanations. For example, it could be suggested that stimulus detection and stimulus recognition actually measure different aspects of consciously perceived information (see Erdelyi, 1986; Eriksen, 1960). However, for this alternative account to be plausible, a reasonable account of why stimulus familiarity interacts with the detection and recognition tasks must also be proposed. Thus, a real value in using qualitative differences to validate a measure of awareness is that such differences also constrain possible alternative interpretations.
Finally, it should be emphasized that predictable qualitative differences may constitute the ultimate empirical criteria for evaluating the conceptual distinction between conscious and unconscious processes. If no consistent qualitative differences can be found, what is the value of the distinction? Clearly, the distinction is much more interesting and important if conscious and unconscious perceptual processes can be shown to lead to qualitatively different behavioral consequences.
Relative Sensitivity of Comparable Direct and Indirect Measures. As an alternative to finding a satisfactory measure of consciousness, evaluations of the relative sensitivity of comparable direct and indirect measures provide a completely different approach for documenting unconscious perceptual processes. Previously, we suggested that comparisons of comparable direct and indirect measures have the potential to provide considerable information regarding the relation between consciousness and perception (Reingold & Merikle, 1988). An important advantage that such comparisons have relative to any approach requiring a satisfactory measure of conscious experience is that the study unconscious processes can be based on a much more minimal a priori assumption. In fact, if the relative sensitivity of direct and indirect measures is compared under appropriate conditions, it is possible to study unconscious perceptual processes without having either a satisfactory exhaustive measure or a satisfactory exclusive measure of consciousness.
To illustrate what we mean by the critical phrase comparable direct and indirect measures, consider the well-known experiment reported by Kunst-Wilson and Zajonc in 1980. Even though these investigators did not use the direct/indirect terminology, their study provides an excellent illustration of the basic task distinction that we wish to emphasize. In their study, subjects were initially shown ten irregular geometric shapes. Each shape was presented five times for a 1 ms. duration, and no subject ever reported seeing any of the shapes. Following these initial presentations, possible perception of the shapes was evaluated by both a recognition and a preference task. For both tasks, the subjects were shown 10 pairs of shapes, with each pair consisting of one old and one new shape. For the recognition task (i.e., the direct measure), the subjects were explicitly instructed to select the member of each pair that had been presented previously, whereas for the preference task (i.e., the indirect measure), the subjects were simply told to choose the shape that they preferred, with no reference being made to the distinction between old and new shapes. The interesting results were that the subjects chose the previously-presented shape in approximately 60% of the pairs when they selected the shape that they preferred, but when they selected the shape that they thought had been presented previously, their performance approximated a chance level (i.e., 50%). In other words, the indirect task (i.e., the preference task) was a more sensitive measure of the old/new discrimination than the direct, recognition task, even though the instructions for the recognition task explicitly directed subjects to select previously-presented shapes.
The important aspect of the Kunst-Wilson and Zajonc methodology is that the direct and indirect measures were matched on all characteristics except task instructions. Previously, we have suggested a number of criteria for evaluating the comparability of direct and indirect tasks (Reingold & Merikle, 1988). As applied to the Kunst-Wilson and Zajonc study, the important considerations are that the direct and indirect measures involved identical study conditions and that the test conditions for both measures involved a two-choice task that measured the subjects' sensitivity to the old/new discrimination. Thus, the two tasks were identical except for the instructions telling subjects either to select the previously-presented shape (i.e, the direct task) or to select the shape that they preferred (i.e., the indirect task).
The real advantage of studying matched direct and indirect tasks is that with a very minimal a priori assumption, any demonstration of greater sensitivity for the indirect than the direct measure provides strong evidence for unconscious perception. Although the exhaustiveness assumption is sometimes made, at least implicitly, when the greater sensitivity of an indirect measure is interpreted as evidence for unconscious perception (e.g., Kihlstrom, 1987), it is not necessary to make such a strong assumption if the direct and indirect measures are truly comparable except for the task instructions. Previously, we have suggested that greater sensitivity for an indirect than for a comparable direct measure necessarily implicates unconscious processes if it is simply assumed that the sensitivity of the direct measure to conscious task relevant information is equal to or greater than the sensitivity of the indirect measure to the same information (Reingold & Merikle, 1988). The rational underlying this assumption is that conscious information relevant to a particular stimulus discrimination, if it exists, should be used equally or more efficiently when subjects are instructed to make a discrimination (i.e., a direct task) than when subjects are not so instructed (i.e., an indirect task). Conversely, it is difficult to imagine circumstances in which conscious information relevant to a particular stimulus discrimination would enhance performance more when subjects are not instructed to make a discrimination than when subjects are explicitly instructed to make the discrimination.
