Perceptual and Motor Skills, 1975, 41, 847-854. @ Perceptual and Motor Skills 1975

EFFECT OF SUBLIMINAL STIMULI ON CONSUMER BEHAVIOR: NEGATIVE EVIDENCE STEPHEN G. GEORGE Malunani Research1

AND

LUTHER B. J E N N I N G S Occidenral College

Summary.-The study corrected methodological weaknesses found in previous experiments designed to test the contentions of motivational research theorists that subliminal stimulation can affect buying behavior. T h e words "Hershey's Chocolate" were presented to a group of 1 8 experimental Ss below a forced-choice detection threshold. The 19 control Ss had a blank slide superimposed over the same background media. In a highly controlled buying situation neither experimental nor control Ss purchased Hershey's products, but o n comparable chocolate products, the experimental Ss bought 5 and the control Ss, 3. A second smdy tested 15 experimental and 1 2 control Ss with the stimulus presented just below a recognition threshold. N o experimental Ss bought Hershey's; two control Ss did. N o support was found for the claims of motivational research theorists.

In 1956 a commercial enterprise claimed to have successfully used subliminal stimulation as an advertising technique in a New Jersey movie theater. They made claims of increased popcorn and Coca-Cola sales after the commands "Eat" and "Drink" the respective products were flashed on the screen at 1/3000 of a second. Although attempts to confirm the study came to naught ( 9 ) its implications in terms of unconscious involuntary control of behavior were so profound that it became one of the most celebrated and quoted studies ever reported in psychological literature and certainly one of the few, if not only one, to lead to a congressional investigation. Although it was never reported in scientific journals, only in the popular press ( 9 ) , its thrust typifies a principal variation of psychoanalytic theory promoted by many psychoanalytically oriented perceptual researchers. Spence and Holland ( 1 3 ) have explicated this point of view somewhat more authoritatively than that implied in the reported movie study. Like Vicary, who originated the story of the movie theater study, and other proponents of motivational research, Spence and Holland construe psychoanalytic theory as leading to the prediction that at some point below the detection threshold, subliminal stimulation arouses unconscious associational networks which are maximally free of conscious restrictions. At chis point the subliminal stimulus maximally influences the unconscious and behavior. Spence and Holland contend that below and above this point of maximum effect subliminal stimuli have less effects on the unconscious and at both the physiological and detection thresholds this effect is reduced to zero. 'Malunani Research, P. 0. Box 127, Makawao, Hawaii 96768.

848

S. G. GEORGE & L. B. JENNINGS

Despite the attractiveness of this general theory, several studies call it under serious question. Jennings and George ( 7 ) have criticized Spence and Holland ( 13) for not employing the accepted forced-choice technique for defining the detection threshold. Correcting this methodological weakness, the Jennings and George study yielded negative evidence. In addition, three studies (7, 8, 14) have demonstrated that serial position effects and interitem association strength favor the recall of the 10 cheese associates over the 10 neutral words used by Spence and Holland ( 13) as their dependent variable. While the above negative evidence appears to eliminate any hope of salvaging the theory that deeply subliminal stimuli maximally influence the unconscious, the possibility still remains that the dynamics implied by che theory only operate in the realm of the basic physiological drives emphasized by Vicary and other motivational research theorisis. That this possibility might exist seems supported by a study by Byrne ( 1 ) and a later study by Spence ( 11). Byrne reported finhng evidence that the subliminally presented word BEEF increased Ss' subjective hunger. Spence claimed that subjective hunger ratings increased after Ss had been exposed co 30 tachistoscopic presentations of the word CHEESE. However, the latter two studies fail to convincingly confirm the influence of subliminal stimulation. Byrne's study has been faulted by Spence ( 11) for using a threshold too close to consciousness to expect the maximum subliminal effect to occur. Spence (11 ) reduced the level of subliminal stimulation to a point which he considered to be below the forced-choice detection threshold and thus in the zone which he felt should maximize the subliminal effect. However, Jennings and George ( 7 ) have shown that, in fact, the level of scimulation Spence used was considerably closer to the recognition threshold than he had intended. When George and Jennings ( 6 ) corrected Spence's omission of a control group and presented stimuli at both the level Spence actually used and the level he had intended to use, neither yielded any support for his theory. Most studies which have attempted more direct tests of the contentions of motivational research theorists can be faulted for several reasons. One difficulty with this type of study inheres in the assessment of changes in buying behavior of Ss. Since little or no restriction has been placed on where S purchased the product, it has been difficult to obtain reliable pre-exposure to post-exposure changes in purchasing patterns of Ss. Also, few of the studies have used adequate control groups with which to compare changes in buying patterns. Moreover, in most studies the level of stimulation used cannot be defined since it involved superimposing tachistoscopically projected images on backgrounds containing subregions which varied immensely in illumination transmission. The resulting point-to-pint variations in figure-to-background contrasts in illumination rendered futile the assignment of a meaningful average index of figure-to-back-

