Although Vicary never actually published these findings, his reports created a frenzy of consumer concern and government legislation aimed at stopping these forms of seemingly insidious mind control. There was, however, one significant and often unknown problem with Vicary's study -- it was all a hoax. Years later, Vicary himself admitted this scam was simply an attempt to save his dying advertising agency.
The notion that stimuli presented outside conscious awareness could influence cognition was not new -- in 1884 Peirce and Jastrow reported that people could perceive small differences in pressure to the skin without conscious awareness of different sensations. Vicary, however, was the first to report impressive behavioral influences in a domain with tremendous money making potential.
Regardless of its truth of falsity, Vicary's report spurred what has now been over 30 years of research into the effectiveness of a new marketing tool: subliminal advertising. The resulting body of work, not surprisingly given its dubious origin, has produced far from impressive results and has led to the strong conclusion that subliminal advertising cannot influence the type of laundry detergent you buy, the type of underwear you wear, or the candidate you vote for.
In short, research on subliminal advertising suggests that subliminally presented stimuli have little or no influence on our thoughts, attitudes, preferences, or behavior. Or do they? Cognitive and social psychologists are now learning that stimuli presented subliminally can not only be perceived, but can have a considerable influence over a variety of cognitive processes (possibly even behavior). This research stands in stark contrast to the failings of subliminal advertising, and in this module we will investigate the replicability, generalizability, and validity of some of these new findings.
We will introduce you to the nuts and bolts of subliminal presentations by letting you try to replicate two, purportedly reliable, subliminal effects. Along the way we will consider the generalizability and reliability of subliminal influences, whether cognitive scientists can accommodate subliminal perception in current theories about the mind, why the phenomenon of subliminal perception is even interesting to cognitive scientists, and whether there are any marketing implications for subliminal perception.
I should note at the outset that we will be restricting our discussion of subliminal presentations to those that are presented exclusively to the visual system. There is some evidence of subliminal perception in the auditory and tactile systems, but a discussion of these effects is beyond the scope of this module. Furthermore, we will restrict our attention to methods that have been found to reliably (but not without controversy) produce subliminal effects. Thus, audio self-help tapes with subliminal suggestions to "lose weight" or "be happy" are not considered, nor are "backmasked" messages instructing one to "smoke marijuana". Research has reliably shown that these methods produce no effects other than those expected by the listeners (Greenwald, Spangenberg, Pratkanis, & Eskenazi, 1991; Pratkanis, 1992).
This paradox has traditionally been resolved using indirect methods showing that stimuli can influence a person's thoughts or judgments even when they are unable to identify or recognize the stimulus. For a truly subliminal stimulus, scientists predict a null effect for measures of conscious awareness (i.e. no conscious recognition), but significant effects on some other judgment, attitude, preference, or behavioral measure.
For example, imagine you were in an experiment where you were flashed the word "help". Researchers might first ask you whether you could recognize the flash just presented on the screen. They might then ask you to identify certain strings of letters as words or non-words. Subliminal influence would be evidenced if you were unable to indicate that you were flashed the word "help" but were nevertheless faster to identify words related to "help" (e.g. "share", "donate", "give") than those that were unrelated (e.g. "stem", "rotund", "leaf"). This is actually an example of a commonly used research paradigm that seeks to find a dissociation between measures of conscious (awareness/recognition) and unconscious (word recognition) responses.
Two general classes of measures have been identified: subjective and objective (Cheesman and Merikle, 1985). Subjective measures of awareness rely on participants' self-reports of their perceptual experiences or conscious processing. Questions like "Did you recognize the word that was presented to you?," or "Do you think the word that was presented influenced your judgments?," are examples of subjective measures of awareness. Objective measures, on the other hand, rely on a participant's ability to discriminate between particular stimuli. Such measures included forced choices between two or more stimuli (e.g. "Which of these three stimuli have you seen before?"), or judgments regarding the presence or absence of a particular stimulus (e.g. "Was this one of the stimuli you were presented or not?"). Measures of objective awareness provide a lower threshold for conscious awareness than subjective measures.