If this minimal working assumption is accepted, then unconscious processes are implicated whenever an indirect measure is more sensitive than a comparable direct measure to the same perceptual discrimination. This is the case because the assumption rules out the possibility that superior performance on the indirect task is attributable to conscious task relevant information. Therefore, by default, whenever an indirect measure is more sensitive than a comparable direct measure, it must reflect the influence of unconscious, task relevant information. It is important to emphasize that this interpretation is warranted only if the direct and indirect measures are truly comparable except for a reference in the direct-task instructions to the discrimination of interest. If the measures are not comparable, then greater sensitivity for the indirect than the direct measure may simply reflect a methodological artifact rather than an unconscious influence.
Given the minimal a priori assumption required when comparable direct and indirect tasks are used to study unconscious influences, comparisons of the relative sensitivity of direct and indirect tasks offer a potentially non-controversial approach for the study of unconscious processes. One limitation, however, is that a review of the literature indicates that there may be severe boundary conditions for finding greater sensitivity of indirect than comparable direct measures. With one exception (Bornstein, Leone, & Galley, 1987), the pattern of results originally reported by Kunst-Wilson and Zajonc (1980) has only been demonstrated with one very specific stimulus set. In all reported replications (e.g., Bonnano & Stillings, 1986; Seamon, Marsh, & Brody, 1984; Mandler, Nakamura, & Van Zandt, 1987), the stimulus set has consisted solely of irregular polygons selected from Vanderplas and Garvin (1959). Clearly, to establish the generality of this approach, it is necessary to compare the relative sensitivity of comparable direct and indirect tasks across a variety of experimental manipulations and sets of stimulus materials.
With these considerations in mind, we recently explored the effects that the allocation of visual attention may have on the relative sensitivity of comparable direct and indirect measures (Merikle & Reingold, 1991). Earlier findings suggest that the relation between direct and indirect measures may be considerably different for previously attended than for previously nonattended stimuli. For example, Eich (1984), in a rather well-known study, demonstrated that memory for non-shadowed, and presumably nonattended, auditory information can be demonstrated via an indirect measure even when a test of recognition memory indicates no memory whatsoever. If Eich's findings reflect a fairly general phenomenon, then it should be possible to demonstrate similar findings with other pairs of comparable direct and indirect measures. Such results could provide strong evidence that minimally attended or even nonattended information can exert a significant unconscious influence.
Our experimental methods are illustrated in Figure 5. As shown in the figure, there was an initial study phase followed by a subsequent test phase. During the study phase, 60 pairs of words were presented, with each pair being presented for approximately 500 ms. The subjects initiated the presentation of each pair, and they were instructed to read the one word in each pair indicated by the arrowheads. The general instructions to the subjects implied that we were interested in determining how rapidly they could read the cued word in each of the pairs.
The test phase of the experiment immediately followed completion of the study phase. It consisted of two blocks of 48 test trials, and on each trial, one word was presented against a background field of randomly-selected dots, which were used to degrade the visual quality of the words. Half of the words were old words that had been presented as uncued words during the study phase and the remaining words were new words that the subjects had not seen in the experiment. Different groups of subjects were tested with the direct and the indirect tasks. As indicated in Figure 5, the direct task was a straightforward recognition task that required subjects to decide whether each word was old or new. The indirect task, on the other hand, was based on recent findings indicating that the perceived contrast between words and background noise is greater for words previously encountered in an experiment (Jacoby, Allan, Collins, & Larwill, 1988). Given these findings, we asked our subjects to decide whether each word was presented under high or low contrast conditions. High contrast was defined as a word that appeared to "stand out" from the background, and low contrast was defined as a word that appeared to "blend" into the background. If the contrast task is a sensitive measure of memory, then old words should be judged more often than new words to have high contrast.
