Current Status of Research on Subliminal Perception

Philip M. Merikle, University of Waterloo
Jim Cheesman, University of Saskatchewan
ABSTRACT - Four experimental approaches to the study of subliminal perception are summarized and evaluated. It is concluded a) that ewidence favoring subliminal perception is only obtained when subliminal or unconscious perception is defined in terms of subjective criteria, and b) that an approach based on subjective criteria can provide a basis for establishing the critical differences that distinguish conscious from unconscious perceptual processes.
[ to cite ]:
Philip M. Merikle and Jim Cheesman (1987) ,"Current Status of Research on Subliminal Perception", in NA - Advances in Consumer Research Volume 14, eds. Melanie Wallendorf and Paul Anderson, Provo, UT : Association for Consumer Research, Pages: 298-302.

Advances in Consumer Research Volume 14, 1987      Pages 298-302

CURRENT STATUS OF RESEARCH ON SUBLIMINAL PERCEPTION

Philip M. Merikle, University of Waterloo

Jim Cheesman, University of Saskatchewan

[Preparation of this paper was supported by grant from the Natural Sciences and Engineering Research Council of Canada.]

[Correspondence concerning this paper should be addressed to Philip M. Merikle, Department of Psychology, University of Waterloo, Waterloo, Ontario N2L 3G1 Canada.]

ABSTRACT -

Four experimental approaches to the study of subliminal perception are summarized and evaluated. It is concluded a) that ewidence favoring subliminal perception is only obtained when subliminal or unconscious perception is defined in terms of subjective criteria, and b) that an approach based on subjective criteria can provide a basis for establishing the critical differences that distinguish conscious from unconscious perceptual processes.

INTRODUCTION

The concept of subliminal perception has generated considerable debate for-many years. However, despite numerous studies directed at establishing whether or not unconscious perceptual processes provide sufficient information to influence higher-level decision processes (see Dixon 1971, 1981; Holender 1986 for extensive reviews), there are still no generally-accepted answers. Much of this continuing controversy is centered on questions concerning how perceptual awareness is best defined and measured. In fact, an answer to the question, "Does subliminal perception occur?", depends entirely on both a) how "subliminal" or the threshold for perceptual awareness is defined and b) the adequacy of the procedures used to measure this threshold. In this paper, three major experimental approaches that have been followed at various times to study subliminal perception are summarized. Each approach, considered by itself, has critical weaknesses. For this reason, a fourth approach, based on recent empirical work (e.g., Cheesman and Merikle 1985), is described. This fourth approach not only resolves many of the controversies that have plagued investigations of subliminal perception but it also provides a basis for establishing the critical differences that may distinguish conscious from unconscious perceptual processes.

Prior to considering the different experimental approaches, it is important to define exactly what psychologists mean by awareness or conscious perceptual processing. Awareness is typically measured in terms of an observer's reports concerning his/her perceptual experiences. This approach simply reflects an attempt to operationalize the phenomenal experiences that we all have when our sense organs are stimulated. Typically, stimulation of the sense organs leads to phenomenal perceptual experiences that can be described in some d tail. On the other hand, when stimulation of sense organs toes not lead to phenomenal perceptual experiences, no descriptions are possible. Thus, definitions of awareness are typically tied to some type of verbal response mate following stimulation of the sense organs. The controversies have concerned both the exact type of response that is the most accurate indicator of perceptual awareness and the most appropriate way the particular response that is considered to reflect awareness should be measured.

STUDIES BASED ON CLAIMED AWARENESS

Historically, in the earliest studies directed at establishing the possible efficacy of unconscious perceptual processes in determining behavior, a definition of awareness was adopted that was based solely on an observer's subjective confidence that the perceived stimulus information was useful for the required task. Many of these studies have been reviewed by Adams (1957), and only one representative study will be discussed. A very early study was reported by Sidis (1898) who had observers guess the specific letter or numeral printed on cards that were displayed so far away from the observers that they had no confidence they were doing anything other than guessing. Sidis found that the observers' "guesses" were correct considerably more often than would be expected on the basis of pure random guessing and therefore concluded that the results 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). Similar findings have been reported by many other investigators (e.g., Miller 1939; Stroh, Shaw and Washburn 1908; Williams 1938), and thus there is every reason to believe that this is a relatively robust, easily replicable phenomenon.