SUBLIMINAL STIMULI IN CONSUMER BEHAVIOR

849

ground contrasts. But the latter is, of course, essential to determining the degree of subliminality of the stimulus. According to the contentions of the theorists mentioned earlier, a stimulus must be below the reach of consciousness to communicate exclusively, and maximally, with unconscious processes. By definition, such a stimulus must be subliminal to a forced-choice detection threshold. Although two studies (2, 3) found evidence contrary to theory, there is no assurance that their critical stimulus materials were impoverished to this extent, In light of the fact that previous studies in this area have had methodological weaknesses and/or have not stimulated Ss at a level which critically tests the contentions of theorists, another study is needed. The present study, which stimulated Ss below a forced-choice detection threshold in order to produce a change in consumer behavior, corrected the methodological weaknesses described above.

METHOD A slide projector, governed by a rheostat, superimposed the words

HERSHEY'S

CHOCOLATEpresented one above the other over a projected meaningful background.

By varying illumination with the rheostat, the appropriate stimulus quality below a forcedchoice detection threshold was established. T h e stimulus, which was 6 in. high and 15 in. long (each letter being 2 % in. high), was presented at 1/50 of a second in the center of a beaded screen which was located in the corner of a completely darkened room so that che apparatus did not obstruct Ss' vision. The research upon which the present study was based involved superimposing'the subliminal stimulus on meaningful material ( a movie). Consequently, in the present study the background stimuli upon which the subliminal words HERSHEY'S CHOCOLATE were superimposed was made meaningful. However, to correct for the, moment-to-moment variations in background quality resulting from using a moving picture and at the same time to employ meaningful stimuli which would continuously change so as to hold the interest of the perceiver, a series of still pictures (slides) were employed. I n order to control for variations in background material, slides were chosen which possessed approximately the same illumination and maximum homogeneity (textural uniformity) i n the central portion where the stimulus would be superimposed. However, as there were differences in slide illumination, the different levels at which the stimulus was to be presented were determined by the following method. I n order to establish the respective level of stimulus illumination for each of the slightly differing slides, E tested two pilot Ss who were precisely at the mean forced-choice detection threshold of 84 Ss tested in a previous experiment using the same apparatus. I n this previous study it was found that Ss were able to detect the flashing of the stimulus ground (background illumination) considerably below the point at which the detection of the figural properties of the stimulus had been reduced to chance accuracy. Hence, in the present study the scimulus was presented at a level at which the average S could not detect even a flash of light accompanying the subliminal presentation. The two pilot Ss were tested at the average S-to-screen distance to be used in the experiment proper. Using the method of serial exploration, E then determined these Ss' individual thresholds for "detecting any flash of light" when the critical stimulus was presented tachistoscopically over each individual slide. In this way the rheostat which governed the illumination of the stimulus slide could be altered according to the threshold averages of the two pilot

