Subliminal Implants in Advertisements: an Experiment

ABSTRACT - In this paper a discussion of the controversies surrounding subliminal perception is presented. In response to the more recent controversy surrounding the use of sexual implants in the pictures of advertisements, an experiment was devised to gain information to address this controversy. In addition to manipulating the sexual implants within the pictures of advertisements, the copy also varied. The hypotheses were (1) that sexual implants would be associated with more positive attitudes and intentions and (2) that sexual implants would be more likely to favorably affect attitudes and purchase intentions when used in conjunction with sexually suggestive copy.


John G. Caccavale, Thomas C. Wanty III, and Julie A. Edell (1982) ,"Subliminal Implants in Advertisements: an Experiment", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 418-423.

Advances in Consumer Research Volume 9, 1982      Pages 418-423


John G. Caccavale (student), Duke University

Thomas C. Wanty III (student), Duke University

Julie A. Edell, Duke University


In this paper a discussion of the controversies surrounding subliminal perception is presented. In response to the more recent controversy surrounding the use of sexual implants in the pictures of advertisements, an experiment was devised to gain information to address this controversy. In addition to manipulating the sexual implants within the pictures of advertisements, the copy also varied. The hypotheses were (1) that sexual implants would be associated with more positive attitudes and intentions and (2) that sexual implants would be more likely to favorably affect attitudes and purchase intentions when used in conjunction with sexually suggestive copy.

The hypotheses were tested using analysis of variance (ANOVA). The results showed no significant positive effects of the interaction of sexual implants and sexually suggestive copy, thus allowing for the dismissal of hypothesis two. The results were ambiguous for hypothesis one. While there is clearly not enough evidence to accept hypothesis one, further research is needed before it can be soundly rejected.


Whether or not people can be influenced to behave in a particular way without their awareness has been a subject of controversy for decades. This question became the subject of much discussion when James Vicary flashed "Drink Coca-Cola" and "Eat Popcorn" on a movie screen for l/3000 of a second every five seconds and reported a 57.7 percent increase in popcorn sales and an 18.1 percent increase in Coke sales (Bachrach, 1959; Brean, 1957). Vicary's findings have been discounted by researchers due to the methodological errors and lack of control. The Federal Communications Commission attempted to replicate Vicary's findings but could not (Subliminal Ad, 1957). However, research by perceptual psychologists has had mixed results (Berelson and Steiner, 1964; Dixon, 1971; Egelhof, 1979; Silverman, 1976). An excellent review of the early work in subliminal perception is provided by McConnell, Cutler, and McNeil (1958). A study by Hawkins (1970) attempted to replicate aspects of Vicary's original work but with much tighter experimental controls. Hawkins (1970) found that subliminal messages can affect drive level but did not exert a significant influence on brand preference.

More recently Zajonc (1980) has reviewed a number of studies which indicated that affective reactions to stimuli can occur even in the absence of recognition memory. He asserts that one can like something or be afraid of it, without actually knowing what it is. Zajonc proposes that affect may be processed via an entirely different, but interrelated system, from that used to assign meaning and process information.

There are a wide variety of ways of producing subliminal stimuli. They may be recognizable less than 50 percent of the time because of their lack of intensity, duration, size or clarity. Most of the work that has been done by perceptual psychologists in the subliminal area has been subliminal by nature of its limited exposure time.

A second controversy has arisen regarding subliminal perception. This controversy has been precipitated by Key (1972, 1976, 1980). In this context subliminal perception has been defined as, "sensory inputs into the human nervous system that circumvent or are repressed from conscious awareness" (Key, 1978, p. 18). Key contends that advertisers are embedding "messages" in advertisements which are too small or lack clarity to be consciously perceived. These "messages", however, are received by an individual's subconscious and stimulate an emotional response and an identification with the advertised brand. Key contends that the subconscious is particularly receptive to experiences that conflict with long standing cultural taboos and since our culture's taboos deal primarily with sex and death these are the elements of most subliminal implants. Key has spent many years analyzing advertisements' pictorial elements and has identified numerous advertisements containing sexual implants. He is convinced that it is a widespread and commonly occurring phenomenon. He concludes that since the phenomenon of using implants is so widespread, it must be an effective selling device. Otherwise why would profit minded businesses spend extra dollars having their photographic work retouched to include subliminal implants.