Regardless of the specific measures used, however, this Dissociative Paradigm has come under attack in recent years. Critics argue that failures to find conscious awareness on either subjective or objective measures may simply reveal the insensitivity of conscious awareness measures rather than the true absence of conscious processing. Indeed, it is very easy to design awareness measures that yield no signs of awareness or stimulus recognition, if you make them impossibly difficult. While researchers don't intentionally develop measures that are insensitive, a null effect can never provide completely convincing evidence that no conscious awareness occurred. There is considerable debate on this issue, however. It seems eminently reasonable to many researchers that if you want to find out if someone is unaware of a particular stimulus, you just ask them. If they say, "I have no idea what you presented to me", that's evidence enough -- end of the issue. No conscious awareness is an inability to verbally report a stimulus -- it is a null effect.
Discussion: Consider the questions below, and also any questions you had while reading the assigned papers. Submit by e-mail written summaries of your discussion.
In light of your readings: 1. Should experiments utilizing subjective measures of awareness be accepted as evidence for subliminal persuasion? Why or why not? 2. Why is the distinction between subjective and objective measures of awareness important? 3. How does the Exclusion Paradigm you read about attempt to resolve the criticisms of the Dissociative Paradigm? Is it effective? (Keep the Exclusion Paradigm in mind-it might come in handy when designing your final projects.)
In the absence of "just because" interest, let me offer two reasons why the study of subliminal perception should be interesting to you, and to cognitive scientists generally. The articles you just read by Philip Merikle provide the first: that conscious and unconscious processes may be fundamentally different from one another. A particular stimulus, when presented consciously, may have a completely different effect than when it is presented unconsciously. In the Stroop Task experiment, Merickle and Joordens (1997) found that participants developed expectancies for certain color words when the presentations were conscious. These expectancies were not present when the stimuli were not consciously perceived, producing considerably different behavior between the two groups.
Similarly, Neely (1977) found that expectations have considerable influence on stimuli that are perceived consciously, but not those perceived unconsciously . In Neely's experiments, participants were presented with two words that are either semantically related (e.g. body-arm) or unrelated (e.g. body-door). Through instructions given by the experimenter, some participants were led to expect that when the word "body" was presented, a building part, such as "door", was likely to be presented. This conscious expectancy allowed participants to quickly recognize words related to building parts when "body" was presented for a long time (750 ms), but hindered this ability when "body" was presented for a short time (250 ms). Following these short presentations, participants were faster to recognize words related to bodies (such as arm), suggesting their conscious expectancy for building parts was not engaged. In this experiment, the word "body" produced very different effects depending on whether it was presented subliminally or not.
In addition to the possibility of producing unique effects, subliminal perception is interesting because it allows cognitive scientists to investigate the nature and function of consciousness. Some psychologists (e.g. Rogers, 1951; Bandura, 1977; Mischel, 1973) have long assumed that consciousness is critically important in a variety of domains, and have argued against those (such as B.F. Skinner) who suggest that it isn't. Of course, one test of whether consciousness is necessary for some process or effect is to see if it can be produced without conscious awareness. Subliminal perception provides an ideal vehicle for such investigation, and research seems to suggest an ever-decreasing number of processes for which consciousness is a critical mediator (Bargh, 1997). Those that remain provide tantalizing clues regarding why consciousness evolved, and what function(s) it serves (For one suggestion, see Baumeister & Sommer, 1997).
Word | "Left" | "Right" | "Yes" | "No" |
---|---|---|---|---|
Key | Q | P | J | F |
***After you've finished labeling the keys, restart your computer with all of the extensions disabled -- this is very important. Extensions require processing time from the computer, and can, without warning, slow down the presentation of stimuli. Consequently, some of the stimuli indented to be presented subliminally may be slowed down and perceived consciously. To turn off the extensions, restart the computer (use the Power key in the upper right of the keyboard, then select "Restart"). As the computer starts up, hold down the SHIFT key. You can release the shift key once you see the message saying that the extensions are disabled.
Once the computer is ready, go to the Apple Menu and find "Subliminal Perception Module f". Inside that submenu, select the "Experiment #1" menu, and then select the item "Experiment #1". This should open SuperLab, the software used to run this experiment.
Once SuperLab is running:
For example, imagine you wanted to parafoveally prime a person with the word "Stop". To make sure the stimulus remained within the parafoveal visual field, you would first need to create a fixation point at the center of the screen, such as a bulls-eye. You would then need to calculate the exact distance a person's eyes would be from the screen, and calculate the distance you would need to move from the fixation point in order to make an angle at the person's eye of less than 6 degrees. Foveal priming is mathematically much easier, and would involve presenting the stimulus directly at the fixation point.