850

S. G.

GEORGE

&

L. B. J E N N I N G S

Ss for each of the slides, thereby ensuring that the stimulus quality was below the detection threshold for each stimulus presentation. Ss participated in the study in their respective sections of introductory sociology which occurred at successive hours. T h e experimental group, ranclomly selected, had an N of 18 and the control group had 19 Ss. The study proceeded as part of a regularly scheduled lecture in which 4 0 slides were used as visual aids. Since the apparatus used for superimposing the subliminal stimulus could not be hidden from Ss, its presence had to be justified. The lecturer referred to it before beginning, indicating that as long as he had to darken the room for slides, he would give a demonstration of some laboratory techniques. A student (assistant) appeared to be setting up and testing the apparatus during the lecture. Acrually, he was superimposing the stimulus over the 20 pretested slides. The assistant adjusted the rheostat according to the levels predetermined by the pilot Ss for each slide; the shutter (1/50 sec.) was triggered five times for each of the 20 slides. The lecturer was not aware of which class received the stimulus as opposed to the blank slide, and therefore could not bias his talk. After the lecture the student assistant helped in demonstrating the technique of presenting stimuli subliminally o n an otherwise blank screen. The stimulus in this demonstration were numbers. 438673. T w o weeks later the lecturer mentioned to both classes that "a student had thought that something had been presented along with the slides rwo weeks ago." H e asked the Ss i n each group how many of them thought something had been presented. Responses were recorded on blank paper as the questions were asked. The second question asked was, "Did you see a light flashing while the slides were being presented?" Third, "Did you feel that any of the flashes were intentional?" Fourth, "If you can describe the image that was presented either accidentally o r intentionally, 1 will give you four dollars." T o check the differences in the purchases of candy bars made by Ss in experimental and control groups, candy sales were monitored in the college student store. Since the student store was the only source of candy bars within a four-mile radius of the rural boarding campus and the students had no means of motor transportation nor public transit system for conveyance, the purchasing situation was highly controlled. Forms were devised under the guise of a survey by a local radio station to be used by the store clerks. T h e clerks filled out the names of the purchasers of the following candy bars: Hershey's Chocolate, Hershey's Almonds, Hershey's Kisses, Nestle's Chocolate, Nestle's Almonds, and Nestle's Crunch (both 54 and 104 sizes). These were the only chocolate bars of similar nature sold in the store. The survey was coincidental with stimulation and lasted 10 days. The radio station manager dealt with the store clerks, so they were not aware that E was connected in any way with the survey.

RESULTS AND DISCUSSION On the questions only one S, an experimental S, felt that something had been presented; he felt this presentation was an accident. He couldn't describe what was presented, but felt it was numbers, as that was the stimulus used in the demonstration afterwards. When asked if they had seen a light flashing, five Ss felt they had; two were from the experimental group, one of them being the previously mentioned S. None of the Ss felt the presentations had been intentional. When attempting to describe the stimulus, only three Ss reported anything remotely related to the stimulus. These two experimental Ss and a control S responded "numbers." As mentioned before, one of the experimental Ss