If one accepts the assumption that a good definition of conscious perceptual processing can be based solely on observers' reports concerning how confident they are that useful perceptual information has been perceived, then the results obtained in these studies provide definitive ewidence for the existence of subliminal perception. However, most psychologists do not accept these results as providing any sort of definitive support for the reality of subliminal perception. A major weakness with this type of study is that it is impossible to establish exactly what criteria observers are using to decide that they are guessing (cf., Merikle 1984). More generally, with this type of approach, it is impossible to determine if observers' reports indicating no subjective confidence are based on preconceived biases or on a true absence of phenomenal perceptual information. It is for this reason that most researchers, with a few notable exceptions (e.g., Dixon 1981, Henley 1984), reject any approach to the study of subliminal perception in which unconscious perception is simply equated with observers' reports indicating a lack of subjective confidence.

STUDIES BASED ON INCORRECT IDENTIFICATION RESPONSES

Another approach to the study of subliminal perception has based the definition of unconscious perception upon a classification of observers' responses in a difficult visual identification task. With this approach, it is assumed that correct identification of a visual stimulus indicates conscious perception and that incorrect identification indicates a lack of awareness for the stimulus. In one of the classic studies in which this approach was followed (Lazarus and McCleary, 1951), ten nonsense syllables were presented to observers and presentation of five of these syllables was paired with the presentation of electric shock. Following this initial conditioning, the ten nonsense syllables were presented tachistoscopically under conditions that made it difficult for the observers to identify the syllables correctly. Two different responses were measured during this phase of the experiment: a) identification accuracy and b) galvanic skin response (GSR). When the observers' identification responses were classified on the basis of whether they were correct of incorrect, Lazarus and McCleary found that the GSR responses were greater in magnitude following the presentation of syllables previously paired with shock, independent of whether or not these syllables hat been correctly identified. Lazarus and McCleary assumed that the GSR, being mediated by the autonomic nervous system, was sensitive to both conscious and unconscious perceptual processes. Therefore, they concluded that subliminal perception, defined as perception below the threshold for identification. had been demonstrated.

The publication of Lazarus and McCleary's findings led to considerable debate that sharpened a number of important methodological issues. Eriksen's (1960) critique is particularly important. He questioned the validity of Lazarus and McCleary' 8 conclusions by noting that the partial correlation between the two responses used to indicate perceptual processing (i.e., identification and GSR) may have occurred because the responses were not equally sensitive to conscious, phenomenal, perceptual information. According to Eriksen, the GSR response, being a continuous variable, may be a more sensitive indicator of perceptual experience than the dichotomous identification measure (i.e., correct vs. incorrect). A differential sensitivity would lead to a partial correlation between measures and would account for the apparent dissociation between measures observed in the Lazarus and McCleary study without recourse to the concept of subliminal perception. Furthermore, it is reasonable to expect that any difference in sensitivity between measures would be maximized when stimuli are barely visible, as they are in any study of subliminal perception.

A particularly important consideration that provides support for Eriksen's criticisms comes from an examination of the overall identification performance in the Lazarus and McCleary study. When identification accuracy was averaged across all ten nonsense syllables, it was found that all observers identified considerably more syllables than would be expected on the basis of chance guessing. Thus, overall identification performance was not at a chance level, which is the usual definition of a detection or identification threshold, and therefore, by definition, it must be concluded that the observers in the Lazarus and McCleary study were aware of the stimulus information. On the basis of such considerations, Eriksen reached the conclusion in 1960 that verbal reports are as sensitive an indicator of perception as any other response that has been studied (e.g., GSR). Given that verbal reports are assumed to indicate conscious perceptual experience, then before subliminal perception is demonstrated, it must be shown perceptual processing occurs in the absence of correlated verbal reports.

