This paper reflects the research and thoughts of a student at the time the paper was written for a course at Bryn Mawr College. Like other materials on Serendip, it is not intended to be "authoritative" but rather to help others further develop their own explorations. Web links were active as of the time the paper was posted but are not updated.
1998 Second Web Reports
Many people have heard of "subliminal" messages that are not consciously perceived by a subject but nonetheless influence his or her behavior. The concept first became publicized in the 1950s, when the advertiser James Vicary claimed that flashing the words "Drink Popcorn" and "Drink Coke" between frames in a movie theater increased Coke sales by 18.1% and popcorn sales by 57.7%. This caused a storm of controversy, although Vicary later admitted that this "study" was a hoax.(1)Many studies carried out in laboratory conditions do show that "subliminal" inputs can be perceived by the nervous system without the awareness of the conscious "I-function". Inputs perceived "subliminally" have also been shown to influence behavior to some extent, in some instances more than supraliminal inputs. These effects, however, are for the most part limited in magnitude and duration.
The literal meaning of the word "subliminal" is rather misleading since it implies that there is an absolute "limen", or threshold, above which inputs are detected by the "I-function" and below which they are not.(2) However, while there are some "supraliminal" inputs of high intensity where perception is always self-reported by experimental subjects, as well as inputs of low intensity that are never reported, there is also a range of input intensities where subjects report only a fraction of the time.(3) The actual experimental definition of the "liminal" or threshold value of an input is the value where conscious perception occurs 50% of the time. The "limen" can also fluctuate with different prevailing conditions.(4) Therefore, an input could be classified as "subliminal" and still be accessible to the I-function.
Some experiments that indicate a conflict between "self-report" of the perception of an input by a subject and other evidence of perception are "forced-choice" experiments, in which subjects who self-report that they cannot perceive an input are directed to guess what it is from a number of presented choices. The accuracy rate of such "forced choices" usually exceed chance.(2) An early experiment conducted in 1898 involved flash cards imprinted with characters that were presented to subjects at such a distance that they reported only "a dim, blurred, spot or dot". However, if the subjects were forced to guess what the character was, their accuracy was greater than chance.(5)
Other experiments utilized conditioning and the Galvanic Skin Response, changes in skin conductance to electricity that can be measured by a polygraph, to create a response other than self-report that could be indicate perception of an input. Subjects were first conditioned to increase their GSR upon presentation of a series of nonsense syllables like ZUM and TOV by pairing them with a mild electric shock. The syllables, mixed with new ones, were then presented at a short time exposure providing no self-report of perception. However, the subliminally presented syllables still exerted their conditioned effects by increasing GSR levels. (4) Once again, there was discrepancy between direct self-report and other indirect measures of perception.
There is also evidence that subliminally perceived inputs can affect behavior. However, inputs have never been shown to affect behavior in the absence of any evidence of perception. There is a point at which inputs are so weak that on "forced-choice" tests accuracy does drop to chance levels, often called an "objective measure" perception compared to the "subjective measure" of self-report.(1) So far, inputs that have failed to override this "objective measure" threshold have also failed to show any impact on behavior.(2)
In a type of experiment called a "priming" experiment, a preliminary input, called a "prime", is presented that is either congruent or incongruent with a target input which is to be evaluated by the subject. In one such experiment, color patches flashed before the subject were "primed" with a word of a color that either matched or did not match the color present. Some of the primes were rendered subliminal by flashing a "masking" input of random letters directly after the prime word. If the delay between prime and mask was relatively long, the input fell below "subjective threshold" but not below "objective threshold." At very short delays the input fell below "objective threshold", with "forced-choice" guessing yielding only chance results. At the supraliminal level, incongruent primes resulted in a slower identification of the color presented than congruent primes. This was also true, to a lesser extent, for subliminal primes below "subjective threshold", but not those below "objective threshold".(2)
Another "priming" experiment in which subjects had to classify a word as either a male/female name, or pleasant/unpleasant, and were primed subliminally by words that had a congruent or incongruent classification, yielded similar results. In this experiment the subjects were required to respond 0.4 seconds after the target word was flashed, and the effect of the incongruent primes were to increase the error rate in classification.(6)
The most interesting experiments indicate that in some instances subliminally perceived inputs are more effective than supraliminal ones in affecting behavior. In a color-patch experiment using only the colors red and green, if the ratio of incongruent to congruent primes was increased, subjects began to associate the prime with its incongruent color, and the results of the original experiment were reversed. However, this reversal only occurred for supraliminal primes, while incongruent subliminal primes still slowed identification.(5) Subliminal inputs may lead to a direct, "automatic" output because they are not consciously processed and cannot be interpreted in context by the "I-function".
There is also experimental evidence for the effect of subliminal inputs on the preferences of a subject. The positive effect of any repeated input on eventual preference for that input has already been documented as the "mere exposure effect".(7) Subjects showed a higher preference for polygonal figures or pictures of human faces that they had been shown repeatedly compared to newly introduced ones. This effect was more pronounced when the repeated prior exposures were subliminal as compared to supraliminal.(8) This may be because the relevancy of subliminal inputs cannot be judged by the "I-function" and are not discarded as irrelevant. Another experiment, while only conducted on the subliminal level, showed that subjects primed before a poetry discussion with repeated exposure to pictures of confederates on one side tended to prefer that side when forced to act as tie-breaker.(8)
However, even these situations differ from the claims of Vicary in that the subjects were always forced to respond in some way. It was the quality of the responses, not whether they occurred in the first place, were affected by subliminal inputs. These experiments do not indicate that arbitrary subliminal inputs can cause behavior that would not otherwise take place. For example, studies of "self-help" tapes billed as containing subliminal messages have not shown to be effective beyond an illusory placebo effect. (9) While subliminal perception definitely exists and subliminal inputs can affect behavior, so far no evidence exists that subliminal inputs can be used to exert any "mind control" on unwilling subjects.
4. SUBLIMINAL PERCEPTION
5. Unconscious Perception
7. Subliminal Mere Exposure
8. The Subliminal Mere Exposure Effect
9. CJBS: Subliminal Self-Help
Sources not cited directly:
Reinventing Subliminal Again
| Course Home Page
| Back to Brain and Behavior
| Back to Serendip |