SUBLIMINAL STIMULI I N CONSUMER BEHAVIOR

85 1

qualified this, saying that he felt it was numbers because thac was the stimulus used in the demonstration afterwards. During the 10-day period in which the purchases of candy bars were made, not a single student in either the experimental or control group purchased any of the Hershey's products. Checks were made on other similar candy bars (Nestle's). The experimental group purchased three Nestle's Milk Chocolate bars and the control group two. When all of the different types of Nestle's candy bars are considered, the experimental group ~urchasedfive and the control group three. This difference is not significant ( p > .05, X< Yate's correction) . Obviously, the results do not support the contentions of motivational research theorists. Neither experimencal nor control Ss purchased any Hershey's products, let alone Hershey's Chocolate bars. One might conclude that too few Ss were tested in the present study and therefore not enough chocolate products could be purchased to analyze using conventional statistical tests. However, enough purchases were made of Nestle's products to use a chi-square test. In addition, when comparing the 5 purchases of Nestle's products with 0 purchases of Hershey's products made by the experimental Ss, a chi-square test using 2.5 purchases as the expected frequency yields a nearly significant difference (+ < .08, Yate's correction). Spence and Ehrenberg ( 1 2 ) found that, when Ss who had at least 2.6 hr. since last food were stimulated either supraliminally or subliminally by the word CHEESE,differential recall of cheese associates over neutral words was greater than that for Ss with 2.5 or less hours since last food. These authors interpret their findings to mean thac food or oral deprivation creates a drive state which is necessary to elicit the effects of a subliminally presented, food-related word. In the present study hours since last food was equated with hours since cafeteria closing because there was no other source of a meal available to Ss. Therefore, the experimental Ss could be considered food-deprived according to the Spence and Ehrenberg figure of at least 2.6 hr. since last food. Other oral activities such as smoking and gum chewing, were not systematically distributed in the experimental group since it was a representative sample of students presently attending college. Therefore, the negative findings obtained in the present study (reported above) are contradictory to those of Spence and Ehrenberg ( 1 2 ) . In addition, George and Jennings ( 6 ) found that 30 subliminal presentations of the stimulus word CHEESEdid not increase subjective hunger ratings in Ss whose hours since last food was more than 2.6 hr. A logical explanation for these contradictory findings is that purchasing behavior and subjective hunger ratings are probably q~ialitativelydifferent from the dependent variable of differential recall of cheese associates over neutral words employed by Spence and Ehrenberg.

852

S. G. GEORGE

&

L. B. JENNINGS

Since Jennings and George ( 7 ) have demonstrated that the method for determining S s awareness used by Spence ( 11, 12, 13) does not ensure subliminality in respect to a forced-choice detection threshold, it is entirely possible that Ss in question were stimulated near the threshold of recognition. Most other investigators reporting positive evidence of subliminal perception have employed stimuli which are only subliminal to a recognition threshold, rather than to the more stringent detection threshold ( 1, 4, 1 0 ) . These studies have been faulted by George and Jennings ( 5 ) for failing to ensure that genuine lack of recognition rather than response bias determined the absence of reported awareness of the stimulus material. From the conventional psychophysical point of view it is certainly more logical to expect that stimuli near, but below, the recognition threshold would affect behavior than the decidedly subliminal stimuli studied in the present research. To test this possibility a second study was done. Since the original subliminal advertising study by Vicary and others ( I ) , (which stimulated Ss near recognition) have used motion pictures as the background media, the stimulus HERSHEY'SCHOCOLATEwas superimposed over a movie in the second study. Ten segments of a movie on juvenile delinquency were selected which possessed similar illuminations and textural uniformity, i.e., a wall, a foggy alley, in the center of the picture, and the same procedures using the two pilot Ss to determine the 10 respective illuminations at which to present the critical stimulus were employed. The Ss ( 15 experimental and 12 control), who participated in their respective sections of introduction to psychology were food-deprived according to the Spence and Ehrenberg ( 12) criterion and were aware that E was superimposing a stimulus. E adjusted the rheostat according to the predetermined levels and presented the stimulus 10 times at each of the 10 segments of the movie. Since none of the Ss could describe the stimulus when offered four dollars for a partial description and as George and Jennings ( 5 ) found that a dollar reward was almost as sensitive at information extraction as a forced-choice procedure at the recognition level, response bias was precluded and the stimulus could be considered genuinely subliminal to a recognition threshold. Only one S in each the experimental and control group guessed that the purpose of the superimposed stimulus was to influence buying behavior, but neither guessed the nature of the product nor the fact that the student store was monitored. NO S in the experimental group purchased Hershey's Chocolate; whereas two Hershey bars were bought by the control group. When only Nestle's chocolate bars were considered, Ss from the experimental group purchased two bars in contrast to one from the control group. The above data do not even indicate a trend, let alone significant differences favoring the notion that stimuli presented near, but below, the recognition threshold can influence buying behavior.