STUDIES BASED ON VISUAL MASKING

Eriksen's (1960) conclusions concerning the validity of subliminal perception were not seriously challenged until 1974 when Marcel reported an extensive series of experiments which appeared to demonstrate that visual stimuli are perceived even when observers cannot detect their presence. Marcel's (1983a) experiments were based on a well-documented phenomenon indicating that decision times to visual target stimuli can be either facilitated or inhibited by the presentation of a related stimulus (i.e., a "prime" in psychological jargon) immediately before the onset of a target (e.g., Dyer and Severance 1973; Meyer, Schvaneveldt and Ruddy 1975). The striking result found by Marcel was that decision times to targets were greatly influenced by primes even when the visibility of the primes was degraded to such an extent by backward masking that they were impossible to detect, as indicated by the observers' verbal reports. In fact, the magnitude of the effect produced by masked primes was approximately the same size as the magnitude of the effect produced when the primes were clearly visible. In subsequent experiments, other investigators, employing similar methodologies, have also obtained results comparable to those originally reported by Marcel (e.g., Balota 1983; Fowler et al. 1981; McCauley et al. 1980). Thus, if verbal reports are assumed to indicate perceptual awareness, the results of these studies appear to provide definitive ewidence that stimuli may be Perceived without awareness.

Given the potential significance of the results from these studies involving masked primes, it is important to consider the general methodology in some detail. The procedures followed in a comparable experiment reported by Cheesman and Merikle (1984) will serve for purposes of illustration. In this experiment, a variant of the Stroop (1935) color-word interference task was used, and the task for the observers was to name a color (i.e., the target) that was immediately preceded by either a congruent or an incongruent color word (i.e., the prime). Previous research using comparable displays has indicated that the reaction time to name a target color is increased when it is preceded by an incongruent color word (e.g., Dyer and Severance 1973) and is decreased when it is preceded by a congruent color word (e.g., Dalrymple-Alfort 1972). The general characteristics of the stimulus displays are illustrated in Figure 1. The important points to note are that a) the color words which served as primes were always presented prior to the centrally-located target colors, b) the presentation of each color word was followed by the presentation of a masking stimulus, and c) each color word was either congruent or incongruent with the name of the target color. The critical aspect of the methodology is that the temporal interval between the prime and the masking stimulus was either sufficiently long to permit identification of the primes or so short that the observers could not distinguish among the four different primes that were used.

FIGURE 1

EXAMPLES OF STIMULUS DISPLAYS

The general pattern of results found in a number of studies in which a methodology similar to the one described above has been used is illustrated by the dotted line in Figure 2. The critical aspect of these results is that prime effectiveness appears to be completely independent of verbal report accuracy. In other words in these studies, detection or identification of the primes, as indicated by verbal report accuracy, had no relation to how effective the primes were at influencing decision times to the targets (e.g., Balota 1983; Fowler et al. 1981; Marcel 1983a; McCauley et al. 1980). Furthermore, even when the primes were neither detected nor identified, they were perceived since, as indicated in the Figure, they were still capable of influencing target decision times. Given the assumption that above chance verbal report accuracy indicates awareness and that chance level verbal report accuracy indicates a lack of awareness, findings such as those illustrated by the dotted line in Figure 2 provide strong, compelling ewidence for the validity of subliminal perception.

FIGURE 2

ILLUSTRATION OF TWO PATTERNS OF RESULTS

Unfortunately, the pattern of results illustrated by the dotted line in Figure 2 has not been found in all studies directed at assessing the effectiveness of masked primes. In other studies, the pattern of results has been much more similar to the one illustrated by the solid line in the Figure (e.g., Cheesman and Merikle 1984, in press). This alternative pattern of results is theoretically very boring, as it shows that verbal report accuracy and prime effectiveness are perfectly correlated. More importantly, these alternative results also indicate that primes are completely ineffective when verbal report accuracy is at a chance level of accuracy. Thus, the results illustrated by the solid line in Figure 2 provide no ewidence whatsoever indicating that primes are perceived without awareness. In fact, these results are completely consistent with Eriksen's (1960) earlier conclusion suggesting that verbal reports are as sensitive an indicator of perception as any other response that has been studied.