Zajonc's (1980) theory concerning separate affective and cognitive systems might explain how the sexual implants impact on brand preference and purchase intentions. Upon exposure to an advertisement containing sexual implants the viewer may have positive feelings aroused in response to these implants without ever being aware they exist. This positive affect may then be transferred to the brand especially over many exposures. This theory does not add any insight into how the use of death elements as subliminal elements might work, unless they too are evaluated positively.

Past Research

Key (1972) reports a study where over 1,000 adults, both male and female, were shown a gin advertisement which contained sexual implants. The subjects were asked to describe their feelings, in writing, as they looked at the ad. Key reported 62 percent of the subjects reported feelings of sexual stimulation or excitement. He fails to make explicit his technique of content analysis or the objectivity of the judge doing the analysis. He gave no basis on which to compare these results, so the reader has no way of knowing if this is more or less than would be occurred had the implants not been present.

Bagley and Dunlap (1980) set out to correct these methodological problems in Key's research. They clipped two copies of a colored advertisement from magazines for four product categories. They assumed these ads were free from sexual implants. They showed the unaltered advertisement to a control group of subjects. The exact nature of the implants added to the other copy of the advertisement was not described. Bagley and Dunlap asked subjects, as Key (1972) had, to express their "personal feelings as to how the advertisement effects you" (p . 297) . Three naive judges rated the comments by searching for key words indicating sexual excitement. The results indicated a significantly higher percentage of subjects with responses indicating sexual stimulation when the ads contained the implants for 2 of the 4 ads and for the study overall.

The Bagley and Dunlap study is a significant improvement over Key's attempt to establish the effects of sexual implants It does not, however, attempt to relate subjects' feelings of "warmth" or "romanticism" to brand attitude, purchase intentions, or brand recall. Additionally, by treating stimulation as an all or none variable they fail to be able to analyze the degree of effect the ads had on subjects.

Kelly and Kessler (1978) conducted an experiment to examine the role of sexual implants on brand and illustration recall. They concluded that sexual implants failed to enhance recall of the brand or illustration.


The first hypothesis of this research is that sexual implants in the brand's advertisements increases the viewer's attitudes and purchase intentions. This is the hypothesis put forth by Key (1972, 1976, 1980) but as yet untested in any soundly methodological manner. There are a number of variables that may impact on the likelihood of the implant affecting the purchase intentions. These included interest in the product class, the number of exposures to the advertisement and the type of copy contained in the advertisement. Only the type of copy contained in the advertisement was included in this experiment. Two types of copy were included in the experiment - copy having sexual connotations and much plainer, more descriptive copy. The sexual copy was based on theories presented by Key (1978, 1980) concerning verbal pronunciation and letter arrangement. He contends that it is not only "taboo four-letter words which are effective in manipulating the response of mass audiences, but other words, with taboo implications have also been shown to possess subliminal power. Words such as 'shot', 'whose', 'cult', 'pints', ant 'taste', which differ by only one or two letters from certain taboo or emotional words, can also evoke strong demonstratable emotional reactions." (Key, 1972, p. 28).

Key also contends that the sexual suggestiveness of the consciously perceived elements of an advertisement enhance the effect of the subliminal elements. Key used the slogan, "Does She or Doesn't She," (Key, 1976, p. 57) as an example of explicit content which enhances the subliminal elements.

The second hypothesis to be tested in this research is that the interaction of sexually suggestive copy with the sexual implants is significant and positively related to purchase intentions.