We see from the above diagram that if the viewer is distance D from the screen, then the outer boundary of the parafoveal region is distance douter= D * tan(6°) from the fixation point and the inner boundary of the parafoveal region (which is the outer boundary of the foveal region) is distance dinner= D * tan(1°) from the fixation point.
It is important to notice, however, that these visual boundaries are only approximate, and these regions are traditionally distinguished for functional, rather than psychophysical, purposes (Holender, 1986). Visual acuity is actually a continuously decreasing function of the distance from fixation (Anstis, 1974), nothing magical happens when you cross the boundaries between these regions.
An alternative to shorter presentation times and Tachistoscopes are visual masks. These are generally meaningless words or images, presented immediately following the critical stimuli, that are intended to "erase" the afterimage. You'll still experience an afterimage, but it will be of the mask, not the critical stimulus, allowing you to present stimuli for longer durations. These masks are usually random strings of letters if you're presenting words (e.g.XQFBZRMQWGBX), or simple meaningless patterns (e.g. cross-hatched lines) if you're presenting pictures. Since we have computers readily available, and have no desire to add unnecessary equipment to the lab, masks will be used in both of the experiments in this module.
Unfortunately, there is no absolute presentation time when a stimulus moves from being subliminal to supraliminal (above threshold). These times vary considerably depending on how you measure awareness (objective, subjective, etc.), whether or not you use a mask, the illumination, size, and complexity the stimulus, as well as the particular individual you are testing. Given this variability, the exact presentation times must be recalibrated to fit the particulars of each experiment. That said, however, there are still some general guidelines or rules of thumb you can follow to start this calibration process. The table below presents these ranges based on current research, with question marks in cells where research is unavailable or unknown.
Mask/Awareness Condition | Priming Method | |
---|---|---|
Foveal | Parafoveal | |
Mask/Subjective Awareness | 13-55 ms | 50-100 ms |
Mask/Objective Awareness | 10-20 ms | 50-70 ms |
No Mask/Subjective Awareness | 10-20 ms | ? |
No Mask/Objective Awareness | < 10 ms | ? |
Note. Parafoveal priming techniques tend to use words as stimuli, while foveal techniques tend to use images. This association should be remembered when interpreting this table. Also note that these ranges are taken from experiments in which the critical stimuli were presented individually, with no other distracting objects or words in the visual field. Some experiments (See Holender, 1986, for a review) involve presenting critical stimuli in the parafoveal region at the same time as another distracting word or object in the foveal region. Experiments using these methods involve considerably longer presentation times, sometimes up to 250 ms.
If activation of one concept spreads to related concepts, then presenting you with the word 'Hawaii' should subsequently make you faster to answer that 'water' and 'sand' are real words. In this case, 'Hawaii' would prime 'water' and 'sand'. What if the word 'Hawaii' was presented subliminally, would it still activate 'water' and 'sand'? The first experiment was designed to investigate this question. In this experiment, 20 pairs of semantically related pairs were selected, such as 'ocean-water' and 'start-begin'. One of these words was presented subliminally, and the other was used as the target word in a lexical decision task.
To determine whether your recognition of the target word was facilitated by subliminal presentations, 20 words were matched to the target in both length and frequency of usage ( Carroll, Davies, & Richman, 1971). Frequency of usage is a critical control, because people are, of course, faster to recognize words that are common than those that are not. When matched on frequency of usage, a word that has been recently activated should be recognized faster than one that hasn't. Considerable evidence suggests that subliminal presentations can influence the recognition of semantically related words (e.g. Kemp-Wheeler & Hill, 1988; Marcel, 1983). Some of this research has been criticized for the use of subjective measures of awareness, but similar findings have been found using methods designed to counter these problems, such as the exclusion task that you read about earlier (see also Greenwald & Draine, 1997; Greenwald, Draine, & Abrams, 1996).
Discussion: Consider the questions below, and also any questions you had while reading the assigned papers. Submit by e-mail written summaries of your discussion.
