Subliminal perception refers to visual and auditory information presented at a speed and or intensity that is below the conscious threshold of perception through one or more channels and thus not readily apparent to the subject (Moore, 1982). Subliminal perception, like multi-image presentations, is also closely related to the theoretical bases of cue summation and multiple-channel research. All are interested in providing the learner with the maximum amount of usable cues, with the idea that these cues will support and reinforce each other. This is similar to multiple-channel theory, which suggests that additional simultaneous cues within and across sensory channels provides greater reinforcement in organizing and structuring information.

Experiments using subliminal exposure to visual and audio stimuli have been reported in psychological journals since 1863 (Application of Subliminal Perreption in Advertising, 1958). Reviews of experimentation in subliminal perception have contributed summaries of various points of view. Three excellent sources on the subject were published by Miller (1942), Adams (1957), and McConnell, Cutler, and McNeil (1958). All three sources indicate that research results have differed widely (DeChenne, 1975).

In reviewing three summaries of research on subliminal perception (Bevan, 1964; Dixon, 1971; McConnell, Cutler & NcNeil, 1958), several generalizations become apparent. Susceptibility to subliminal stimulation varies among people and is dependent on factors such as anxiety, attentiveness, and need state. Sensitivity to subliminal effects tends to be cumulative, since repeated viewing of subliminal materials tend to make a person more aware of the technique. Differences in awareness thresholds also determine whether subliminal messages are perceived. Perception thresholds can be lowered if the duration of the subliminal exposure increases or is of different brightness than the surrounding visual field. In other words, the closer to being consciously visible the material is, the more likely it is to be perceived (Moore, 1982).

Early experiments were designed to provide evidence that the psychological phenomenon of subliminal perception was a reality. One of the earliest of these experiments was reported by Hollingworth (1919). Others included experiments by Maker (1937), Coyne, King, Zubin, and Landis (1943), McGinnus (1949), Lazarus and McClearey (1951), and )Wilcot (1953). All except Wilcot reported results that there had been definite unconscious recognition or influence by stimuli below the conscious threshold. These studies gained attention for the concept of subliminal perception but brought about additional research that was often inconclusive and contradictory (Moore, 1982). More recent experiments have focused on determining relationships between subliminal perception and behavior. Studies of this type included those of Klein, Spence, Holt, and Gourevtich (1958), and Smith, Spence, and Klein (1959), all of which reported tendencies of a positive nature concerning the effectiveness of subliminal perception.

Several studies have been conducted to determine whether subliminal shapes or words could be detected when superimposed on a still or moving picture. One method of operationalizing subliminal stimulation is to superimpose a message at a very low relative brightness for a long period of time. This method was used by DeFleur and Petranoff (1959) in one of the first studies of subliminal perception using television as a carrier medium. The subliminal material in this experiment was superimposed as an extremely faint image, relative to the main program. Analysis of the results indicated that significantly more correct guesses had occurred than would have been expected by chance. It was not reported if the participants were asked whether they had consciously seen any of the shapes during the film. Nevertheless, the results seemed to indicate that TV images of extremely low brightness influenced their responses.

Moore (1982) commented on the procedures used in DeFleur and Petranoff's (1959) study. The low-intensity, constant-image technique that was used by DeFleur and Petranoff could result in the "subliminal" image being consciously visible. Because the visual field of the motion picture was dynamic (the images moved and changed), the faint subliminal words or shapes that were on the screen may have become partially unmasked at times as the foreground images changed. For example, if the constantly superimposed, subliminal images were white and the foreground images (the motion picture) in the same area of the screen were momentarily dark, then the resulting contrast differences may have been sufficient to unmask and reveal the subliminal word or shape or an identifiable segment of it. If the superimposed words or shapes were quickly flashed rather than constantly exposed, then the visual threshold of viewers would remain higher and the images would more likely remain subliminal (Moore, 1982).

Similar experiments have been reported by several other researchers. In these experiments, the subliminal shapes or words were nonmoving images on a neutral background, as compared to the moving foreground images used by DeFleur and Petranoff (1959). Schiff (1961) and King, Landis, and Zubin (1944) reported positive results, while Champion and Turner (1959) and Calvin and Dollerimayer (1959) concluded that there was no definitive evidence that behavior was altered by subliminal presentations. The relationship between subliminal stimulation and cognitive functions has been studied in a number of experiments. Kolars (1957) (two studies) and Gerard (1960) used a problem-solving task in which rows of geometric figures were simultaneously presented by a tachistoscope. Kolars concluded that the presentations of subliminal stimuli did influence the frequency of correct answers in both studies. Gerard tested participants' ability to reconstruct mentally a composite, geometric figure into alternative assemblies. One group saw the correct solution, another group saw an incorrect solution, and the control group saw no subliminal solution.

The results indicated that the control group did better than either of the subliminal treatment groups. However, the group shown the correct answer did better than the group shown the incorrect answers, as hypothesized. Gerard's results partially confirmed Kolers' findings, however, that subliminal presentations could affect performance on problem-solving tests (DeChenne, 1975; Moore, 1982; Moore & Moore, 1984).

The research described above (Calvin & Dollenmayer, 1959; DeFleur & Petranoff, 1959; Gerald, 1960; Kolers, 1957) indicates that subliminal perception can occur among certain people in laboratory settings. Research dealing with educational uses has been conducted by Murch (1965) and Sharp (1959) who demonstrated that the test-taking behavior of students can be subliminally influenced. DeChenne (1975), Skinner (1969), and Taris (1970) studied either teaching subject matter or teaching a skill entirely by subliminal means (DeChenne, 1975, Moore, 1982).