The basic design of the study was a 2 x 2 after only design. The manipulations were: a picture containing a sexual implant or the same picture without the implant and copy containing sexual innuendoes or plain copy. This design was replicated for 5 product categories. Each subject was exposed to five advertisements - one from each product category. These advertisements were arranged so that each subject saw one advertisement from each of the 4 picture/copy conditions and saw 2 advertisements from one cell. The cell with 2 advertisements drawn from it, was varied so that each picture/copy combination was selected once. The specific advertisements seen by each group of subjects are given in Figure 1.

Because of the different numbers of subjects in different groups and the differences in the copy and implants used for each product category, a decision was made to analyze the data by product category. Thus, for each product category there are 106 observations. All factors are between subjects within each product.




The subjects were 106 MBA students who took part in the experiment during a regularly scheduled class on the same day.

Advertisements and Products

To determine the effects of sexual implants on consumers' purchasing intentions, the study required two sets of advertisements: Those containing the implants and those free of implants. Originally, the intention was to present existing advertisements with implants and without implants to the subjects in much the same way that Bagley and Dunlap (1980) had done. A portfolio of print advertisements was collected. After attempting to divide them into those with implants and those free of implants, two major problems were encountered. First, it became obvious that we could not eliminate brand familiarity and preference from interfering with the hypothesized effects. Additionally, it was found that no matter how intensively each apparently implant-free advertisement was examined, one could never be certain that it was truly free of subliminal implants. Consequently, we decided that it was impossible to test our hypothesis using current print advertisements.

The only recourse was to produce a set of advertisements containing sexual implants and an identical set without the implants. Using this technique, the effects of brand familiarity and preference, the differences in the physical construction of the advertisement, and the problem of a suitable control were eliminated. The advertisements were constructed to include both subliminal elements found by Key (1972) and those that had been observed in the portfolio of advertisements. Physical problems with ice and liquids necessitated the substitution of snack foods for the construction of our advertisements.

Five types of snack food were chosen: pretzels, cheese curls, cheese and crackers, popcorn, and apples. The following sexual elements were included in the picture of each of the advertisements.

pretzel - SEX scratched in the site of one pretzel

cheese curls - FUCK constructed out of cheese curls themselves

cheese - SEX written on two crackers and a phallic symbol etched on the cheese

popcorn - SEX written on two kernels

apples - SEX written on one apple

Photographs were taken of the actual food items with and without the sexual implant. Careful attention was paid to eliminate any other difference from occurring between the sexual version and the plain version of the pictures. The distance of the camera from the stimulus, the lighting and camera settings were held constant.

The 20 advertisements were created by combining the 2 copy conditions with the 2 picture conditions for each of the 5 products. Hypothetical names were used throughout. The copy was constructed using Press-types to insure no deviation in the lettering of the advertisements. Color photographs were used for the pictures. Finally, color slides were taken of the twenty print advertisements to facilitate showing them to groups. Again, strict control of the camera's distance from the advertisements, lighting and camera settings was maintained.


Subjects performed the task in groups. All subjects were told that they would be seeing mock-ups of advertisements for snack foods. The advertisements were presented in the same order to all groups. Each advertisement was on the screen for 30 seconds followed by a 5 second interval. After exposure to the 5 advertisements, each subject was given a questionnaire to complete. The questionnaire contained measures of the subject's evaluation of the picture, the copy, the total advertisement, and the likelihood of purchase.

Dependent Variables

Attitude toward the picture, the copy and the total advertisement were measured using the average of four, five-point scales (l - very interesting to 5 - not very interesting, l - good to 5 - bad, 1 - not irritating to 5 - very irritating, and 1 - like to 5 - dislike). The internal consistency of these scales was .89, .87 and .84 respectively. Purchase intention was measured on a ten-point scale anchored by "not likely at all to buy" to "very likely to buy


Manipulation Checks

In order to assess the appropriateness of the pictures for the purpose of the study, the overall quality of the pictures needed to be good while the implants needed to be subtle enough to eliminate readily conscious observation. Twenty pretest subjects were asked to examine the five pairs of pictures very carefully and to indicate any differences between the two forms. No subject detected the sexual implants. While this results does not indicate that the implants were actually subliminal it does add credibility. Subjects were then shown the sexual implants. After the exposure and an elapsed time of 10 minutes subjects were again shown the pictures. They could immediately point out the sexual implant. This led us to the conclusion that the implants were observable.