Prime | Target | Unrelated | Nonsense #1 | Nonsense #2 |
---|---|---|---|---|
help | donate | rotund | frentd | pawet |
start | begin | hours | guipd | -- |
ocean | water | could | phewe | -- |
angry | mad | dug | gix | -- |
child | infant | levers | entupe | jetqen |
hint | clue | stem | tepr | -- |
stop | end | few | rtn | chp |
rear | back | know | daot | fuut |
rope | string | total | spont | allit |
shy | quiet | teeth | stoow | -- |
boring | dull | shed | sask | -- |
sad | unhappy | mineral | slekies | compber |
moral | ethical | boilers | whistpy | hedfer |
throat | neck | rate | easp | -- |
share | give | line | cahe | refr |
shoe | boot | sped | kiwn | nomr |
road | street | trying | clopir | -- |
alarm | warning | lengths | petsuns | greum |
female | woman | pages | boaps | -- |
coat | jacket | habits | mepalz | -- |
Subliminal Presentations The experiment is designed so that each lexical decision task is preceded by 8 subliminal presentations containing the same word. These words, labeled "Prime" in the table above, are presented parafoveally, followed by a mask (XQFBZRMQWGBX). While these words are being presented, you're asked to focus carefully on a bulls-eye at the center of the screen and to indicate whether the flashes (the word and the mask) occur to the left or the right of the bulls-eye. This visual reaction time task is simply used as a cover-story to help mask the subliminal presentations. The subliminal stimuli are presented 5 cm away from the bull's-eye, allowing the words to remain in the parafoveal visual field when your head is positioned 64 cm away from the screen.
While the center of each word is always presented 5 cm from the bulls-eye, each word is not always presented in the same location. Instead, words are presented 45 degrees off horizontal in each of the 4 corners of the screen. The exact location is determined at random. Each of the priming words is presented for approximately 80 ms, or 6 screen refreshes on our 75 Hz monitors. Notice that we didn't try to calibrate each person to determine his/her own threshold for subliminal perception, this would have taken considerable time and would have been rather complicated. Instead, we relied on other researchers, including myself, who have found this procedure produces no subjective awareness of stimuli, and more importantly on pilot testing done at Cornell University using this exact procedure.
Given that we didn't individually calibrate subliminal thresholds, some of you may have seen a few of the stimuli, particularly given that you knew beforehand that some words are being presented. A number of the pilot participants run through this procedure mentioned similar feelings, but when actually asked to identify which stimuli are presented, they are unable to do so reliably.
Lexical Decision Task. After being subliminally presented with 8 stimuli, you are asked to perform a lexical decision task. This is the point at which the data critical to our hypothesis are collected. In the lexical decision task, you are presented with either 3 or 4 strings of letters and asked to determine, as quickly as possible, whether each string creates a real word or not. Of the strings you see, one is the target word -- a word semantically related to the prime word. Another string is the unrelated word -- a word that is matched on both frequency and length to the target word.
There are also, depending on the trial, either one or two nonsense words -- random strings of letters. The order in which these words are presented is random, except that the second nonsense word is always last. This occasional second nonsense word is used to disguise the composition of the task, as it would otherwise be obvious that there are always 2 real words and 1 non-word.
Equipment. This experiment is run using SuperLab, a software program designed primarily for running experiments. Stimuli used in experiments are not created directly in SuperLab, but are instead created in a separate drawing program and saved as a PICT file. The stimuli for this experiment are created using Canvas, but any drawing program will work. SuperLab can then present these PICT files in any order and speed you like, while also recording key strokes and reaction times. If you are interested in learning more about SuperLab, a manual can be obtained from your course instructor. If you'd like more information about the structure of this experiment in SuperLab, see Appendix A.
To analyze the data:
Discussion: Before breaking into discussion groups, everyone in the class should report his or her average difference score. Every person should write all of these scores down, enter them into a new StatView file, and perform a one group t-test on the entire class data to determine if our major hypothesis was confirmed: that people, on average, will be faster to respond to words that are semantically related to the subliminal prime than to those that are unrelated. After you have collected the data from your classmates and analyzed the data, consider the questions below in light of your findings. Submit by e-mail written summaries of your discussion.