In draft, Figure 5 would appear here
The sensitivity of the direct and indirect tasks to memory for the previously uncued words is shown in Figure 6 for each of the two blocks of 48 test trials. Inspection of this figure shows that evidence for unconscious processes was found in the first block of trials, as the sensitivity of the contrast task in this block of trials was greater than the sensitivity of the recognition task. Statistical analyses confirmed that the relatively modest sensitivity of the contrast measure in the first block of trials was both significantly greater than chance and significantly greater than the sensitivity of recognition measure, which did not differ from chance. Performance in the second block of trials, on the other hand, provided no evidence either for or against unconscious processes, as the results indicated that the recognition task was slightly but not significantly more sensitive than the contrast task. Although the similar sensitivities of the contrast and recognition tasks in the second block of trials can lead to some interesting speculations (see Merikle & Reingold, 1991, for one speculative interpretation), for present purposes, the only important finding of this study is the greater sensitivity of the contrast task relative to the recognition task in the first block of trials.
In draft, Figure 6 would appear here
Given the logic underlying comparisons of the relative sensitivity of direct and indirect measures, the greater sensitivity of the contrast measure relative to the recognition measure necessarily implicates unconscious processes. This conclusion follows from the minimal a priori assumption that the sensitivity of the direct, recognition task to relevant conscious information is either greater than or equal to the sensitivity of the indirect, contrast task. Likewise, given this assumption, previous findings demonstrating greater sensitivity for indirect than for comparable direct measures (e.g., Kunst-Wilson & Zajonc, 1980) must also be considered as unequivocal demonstrations of unconscious influence. Thus, both our results and previous results provide convincing demonstrations of the importance of unconscious processes. In addition, our results extend the generality of previous findings considerably and suggest that greater relative sensitivity of indirect than direct measures is a phenomenon that may occur under a wide variety of conditions (see Bornstein, this volume, for another possible example).
As an approach for establishing unconscious processes, comparisons of the relative sensitivity of comparable direct and indirect measures have distinct conceptual and methodological advantages relative to many other approaches. Conceptually, much of the controversy over the measurement of consciousness is bypassed, since the approach does not require either a satisfactory exhaustive or a satisfactory exclusive measure of conscious awareness to be identified. In fact, given the minimal proposed a priori assumption, all behavioral measures potentially may reflect both conscious and unconscious influences. This is an inherently reasonable a priori position that should be maintained unless there is strong evidence to the contrary. An emphasis on the relative sensitivity of different measures also leads to an important methodological advantage; it is no longer necessary to establish that any measure necessarily indicates null sensitivity. When the relative sensitivity of different measures is emphasized, the influence of unconscious processes can be demonstrated convincingly, even when both measures can show significant sensitivity. The important empirical result is simply to demonstrate a greater sensitivity for the indirect than the direct measure. With these conceptual and methodological advantages, comparisons of the relative sensitivity of direct and indirect measures offer a potentially valuable methodology for documenting a variety of unconscious influences in perception and memory.
Despite more than 100 years of experimental studies directed at demonstrating unconscious perceptual processes, decisive experiments remain elusive. In fact, given the available evidence, it is still possible to argue that unconscious perceptual processes have not been shown to play any important role in directing human behavior (e.g., Holender, 1986). The major impediment to progress in the study of unconscious perceptual processes has been the failure to find a satisfactory measure of conscious awareness. Such a measure is required by the most frequently used methodologies. However, given that a totally adequate measure of consciousness has yet to be identified, alternative methodologies that do not require either an exclusive or an exhaustive measure consciousness can provide much more compelling evidence for unconscious influences than is provided by the more traditional methodologies.
The study of unconscious processes has also been impeded by a preoccupation, on the part of many researchers, with proving either the existence or the nonexistence of unconscious influences. Given the absence of definitive experiments, questions concerning the "existence" of unconscious perception have certainly not provided a successful research focus to date. As an alternative, it may prove more fruitful to ask how conscious and unconscious perceptual processes differ. The distinction between conscious and unconscious perceptual processes is certainly much more important if conscious and unconscious processes differ in interesting ways than if unconscious perceptual processes are simply quantitatively weaker versions of conscious perceptual processes. For this reason, we suggest that the research focus should change from "existence proofs" to analyses of the possible characteristics that distinguish conscious from unconscious processes. In this way, it should be possible to evaluate the value of the long-standing conceptual distinction between conscious and unconscious perceptual processes.