This manipulation check is not particularly impressive. However, any test of differences between the two sets of advertisements is basically a test of the effect of the implant. The manipulation check was designed firstly to see if the implants were subliminal - not consciously detectable even under close examination. Secondly, it was designed to see if the implants were observable when the subject knew where to look. The pictures passed on both accounts.

The copy was also varied, being either sexually suggestive or plain. The exact copy used for each ad is given in Appendix A. A second pretest was also conducted on two groups of ten subjects each. In this pretest subjects were shown one version of the copy portion for each of the five advertisements. Each group saw copy from the sexual copy condition and from the plain copy condition but for different products. There were asked to rate the copy on four 7-point scales. These scales were: (1) informative, (2) sexually suggestive, (3) easy to understand, and (4) believable. They were anchored by 1 - not at all and 7 very. The mean for the five plain copies together were significantly lower (2.94 vs. 4.08) in sexual suggestiveness than were those of the sexual copies (t = 5.96, p < .001). Additionally, the difference between the means for the sexual copy and the plain copy was significant for each of the S products analyzed separately. There were no statistically significant differences (p < .05) for the other three scales.

Test of the Hypotheses

The two hypotheses that this research was designed to address are (1) that sexual implants in a brand's advertisement significantly increase the intentions to purchase that brand and (2) that the interaction of the implants with sexually suggestive copy is positively related to attitudes and purchase intentions.

To test these hypotheses ANOVA was performed on the data for each product category for the three attitude and the intentions measures. These results are displayed in Tables 1-4.

Table 1 displays the ANOVA results for the attitude about the picture in the advertisement. There are no systematic effects across product categories. The interaction of copy and picture for pretzels was significant but the means indicate the most favorable attitude occurs when the copy is not sexually suggestive and the picture contains no sexual implants. The main effect of picture for cheese is highly significant. This effect may be an anomaly or it may be related to the fact that cheese was the only product category to contain a phallic symbol as well as the word SEX. The mean for the cell with the subliminal implant was 2.93 vs. a mean of 3.50 for the cell without the implant. This is on a scale of l to 5 with l representing the most favorable attitude.

The ANOVA table for the dependent variable, attitude toward the copy of the advertisement, is given in Table 2. The only significant effect is the effect of copy for the cheese product. The means indicate that plain copy is more favorably evaluated (2.84 vs. 3.31). Since there is no hypothesis regarding a main effect of copy, and this effect is an isolated one, it is best to consider this effect as the result of chance.

The ANOVA table for the overall attitude toward the advertisement is Table 3. Again only one effect is significant and for only one product. The interaction effect of copy and picture is significant for the pretzel category. The means indicate that the most favorable attitude occurred when the copy was without sexual innuendos and the picture was without sexual implants. This effect is the opposite of that hypothesized. Additionally, the attitude toward the advertisement as a whole should have some relationship to the attitude toward its component parts. Thus one would expect that if an effect were reliable, and it had a significant effect on the attitude toward the picture or the copy, that it would also have an effect on the attitude toward the entire ad. None of the effects noted earlier show this trend.

Table 4 gives the analysis for the purchase intentions variable. Again no systematic effects are evident. The main effect of copy is significant for the popcorn category. The purchase intentions are lower (4.28) when the copy is sexually suggestive than when it is plain (5.16). The main effect of picture is significant for the apples product category. The means indicate that the subjects exposed to the subliminal implants had greater purchase intentions (5.48) than those in the no implant condition (4.41). The interaction of picture and copy was significant for cheese curls and popcorn. The highest purchase intentions was obtained when the copy was sexually suggestive, but the picture was without sexual implant for the cheese curls. For the apples, the cell with the most favorable mean was the one with plain copy and the sexual implant.