Discussion Questions:
According to the Thorndike-Lorge count [a compilation of how often words are used in English texts], the word "happiness" occurs 761 times, "unhappiness" occurs only 49 times. "Beauty" is to be found at least 41 times as often as "ugliness," and "wealth" outdoes "poverty" by a factor of 1.6. We "laugh" 2.4 times as often as we "cry"; we "love" almost 7 times more often than we "hate." We are "in" at least 5 times more often than we are "out"; "up" twice as often as we are "down"; much more often "successful" than "unsuccessful"; and we "find" things 4.5 times more often than we "lose" them -- all because most of us are "lucky" (220) rather than "unlucky" (17). We have all the reasons in the world to be "happy" (1449) and "gay" (418) rather than "sad" (202) and "gloomy" (72), for things are 5 times more often "good" than "bad," almost 3 times more often "possible" than "impossible," and about 5 times more "profitable" than "unprofitable."
This correlational evidence is supported by hundreds of studies showing the more we are exposed to a stimulus, the more we like it. In short, familiarity breeds approval, not contempt (Gilovich, 1998). This tendency for repeated exposures to enhance liking has been termed the Mere Exposure Effect. Of course, one could argue that repeated exposure provides people with more and more information about a particular stimulus, causing them to see things they have never seen before -- "Gee, I never realized just how interesting that 'swoosh' symbol really is". Thus, it might be that we like objects and people we know more about, not necessarily those that are merely more familiar to us. This explanation doesn't seem completely unreasonable for stimuli we can see consciously, but what about stimuli we can't? It seems odd to suggest we glean meaningful information from stimuli that are presented beneath our level of conscious awareness, i.e. subliminally. Many researchers have investigated whether mere exposure to subliminally presented stimuli can enhance one's liking for those stimuli. The answer seems to be, yes, it can.
Discussion: Consider the questions below, and also any questions you had while reading the assigned papers. Submit by e-mail written summaries of your discussion.
To create this experiment, 10 pairs of abstract polygons were designed that were matched on both size and number of sides. This matching was important because we want to make sure we control all aspects of the stimuli except the number of exposures. These polygons were then numbered from 1-20, with adjacent numbers paired together (1 & 2 are a pair, 3 & 4 are a pair, etc.). In each pair, only one polygon is presented subliminally allowing us to create 2 conditions: those who are subliminally exposed to the odd numbered polygons (Odd condition), and those who are exposed to the even numbered polygons (Even condition). Images of these polygons are contained in Appendix B.
After 7 subliminal exposures to one polygon, people are asked which polygon in each pair they prefer. Our prediction is that, regardless of which particular polygon was presented, people will more often prefer the polygon that was presented subliminally. Notice this design rules out the concern that the subliminal polygons were actually the most attractive polygons, regardless of repeated exposures, since we're predicting that the polygon judged most attractive will vary depending on which one was presented subliminally. In addition to showing the mere exposure effect, we also need to show that the stimuli cannot be consciously perceived.
Therefore, we've created a second part of the experiment in which people are subliminally presented with the same stimuli, but instead of being asked which of two polygons they prefer, they're asked to indicate which of two polygons they've seen before. We predict that people will prefer the polygon they've been subliminally presented, but will be unable to reliably detect which polygon they've seen before. Thus, we expect an overall preference for previously seen polygons that is significantly higher than chance guessing, or 50%, but recognition of stimuli that is not different from 50%.
Unfortunately, we also have to add a bit more complication to this design since we can't have everyone do the preference ratings first and the awareness ratings second. Why not? Well, there is a possibility that the order in which the questions are asked may influence our results. For example, people may catch on to the design of the experiment and start looking for the polygons. If this happened, people may become more and more aware of the stimuli as the experiment progresses. To make sure this is not a problem, we can simply add another independent variable in which we manipulate which of the two types of questions are asked first. In one condition, people will answer the awareness questions first. In the other, people will answer the preference questions first. We will call these, respectively, the Awareness and Preference conditions to indicate which of the two tasks is performed first. We expect this variable to have no influence on the data. The table below will help you visualize the four different conditions in this experiment.
Experimental Conditions | |||||||||||
|
In this procedure, the critical polygon is presented for 13 ms (1 screen refresh) in the center of the screen. The polygon is then followed by a pattern mask made up of cross-hatched black and white lines. This mask is approximately the same size as the original polygon. This first mask is then followed by a second mask that is simply the negative image of the first. Pilot testing revealed this was sufficient to disrupt the perceptual field and keep the polygons outside conscious awareness. Immediately following the second mask is an image of small black and white dots.