Previously, we have suggested that the apparent conflict between the two patterns of results illustrated in Figure 2 can be resolved by considering the procedures used to establish awareness thresholds in the different studies (Cheesman and Merikle 1985, in press). In the studies in which results similar to those illustrated by the dotted line in the Figure were obtained, awareness thresholds were determined on the basis of very few observations. Therefore, it is very probable that the threshold or boundary separating chance from above chance verbal report accuracy was not measured reliably (cf., Merikle 1982). On the other hand, in the studies in which results similar to those illustrated by the solid line in the Figure were obtained, awareness thresholds were based on a considerably greater number of observations. Thus, it is much more likely that the thresholds in these latter studies were measured reliably. This consideration of the procedures used to determine awareness thresholds suggests that the thresholds established in the studies which appear to demonstrate subliminal perception may have been more indicative of the observers' claimed awareness than of the observers' actual ability to indicate verbally when a stimulus was detected or identified (e.g., Cheesman and Merikle 1985, in press). If this conjecture is correct, then it is entirely possible that the two different patterns of results found in studies involving masked primes are not necessarily contradictory.

However, if thresholds for claimed awareness rather than thresholds for detection or identification were established in the studies that appear to indicate that prime effectiveness is independent of verbal report accuracy, then the results of these studies do not necessarily provide ewidence favoring subliminal perception. A central assumption underlying these studies was that perceptual awareness is best defined in terms of an observer's ability to give verbal reports concerning stimuli at a better than chance level of accuracy. Since it is unlikely that the procedures used in these studies ensured that this assumption was satisfied, the results of these studies cannot be considered definitive. On the other hand, the results of these studies involving masked primes are entirely consistent with earlier studies of subliminal perception in which it was established that considerable perceptual processing occurs even when observers claim to be unaware of the stimulus information (e.g., Miller 1939; Sidis 1898; Williams 1938).

STUDIES BASED ON SUBJECTIVE THRESHOLDS AND QUALITATIVE DIFFERENCES

Recently, we have proposed an approach to the study of subliminal perception which has the potential to resolve many of the long-standing controversies that have continually plagued this area of research. The proposed approach is similar in some ways to the approach underlying the studies involving an assessment of the effectiveness of masked primes. However, there are two critically important differences. First, the boundary between conscious and unconscious perceptual processes is equated explicitly with a subjectively-defined threshold based on claimed awareness rather than an objectively-defined threshold based on chance level verbal report accuracy. Second, subliminal perception is only assumed to be demonstrated when it can also be shown that consciously and unconsciously perceived information lead to qualitatively different behavioral consequences.

The proposed approach shares characteristics with several previous approaches. Obviously, by defining the boundary between conscious and unconscious perceptual processes in terms of subjective criteria, the proposed approach shares certain important similarities with all earlier studies of subliminal perception in which awareness was measured in terms of observers' subjective confidence that useful stimulus information was perceived. However, the previously mentioned criticisms of approaches based on subjective measures of awareness are avoided in the proposed approach by the additional requirement that any demonstration of subliminal perception also requires ewidence indicating qualitative differences in the effects of consciously and unconsciously perceived information. Although the importance of demonstrating qualitative differences between the effects of consciously and unconsciously perceived information has been emphasized previously (e.g., Dixon 1971, 1981; Shevrin and Dickman 1980), this criterion has had surprisingly little empirical impact. In contrast, demonstrations of qualitative differences play a central role in the proposed approach, as they provide the critical ewidence for distinguishing between conscious and unconscious processes. The reason that demonstrations of qualitative differences are so important is that they provide much stronger support for any distinction between conscious and unconscious processes than can ever be provided solely by demonstrating that stimulus information is perceived both above and below a particular threshold.