Adams, J. K. (1957). Laboratory studies of behavior without awareness. Psychological Bulletin, 54, 383-405.
Avant, L. L., & Thieman, A. A. (1985). On visual access to letter case and lexical/semantic information. Memory & Cognition, 13, 393-404.
Balota, D. A. (1983). Automatic semantic activation and episodic memory. Journal of Verbal Learning and Verbal Behavior, 22, 88-104.
Bonnano, G. A., & Stillings, N. A. (1986). Preference, familiarity, and recognition after brief exposures to random geometric shapes. American Journal of Psychology, 99, 403-415.
Bornstein, R. F., Leone, D. R., & Galley, D. J. (1987). The generalizability of subliminal mere exposure effects: Influence of stimuli perceived without awareness on social behavior. Journal of Personality and Social Psychology, 53, 1070-1079.
Bowers, K. S. (1984). On being unconsciously influenced and informed. In K. S. Bowers & D. Meichenbaum (Eds.), The unconscious reconsidered (pp. 227-272). New York: Wiley.
Cheesman, J., & Merikle, P. M. (1984). Priming with and without awareness. Perception & Psychophysics, 36, 387-395.
Cheesman, J., & Merikle, P. M. (1985). Word recognition and consciousness. In D. Besner, T. G. Waller, & G. E. MacKinnon (Eds.), Reading research: Advances in theory and practice (Vol. 5, pp. 311-352). New York: Academic Press.
Cheesman, J., & Merikle. P. M. (1986). Distinguishing conscious from unconscious perceptual processes. Canadian Journal of Psychology, 40, 343-367.
Dagenbach, D., Carr, T. H., & Wilhelmsen, A. (1989). Task-induced strategies and near-threshold priming: Conscious influences on unconscious perception. Journal of Memory and Language, 28, 412-443.
Dixon, N. F. (1971). Subliminal perception: The nature of a controversy. New York: McGraw-Hill.
Dixon, N. F. (1986). On private events and brain events. Behavioral and Brain Sciences, 9, 29-30.
Doyle, J. R. (1990). Detectionless processing with semantic activation? A footnote to Greenwald, Klinger, and Liu (1989). Memory & Cognition, 18, 428-429.
Eich, E. (1984). Memory for unattended events: Remembering with and without awareness. Memory & Cognition, 12, 105-111.
Erdelyi, M. H. (1985). Psychoanalysis: Freud's cognitive psychology. New York: Freeman.
Erdelyi, M. H. (1986). Experimental indeterminacies in the dissociation paradigm. Behavioral and Brain Sciences, 9, 30-31.
Eriksen, C. W. (1960). Discrimination and learning without awareness: A methodological survey and evaluation. Psychological Review, 67, 279-300.
Fowler, C. A., Wolford, G., Slade, R., & Tassinary, L. (1981). Lexical access with and without awareness. Journal of Experimental Psychology: General, 110, 341-362.
Greenwald, A. G., Klinger, M. R., & Liu, T. J. (1989). Unconscious processing of dichoptically masked words. Memory & Cognition, 17, 35-47.
Groeger, J. A. (1984). Evidence of unconscious semantic processing from a forced-error situation. British Journal of Psychology, 75, 305-314.
Groeger, J. A. (1986). Predominant and non-predominant analysis: Effects of level of presentation. British Journal of Psychology, 77, 109-116.
Groeger, J. A. (1988). Qualitatively different effects of undetected and unidentified auditory primes. Quarterly Journal of Experimental Psychology, 40A, 323-339.
Holender, D. (1986). Semantic activation without conscious identification in dichotic listening, parafoveal vision, and visual masking: A survey and appraisal. Behavioral and Brain Sciences, 9, 1-23.
Jacoby, L. L. & Whitehouse, K. (1989). An illusion of memory: False recognition influenced by unconscious perception. Journal of Experimental Psychology: General, 118, 126-135.
Jacoby, L. L., Allan, L. G., Collins, J. C., & Larwill, L. K. (1988). Memory influences subjective experience: Noise judgements. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 240-247.
Kemp-Wheeler, S. M., & Hill, A. B. (1988). Semantic priming without awareness: Some methodological considerations and replications. Quarterly Journal of Experimental Psychology, 40A, 671-692.