It seems reasonable to assume if the effect of subliminal implants are of any managerial importance it should explain at least 5 percent of the variance in the purchase intentions data. In order to explain 5 percent of the variance in this data, the difference in the means must exceed l scale point. Therefore, in calculating the power of the test, the difference between the means that is considered necessary is 1 scale point. With 106 subjects, and a = .05 the power of the one-tail t-test is .88.


The results presented above allow us to reject hypothesis two without question. In none of the product classes and for none of the dependent measures did we find any evidence to support the hypothesis that sexual implants and sexually suggestive copy work together is a way to increase attitudes or purchase intentions.

The evidence on hypothesis one is less conclusive. There were 2 cases where the subliminal implants had a significant effect in the hypothesized direction. Table 5 shows that of the twenty (4 measures times 5 product categories) dependent measures analyzed, there were 11 cases where the subliminal implant was associated with less favorable attitudes or intentions, and 9 cases where the implant was associated with more favorable attitudes or intentions.







While this evidence does not, in any way, allow for the acceptance of hypothesis one, it does raise some interesting questions.

The cheese product is the only one where a positive systematic effect is seen on all variables. As noted earlier, cheese was the only product to contain a phallic symbol, as well as, the word SEX. This suggests that the nature of the implant may be an important variable in determining the effect of the implant.

Suggestions for future research in this area would include the following. The type of implant used should be carefully varied so that one can determine to what the resultant effects are do. Secondly, before this question can be resolved the exposure conditions need to be more natural. Subjects should view the advertisements as they normally would. This should include having the advertisements embedded in various editorial contents and exposure in a relaxed, non-laboratory environment. A third variable which needs examination is the number of exposures required.

The purpose of this study was to provide an empirical test of a popular hypothesis-that subliminal implants increase sales. The result may be that more questions have been raised than have been resolved.




Cheese Curls

Sexual Copy

Come taste new Bedda-Chedda. A sure ire party sensation. It's cheesy, cheddar taste can stand up to the most discriminate connoisseur. So long, boring party treats' Welcome to the cheesy, cheddar taste of Bedda-Chedda!

Plain Copy

Bedda-Chedda. A revolutionary idea in party snacks. We use real Wisconsin cheddar cheese with absolutely no artificial ingredients. Tired of boring cheese snacks? Try new Bedda-Chedda. Bedda Gedda some Bedda-Chedda.


Sexual Copy

Big Barn brings you a delicious new, country cheddar taste in cheese. We use only the essential, natural ingredients to produce a cheese that is guaranteed to excite your taste buds. The creamy taste of Big Barn cheddar is a surefire hit at parties, picnics or any type of affair.

Plain Copy

Brentwood Farms brings you a deliciously different Cheddar Cheese. Our cheese contains no artificial ingredients. We use the finest, natural dairy products to produce a cheese that is fast becoming America's favorite' Brentwood Farms cheddar cheese. . . the way it was meant to be.


Sexual Copy

New' Mel's Zels. Made from Mel's secret old-fashioned country recipe that has kept people begging for more. Mel's Zels whipped Mr. Salty, Rold Gold, and Wise in a nationwide taste test. Try 'em' They're a taste too good to miss' "I give you my personal guarantee that the salty, crunchy, taste of Mel's Zels will stand up to any comparison," Mel. Mel's Zels. The perfect pretzel, brought to you from a long standing family tradition.

Plain Copy

New' Mel's Zels. A revolutionary new idea in pretzels. Mel's secret recipe has brought a delectably delicious change to the modern pretzel. "I give you my personal, money-back guarantee that these will be the best pretzels you have ever eaten, "Mel. Mel's Zels. The perfect Pretzel


Sexual Copy

Apples from Ohio. Ridiculous you say? But indeed, the most succulent, juicy, mouth-watering apples come from Ohio You've probably been eating Ohio apples for years, they're the ones with the long stems. You remember Grandma's delicious pie? She's been making it with Ohio apples for years. Ohio Apples. When it comes to apples, they come to us.

Plain Copy

Apples from Ohio? You'd never believe it, but they're the best' After one bite, you'll be convinced. Ohio Apples, ask for them by name, and remember . . . Grandma never made a great apple pie . . . unless she used Ohio Apples.