The proposed approach can be illustrated by a recent study we have conducted involving another variant of the Stroop color-word interference task. In this study, the target colors were either red or green and the color words used as primes were also either "red" or "green". In this two-color variant of the Stroop task, observers can name a target color faster when it is preceded by the congruent color word than when it is preceded by the incongruent color word, as long as congruent and incongruent trials are equally probable (i.e., 50/50). However, if the probabilities are changed 80 that incongruent prime-target pairings occur on 80% of the trials and congruent pairings occur on 20% of the trials (i.e., 80/20), then observers name a color faster when it is preceded by the incongruent color word than when it is preceded by the congruent color word (cf., Logan, Zbrodoff and Williamson, 1984). The different patterns of results obtained under these two probability conditions are illustrated in the left and center panels of Figure 3. A comparison of the panels shows that it is possible to completely reverse the relative effects of congruent and incongruent primes simply by varying the proportion of congruent and incongruent prime-target pairings.

FIGURE 3

STROOP EFFECTS WITH DIFFERENT PROPORTIONS OF CONGRUENT AND INCONGRUENT PRIME-TARGET PAIRINGS

Probability effects, such as the one described above, are generally assumed to indicate that observers voluntarily adopt a strategy on any particular task to maximize performance. Thus, it is reasonable to assume that the observers in the above task adopted a strategy in the 80/20 probability condition which utilized the predictive information provided by the primes. Since only two colors were used and incongruent prime-target pairings occurred on 80% of the trials, the best prediction that could be made concerning the target on the basis of the identity of the prime was that the target would be the color not named by the prime. Such a strategy would have allowed observers to anticipate the target correctly on 80% of the trials and would have only led to incorrect anticipations on 20% of the trials. This strategy provides a straightforward account of the faster decision times observed when incongruent rather than congruent prime-target pairings were presented, as it implies that performance should be facilitated on the large proportion of trials involving incongruent prime-target pairings and disrupted on the small proportion of trials involving congruent pairings.

Given that different predictive strategies can be induced in a Stroop task by variations in the proportions of congruent and incongruent trials, it was possible to use this task to establish that conscious and unconscious perceptual processes lead to qualitatively different behavioral consequences. If it is assumed that one function of consciousness is the selection of appropriate actions based on the analysis of perceptual information, then a reasonable expectation is that predictive strategies based on the identity of a prime may only occur when observers are consciously aware that the prime has been presented. Conversely, when observers have no awareness of a prime having been presented, it should be impossible for them to initiate any predictive strategy that would allow them to anticipate the color of the target. A demonstration that predictive strategies are only initiated when observers are consciously aware of the primes would indicate that conscious and unconscious perceptual processes are qualitatively different and would therefore support a definition of awareness based on observers' subjective reports.

To test whether the initiation of predictive strategies is limited to conditions in which observers claim to be consciously aware of the primes, the variant of the Stroop task described above was used in conjunction with a visual masking stimulus. In this experiment, the temporal interval between a prime and the mask was either so short that the observers claimed they were unable to discriminate one prime from the other or sufficiently long to allow the observers to identify which prime had been presented. The interesting pattern of results that was obtained is illustrated by comparing the middle and right panels of Figure 3. As in the preceding study, decision times were faster on incongruent than congruent trials when the primes were clearly visible. However, as indicated by the data presented in the right panel of the Figure, when the primes were masked so that the observers claimed not to be aware of their identity, the pattern of results was similar to that found when congruent and incongruent trials were equally probable. Thus, these data indicate that observers cannot initiate predictive strategies based on the identification of the primes when they claim not to be able to discriminate between the primes. However, these data also demonstrate that the primes were perceived even when the observers claimed not to be aware of them, since a typical Stroop effect was observed under these conditions.