Kihlstrom, J. F. (1987). The cognitive unconscious. Science, 237, 1445-1452.
Kunst-Wilson, W. R., & Zajonc, R. B. (1980). Affective discrimination of stimuli that cannot be recognized. Science, 207, 557-558.
Leibniz, G. W. (1981). New essays on human understanding. Translated and edited by P. Remnant & J. Bennett. Cambridge: Cambridge University Press.
Macmillan, N. A. (1986). The psychophysics of subliminal perception. Behavioral and Brain Sciences, 9, 38-39.
Mandler, G., Nakamura, Y., Van Zandt, B. J. S. (1987). Nonspecific effects of exposure on stimuli that cannot be recognized. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 646-648.
Marcel, A. J. (1974, July). Perception with and without awareness. Paper presented at the meeting of the Experimental Psychology Society, Stirling, Scotland.
Marcel, A. J. (1980). Conscious and preconscious recognition of polysemous words: Locating the selective effects of prior verbal context. In R. S. Nickerson (Ed.), Attention and performance VIII (pp. 435-457). Hillsdale, NJ: Erlbaum.
Marcel, A. J. (1983). Conscious and unconscious perception: Experiments on visual masking and word recognition. Cognitive Psychology, 15, 197-237.
McCauley, C., Parmelee, C. M., Sperber, C. D., & Carr, T. H. (1980). Early extraction of meaning from pictures and its relation to conscious identification. Journal of Experimental Psychology: Human Perception and Performance, 6, 265-276.
Merikle, P. M. (1982). Unconscious perception revisited. Perception & Psychophysics, 31, 298-301.
Merikle, P. M. (1984). Toward a definition of awareness. Bulletin of the Psychonomic Society, 22, 449-450.
Merikle, P. M., & Cheesman, J. (1986). Consciousness is a "subjective" state. Behavioral and Brain Sciences, 9, 42-43.
Merikle, P. M., & Reingold, E. M. (1990). Recognition and lexical decision without detection: Unconscious perception? Journal of Experimental Psychology: Human Perception and Performance, 16, 574-583.
Merikle, P. M., & Reingold, E. M. (1991). Comparing direct (explicit) and indirect (implicit) measures to study unconscious memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 224-233.
Miller, J. G. (1939). Discrimination without awareness. American Journal of Psychology, 52, 562-578.
Nolan, K. A., & Caramazza, A. (1982). Unconscious perception of meaning: A failure to replicate. Bulletin of the Psychonomic Society, 20, 23-26.
Purcell, D. G., Stewart, A. L., & Stanovich, K. E. (1983). Another look at semantic priming without awareness. Perception & Psychophysics, 34, 65-71.
Reingold, E. M., & Merikle, P. M. (1988). Using direct and indirect measures to study perception without awareness. Perception & Psychophysics, 44, 563-575.
Reingold, E. M., & Merikle, P. M. (1990). On the inter-relatedness of theory and measurement in the study of unconscious processes. Mind & Language, 5, 9-28.
Seamon, J. G., Marsh, R. L., & Brody, N. (1984). Critical importance of exposure duration for affective discrimination of stimuli that are not recognized. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 465-469.
Shevrin, H., & Dickman, S., (1980). The psychological unconscious: A necessary assumption for all psychological theory?. American Psychologist, 35, 421-434.
Shimamura, A. P. (1986). Priming effects in amnesia: Evidence for a dissociable memory function. Quarterly Journal of Experimental Psychology, 38A, 619-644.
Sidis, B. (1898). The psychology of suggestion. New York: Appleton. (Reprint Edition, Arno Press, 1973).
Stroh, M. A., Shaw, M., & Washburn, M. F. (1908). A study of guessing. American Journal of Psychology, 19, 243-245.
Vanderplas, J. M., & Garvin, E. A. (1959). The association value of random shapes. Journal of Experimental Psychology, 57, 147-154.
Weiskrantz, L. (1986). Blindsight. New York: Oxford University Press.
Whyte, L. L. (1960). The unconscious before Freud. New York: Basic Books.
Williams, A. C. (1938). Perception of subliminal visual stimuli. Journal of Psychology, 6, 187-199.
Preparation of this manuscript was facilitated by Grant APA-231 from the Natural Sciences and Engineering Research Council of Canada to the first author.