SUBLIMINAL STIMULI IN CONSUMER BEHAVIOR

853

Since the follow up of the second study was only one month after the original one, an additional analysis of buying behavior of the original Ss could be made. Although motivational research theorists apparently maintain that the effects of subliminal stimulation are almost immediate (well within the 10-day monitoring period used in the present study), the second study offered an additional cross-section of buying behavior a month after initial stimulation of the first Ss below a detection threshold. During the second purchase period only one S from each of the experimental and the control group bought a Hershey's Chocolate bar. Total Hershey's products purchased by experimental and control Ss were 1 and 3 respectively. However, if total purchases of both Hershey's and Nestle's products from both 10-day purchase periods are considered, then the experimental group bought 7 and the control group purchased 12. Certainly, the preceding evidence offers no support to the contentions of motivational research theorists or other theorists (13) that stimuli which are completely subliminal in respect to any sort of conscious awareness, i.e., below a forced-choice detection threshold, can influence behavior through communication with unconscious processes. Nor do the findings offer any support for the more plausible, but still highly unlikely, notion chat stimuli subliminal to a recognition threshold could influence buying behavior. REFERENCES The effect of a subliminsl food stimulus on verbal responses. Iousnal o f Applied Psychology, 1959, 43, 249-252. CHAMPION,J. M., & TCTRNER,W. W. An experimental investigation of subliminal perception. ]ournal o f Applied Psychology, 1959, 43, 382-384. DE FLEUR,M. L., & PETRANOFF,R. M. A televised test of subliminal persuasion. Public Opinion Qtrarterly, 1959, 23, 168-180 EAGLE,.M. The effects of subliminal stimuli of aggressive content upon conscious cognition. Journal o f Personality, 1959, 27, 578-600. GEORGE,S. G., & JENNINGS, L. B. Effects of subliminal stimuli on dreams: further evidence against the Spence-Holland theory. Perceptual and Motor Skills, 1972, 35, 251-257. GEORGE,S. G., & JENNINGS,L. B. Reexamination of effect of a subliminal verbal food stimulus on subjective hunger ratings. Psychological Reports, 1972, 30, 521522. JENNINGS,L. B., & GEORGE,S. G. The Spence-Holland theory of subliminal percepuon: a reexamination. Psychological Record, 1970, 20, 495-504. JUNG, J. Restricting effects of awareness: serial position bias in Spence's study. ]ournal of Personality and Social Psychology, 1966, 3, 124-128. MCCONNELL,J. V., CUTLER,R. L., & MCNEIL, E. B. Subliminal stimulation: an overview. A m e ~ i c a nPsychologist, 1958, 13, 229-242. SMITH, G. J. W., SPENCE, D. P., & KLEIN, G. S. Subliminal effects of verbal stimuli. Journal o f Abnormal and Social Psychology, 1959, 59, 167-176. SPENCE,D. P. Effects of a continuously flashing subliminal verbal food stimulus on subjective hunger ratings. Psychological Reports, 1964, 15, 993-994. SPENCE,D. P., & EHRENBERG,B. Effects of oral deprivation on responses to subliminal and supraliminal verbal food stimuli. Journal o f Abnormal and Social Psychology, 1964, 69, 10-18.

1. BYRNE,D.

2.

3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

854

S. G. GEORGE & L. B. JENNINGS

13. SPENCE, D. P., 9: HOLLAND, B. The restricting effects of awareness: a aradox and an explanation. Journal o f Abnormal and Social Prychology, 1962, 163-174. 14. WORRELL,L., & WORRELL,J. An experimental and theoretical note o n conscious and preconscious influences on recall. Journal of Personality and Social Psychology, 1966, 3, 119-123.

&,

Accepted September 23, 1975.

Effect of subliminal stimuli on consumer behavior: negative evidence.

The study corrected methodological weaknesses found in previous experiments designed to test the contentions of motivational research theorists that s...
303KB Sizes 0 Downloads 0 Views