These results clearly demonstrate a qualitative difference in the relative effects of congruent and incongruent primes which is dependent upon observers' claims concerning their perceptual awareness. These findings thus support the assumption that the boundary between conscious and unconscious perceptual processes can be defined in terms of observers' statements concerning their awareness of perceptual information. Furthermore, if this assumption is accepted, then these results provide definitive support for subliminal perception. It is important to emphasize, however, that the observers' verbal guesses, if they had been forced to guess in this experiment when they claimed no awareness, would have been considerably better than a chance level of accuracy. Thus, the conditions under which subliminal perception seems to occur are those in which perceptual analysis leads to phenomenal experiences that observers do not believe provide sufficient information to guide their behavior (cf.. Marcel 1983b).

CONCLUSIONS

This review of the different approaches followed in studies of subliminal perception indicates that evidence favoring subliminal perception is often found when perceptual awareness is defined in terms of subjective criteria. In the very earliest studies of subliminal perception, subjective criteria were used, and it was established that the observers could respond consistently at a better than chance level of accuracy even when they thought that their responses were only guesses (e.g., Miller 1939; Sidis 1898; Williams 1938). Likewise, our studies involving masked primes have consistently shown that primes are effective even when observers claim that they are not perceiving sufficient information to identify which prime may have been presented (Cheesman and Merikle, 1984, 1985, in press). Thus, if perceptual awareness is measured in terms of subjective criteria, then subliminal perception is a valid, easily-demonstrated phenomenon.

On the other hand, some critics of subliminal perception believe that the only appropriate way to define perceptual awareness is in terms of an observer's actual ability to give correct verbal reports indicating detection or identification of stimuli (e.g., Holender 1986). Thus, if observers are forced to guess even when they are certain they are unable to identify or detect a stimulus, and these guesses are somewhat more accurate than a chance level of performance, then these observers, by definition, are perceptually aware of the stimuli. To date, no evidence favoring subliminal perception has been found when this definition of awareness has been adopted and precautions have been taken to ensure accurate measurement of detection or identification thresholds (cf., Cheesman and Merikle 1985; Merikle 1982).

Previously, we have proposed that this definitional debate can be resolved by demonstrating that consciously and unconsciously perceived information lead to qualitatively different behavioral consequences (e.g., Cheesman and Merikle 1985; Merikle and Cheesman 1986). In fact, it is unlikely that any agreement concerning the proper definition of awareness will ever be reached unless the qualitative differences that distinguish conscious from unconscious perceptual processes are demonstrated. Furthermore, a real advantage of directing research toward establishing qualitative differences in perceptual processing is that the important research question is changed from "Does unconscious perception occur?" to "What are the differences that distinguish conscious and unconscious perceptual process?" The qualitative differences in perceptual processing described above, as well as other reported qualitative differences (e.g., Groeger 1984; Marcel 1980), provide optimism that future studies directed at establishing qualitative differences in perceptual processing will allow further specification of the critical differences that distinguish conscious from unconscious perceptual processes.

REFERENCES

Adams, Joe K. (1957), "Laboratory Studies of Behavior Without Awareness," Psychological Bulletin, 54, 383-405.

Balota, David A. (1983), "Automatic Semantic Activation and Episodic Memory Encoding," Journal of Verbal Learning and Verbal Behavior, 22, 88-104.

Cheesman, Jim and Philip M. Merikle (1984), "Priming With and Without Awareness," Perception & Psychophysics, 36, 387-395.

Cheesman, Jim and Philip M. Merikle (1985), "Word Recognition and Consciousness," in Reading Research: Advances in Theory and Practice, Vol. 5, eds. D. Besner, T. Gary Waller, and G. E. MacKinnon, Orlando, FL: Academic Press, 311-352.

Cheesman, Jim and Philip M. Merikle (in press), "Distinguishing Conscious from Unconscious Perceptual Processes," Canadian Journal of Psychology.