In contrast to Murch (1965) and Sharp (1959), who demonstrated that choice behavior could be altered in a test-taking situation, the experiments of DeChenne (1975), Skinner (1969), and Taris (1970) failed to demonstrate that direct teaching by subliminal perception can occur. Although various laboratory experiments have produced evidence that subliminal perception can occur, field experiments conducted to test direct teaching by subliminal perception have not yielded collaborative results.

Moore (1982) contends that when teaching by a subliminal means under conditions when the subject matter to be taught is transmitted with films that are unrelated and/or irrelevant to the subject matter, the possibility for content interference is great and the lack of conductive and focused learning setting would seem to hinder learning further. "Expecting subliminally produced learning to occur now seems less realistic than expecting a classroom teacher to teach while students are watching an Abbott and Costello comedy" (pp. 19, 20).

A number of studies investigated the possibilities that motivation might be influenced by subliminal perception. Among these were studies by Byrne (1959) and Goldstein and Davis (1961), whose results indicated no influence on the subjects. Goldstein and Barthal (1968) and Zuckerman (1960) conducted studies to determine whether subliminal stimulation could influence elaborative thinking. In both studies, positive and negative words were subliminally flashed with pictures from the Thematic-Apperception Test. Both studies reported contradictory results when participants were asked to create and elaborate on stories and the amount written as directed in the subliminal constructions. Shevrin and Luborsky (1958) and Johnson and Erikson (1961) reported similar results to support their theory that there was a tendency for tachiscopically presented material to appear in daydreams and dreams.

In addition to content reinforcement, Moore (1982) asks what effect individual cognitive style differences may have on learning from subliminal media treatments. Most early subliminal perception research limited consideration of individual participant differences to sex, race, and IQ. Other (undetected) differences in sample populations might explain why many replication attempts have failed to confirm original findings, and why many findings are contradictory. In a review of subliminal research, McConnell et al. (1958) stated that individual differences "must be taken into account by anyone who wishes to deal with individuals. It is quite likely that many differences in the perception of subliminal stimuli do exist between individuals of differing classes, ages, and sexes" (P. 236). Allison (1963), Murch (1965), and Sackeim, Packer, and Gur (1977) have shown that individual differences such as thought strategies, cognitive set, and hemisphericity were related to susceptibility to subliminal stimulation. DeChenne (1975) and Skinner (1969) did not collect data on individual differences in learning styles or abilities within their samples. By not doing so, detecting the effect of the treatment would have been more difficult if aptitude-treatment interaction effects were occurring, as the slight increase in treatment effectiveness in these two studies may have indicated. The term individual differences is also associated with the concept of cognitive styles.

Past studies questioned whether subliminal perception could be a useful tool for producers of educational television and explored the feasibility of teaching one topic while students were watching a program unrelated in content (DeChenne, 1975; Skinner, 1969; Thris, 1970). The results indicated that subliminal messages were generally not powerful enough to cause learning when students were concentrating on an unrelated topic. In other words, it is unrealistic for educational producers to expect that students could be taught two topics simultaneously, one through normal channels and the other through subliminal perception Moore, 1982; Moore & Moore, 1984). However, there was some evidence (DeChenne, 1975) that some students seeing subliminal cues performed better on a criterion task. This suggested that individual differences such as intelligence or perceptual abilities may be related to the ability to profit from subliminal messages implanted in a television program. This is generally consistent with Calvin and Dollerimayer (1959), Gerard (1960), Murch (1965), and Sharp (1959).

The properties of visual subliminal messages include being faintly and quickly embedded within a surrounding visual field. A student's ability to profit from subliminal messages could be related to the ability to dissembed the message from the surrounding television picture. Therefore, it was thought that the cognitive style of field dependence may have some relationship to the potential usefulness of subliminal perception. Since people have different ways of perceiving their environment, these differences may have been associated with the differences in subliminal learning seen in various studies (Calvin & Dollenmayer, 1959; DeChenne, 1975; Gerard, 1960; Kolers, 1957). Based on the literature, it also could be expected that field-independent individuals, because they have highly developed skills at dissembedding one object or image from a surrounding array of objects or images, should likewise be able to distinguish the embedded subliminal messages in a television picture (Greco & McClung, 1979; Hessler, 1972). The real benefit in learning, however, could occur for those students who are field dependent, since they typically benefit from more salient content organization cues (Witkin, Moore, Goodenough & Cox, 1977). Thus, the use of subliminal reinforcement cues (captions) could be of most value to field-dependent students, because the captions would supplant students' reduced ability to distinguish between relevant and nonrelevant cues and would make the relevant cues more salient.

In Moore's (1982) experiment, these differences in cognitive style were studied as a possible intervening factor for consideration in the production and utilization of subliminal materials. In the analysis of data, it was found that students having prior experience with the subject matter, such as in a previous course, averaged highest on the recall test, as one would expect. These students were eliminated from subsequent analysis, since their recall may have reflected prior knowledge or outside influence.

The available experiments and observations on subliminal perception seem to indicate that in certain instances human subjects are capable of responding to audio and visual stimuli that are so weak in duration, intensity, or clarity that they are not consciously aware of them. Researchers have varying opinions as to the effectiveness of subliminal stimulation, and there is no conclusive evidence as to its ineffectiveness or effectiveness. However, the body of evidence does indicate that, effective or not, there is perception below the threshold of awareness (DeChenne, 1975). There appears to be major concerns, however, involved in determining the amount of information a human can process at any one time. To recognize information simultaneously, the various receptors (eyes, ears) would have to analyze a great variety of different cues. All the findings noted in preceding sections, e.g., multiple-channel, multi-image, and subliminal perception, have import to the design of multimedia presentations. Basic decisions have to be made to determine how the presentation is to be developed, the number of cues to be available, and the number of channels to be used.