Sexual Copy

Pop's Popcorn. At long last, we have uncovered Pop Smith's secret country recipe for perfect popcorn. The secret's on the inside. Each delectable kernel already contains just the right amount of butter-salt. So when you taste the first morsel, you'll always come back for more! Pop's Popcorn. It comes with everything you need. For parties, get-togethers, or just munching on your own. It tastes exceptional!

Plain Copy

Pop's Popcorn. At long last Pop Smith has revealed his secret for perfect popcorn. Using a legendary farming technique, Pop brings you a kernel that contains its own butter-salt. The days of greasy butter and messy salt are gone. When that popcorn pops, it's ready to eat. Pop's Popcorn. The secret is in the kernel.


Bachrach, J. J. (March, 1959), "The Ethics of Tachistocopy, Bulletin of Atomic Scientists, 15, 212-215.

Bagley, Gerrold S. and Dunlap, B. J. (1980), "Subliminally Embedded Ads: A 'Turn On'?' Southern Marketing Proceedings, 296-298.

Berelson, Bernard and Steiner, Gary A. (1964), Human Behavior: An Inventors of Scientific Findings (New York: Harcourt, Brace & World).

Berkman, Harold W. and Gilson, Christopher C. (1978), Consumer Behavior: Concepts and Strategies (Encino, CA: Dickenson Publishing).

Bostock, Roy (March 18, 1981), Personal Communication.

Brean, R. (April 27, 1958), "What Hidden Sell is All About Life, 29, 194-214.

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

Egelhof, Joseph (June 23, 1979), "Is Message Faster than Eye or Ear?: Subliminal ImpLants' Furor Growing Again," Chicago Tribune, E-27.

Hawkins, Del (August, 1970), "The Effects of Subliminal Stimulation on Drive Level and Brand Preference," Journal of Marketing Research, 7, 322-326.

Kelly, J. Steven and Kessler, Barbara M. (1978), "Subliminal Seduction: Fact or Fantasy?" Southern Marketing Proceedings.

Key, Wilson B. (1972), Subliminal Seduction: Ad Media's Manipulation of a Not So Innocent America (Englewood Cliffs: Prentice-Hall).

Key, Wilson B. (1976), Media Sexploitation (Englewood Cliffs: Prentice-Hall).

Key, Wilson B. (1980), The Clam-Plate Orgy: And Other Subliminal Techniques for Manipulating Your Behavior (Englewood Cliffs: Prentice-Hall).

McConnell, James V., Cutler, Richard L. and McNeil, Elton B. (May, 1958) "Subliminal Stimulation: An Overview," American Psychologist, 13, 229-242.

Silverman, L. H. (September, 1976), "Psychoanalytic Theory: 'The Reports of My Death Are Greatly Exaggerated'," American Psychologist, 31, 621-637.

(April, 23,1957), "Subliminal Ad Okay If It Sells: FCC Peers into Subliminal Picture on TV," Advertising Age, 27-31.

Zajonc, R. B. (February, 1980), "Feeling and Thinking: Preferences Need No Inferences," American Psychologist, 35, 151-175.



John G. Caccavale, (student), Duke University (student), Duke University
Thomas C. Wanty III, Duke University
Julie A. Edell


NA - Advances in Consumer Research Volume 09 | 1982

Share Proceeding

Featured papers

See More


How Matte Product Surface Enhances Perceived Durability

Taehoon Park, University of South Carolina, USA
Junghan Kim, Singapore Management University, Singapore

Read More


Signaling Fun: Anticipated Sharing Leads to Hedonic Choice

Nicole Kim, University of Maryland, USA
Rebecca Ratner, University of Maryland, USA

Read More


Thanks for Nothing: Expressing Gratitude Invites Exploitation by Competitors

Kelly Kiyeon Lee, Georgetown University, USA
Jeremy A. Yip, Georgetown University, USA
Cindy Chan, University of Toronto, Canada
Alison Wood Brooks, Harvard Business School, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.