Dalrymple-Alford, E. C. (1972), "Associative Facilitation and Interference in the Stroop Color-Word Task," Perception & Psychophysics, 11, 274-276.

Dixon, Norman F. (1971), Subliminal Perception: The Nature of a Controversy, new York: McGraw-Hill.

Dixon, Norman F. (1981), Preconscious Processing, New York: Wiley.

Dyer, Frederick N. and Lawrence J. Severance (1973), "Stroop Interference With Successive Presentations of Separate Incongruent Words and Colors," Journal of Experimental Psychology, 98, 438-439.

Eriksen, Charles W. (1960), "Discrimination and Learning Without Awareness: A Methodological Survey and Evaluation," Psychological Review, 67, 279-300.

Fowler, Carol A., George Wolford, Ronald Slade and Louis Tassinary (1981), "Lexical Access With and Without Awareness," Journal of Experimental Psychology: General, 110, 341-362.

Groeger, J. A. (1984), "Evidence for Unconscious Perception in a Forced-Error Situation," British Journal of Psychology, 75, 305-314.

Henley, S. H. A. (1984), "Unconscious Perception Re-revisited: A Comment on Merikle's (1982) Paper," Bulletin of the Psychonomic Society, 22, 121-124.

Holender, Daniel (1986), "Semantic Activation Without Conscious Identification," Behavioral and Brain Sciences, 9, 1-23.

Lazarus, Richard S. and Robert A. McCleary (1951), "Autonomic Discrimination Without Awareness: A Study of Subception," Psychological Review, 58, 113-122.

Logan, Gordon D., N. Jane Zbrodoff, and James Williamson (1984), "Strategies in the Color-Word Stroop Task," Bulletin of the Psychonomic Society, 22, 135-138.

Marcel, Anthony J. (1974), "Perception With and Without Awareness," paper presented at the July meeting of the Experimental Psychology Society, Stirling, Scotland.

Marcel, Anthony J. (1980), "Conscious and Preconscious Recognition of Polysemous Words: Locating the Selective Effects of Prior Verbal Context," in Attention and Performance VIII, ed. Raymond S. Nickerson, Hillsdale, NJ: Erlbaum, 435-457.

Marcel, Anthony J. (1983a), "Conscious and Unconscious Perception: Experiments on Visual Masking and Word Recognition," Cognitive Psychology, 15, 197-237.

Marcel, Anthony J. (1983b), "Conscious and Unconscious Perception: An Approach to the Relations Between Phenomenal Experience and Perceptual Processes," Cognitive Psychology, 15, 238-300.

McCauley, Charley, C. M. Parmelee, Richard D. Sperber, and Thomas H. Carr (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, Philip M. (1982), "Unconscious Perception Revisited," Perception & Psychophysics, 31, 298-301.

Merikle, Philip M. (1984), "Toward a Definition of Awareness," Bulletin of the Psychonomic Society, 22, 449-450.

Merikle, Philip M. and Jim Cheesman (1986), "Consciousness is a 'Subjective' State," Behavioral and Brain Sciences, 9, 42-43.

Meyer, David E., Roger Schvaneveldt, and Margaret Ruddy (1975), "Loci of Contextual Effects on Visual Word Recognition," in Attention and Performance V, eds. P. H. A. Rabbitt and S. Dornic. New York: Academic Press, 98-118.

Miller, James G. (1939), "Discrimination Without Awareness," American Journal of Psychology, 52, 562-578.

Shevrin, Howard and Scott Dickman (1980), "The Psychological Unconscious: A Necessary Assumption for All Psychological Theory?" American Psychologist, 35, 421-434.

Sidis, Boris (1898), The Psychology of Suggestion, new York: Appleton.

Stroh, Marie, A. Margaret Shaw, and M. F. Washburn (1908), "A Study of Guessing," American Journal of Psychology, 19, 263-265.

Williams, A. C. (1938), "Perception of Subliminal Visual Stimuli," Journal of Psychology, 6, 187-199.

----------------------------------------