Participants are told their task is to determine, as quickly as possible, whether there are an even or an odd number of dots on the screen. This task is important only to mask the real purpose of this experiment -- to subliminally present people with polygons. This procedure -- polygon, mask1, mask2, dots, -- is repeated ten times. To reduce the probability of consciously perceiving the polygon, however, the figure is removed in three of the presentations. This means the participant is actually presented with each polygon only seven times before making his/her preference or awareness judgments.
Label | "Even" | "Odd" | "Right" | "Left" |
---|---|---|---|---|
Key | M | C | I | E |
It's very important that these keys be labeled correctly, the computer determines whether a response is correct or incorrect based on the specific key that is pressed.
To set up the experiment:
Once the file is opened:
To organize the data:
Now things get a bit tricky. It would be a problem in this experiment if all of the polygons in the odd condition are presented on the left side of the screen during the choice measures, and all of the even ones on the right. To remedy this, the positions of the polygons are randomized: sometimes polygons in the even condition are presented on the right and sometimes they are presented on the left. These positions are included in the SuperLab program, and thus it can determine whether the polygon selected is the one presented subliminally or not. If it is the same, this is indicated with a letter "C", for correct, in the column titled "Error Code". If it is different, this is indicated with an "E", for error, in the same column.
To determine whether people choose polygons that were presented to them subliminally, you should be able to scan down the Error Code column, record the response for each trial, and tally them up. Of interest is the percentage of times the participant chose the subliminally presented polygon (Labeled C, for correct response, in the Error Code column). Each trial is labeled, in the "Trial Name" column, by the trial number as well as the task: Aware or Pref. You simply need to find all 10 trials and count up the number of C's and E's.
Ideally the program would work this way, but sometimes the participants press keys other than those you labeled -- pressing "O" instead of "I" for example. Unfortunately, pressing a key that is not included in the program is registered as correct, even though it might not be. To make sure the calculations you just made are correct, a list of correct key responses (selecting the subliminally presented stimuli) are listed in the table below. You should double-check and make sure there are no mistakes by comparing this table with the responses in the column titled "Response".
Trial | Even | Odd |
---|---|---|
1 | I | E |
2 | I | E |
3 | I | E |
4 | E | I |
5 | I | E |
6 | E | I |
7 | I | E |
8 | E | I |
9 | E | I |
10 | I | E |
Note: This table includes the correct key responses for each trial in the even and odd conditions. "I" indicates selecting polygon on the right of the screen, "E" indicates selecting a polygon on the left side of the screen. So, for example, in Trial 1, the Even numbered polygon is presented on the right side of the screen and the odd numbered polygon on the left. Consequently, a correct response (selecting the polygon that was presented subliminally) for those in the Even condition would be 'I', to indicate the preferred or recognized stimulus is on the right side of the screen. A correct response in the Odd condition would be 'E', to indicate the preferred or recognized stimulus is on the left side of the screen.
When you are finished, you should have 8 different percentages: 2 percentages from each of your 4 participants indicating the percentage of times they preferred the subliminally presented polygon and the percentage of times they indicated awareness for the subliminal polygon. Average together your 4 preference and 4 awareness percentages to come up with an average preference and awareness percentage. We predict the first will be significantly greater than 50% and the second will be no different than 50%.
To test our main hypotheses, two one group t-tests should be conducted on the %Prefer and %Aware columns to determine whether either of these are significantly different from 50%. In the first t-test, you want to see if people prefer the polygons presented subliminally more than chance guessing (50%) would dictate. In the second, you want to see if people can recognize, or are aware, of the stimulus that was presented subliminally more than chance guessing would dictate. A subliminal mere exposure effect would yield a significant difference on the preference measures but not the awareness measures -- they prefer the polygon presented subliminally, but can't identify it consciously.
After analyzing the data, consider whether our hypotheses were confirmed or disconfirmed. Also feel free to discuss possible extensions to the experiment, or ways in which the experiment could be improved. Submit by e-mail written summaries of your discussion.
Let's take the two questions mentioned above and consider them separately. First, is accuracy in the awareness measures correlated with confidence? Are people who are more confident that their judgment is correct actually more likely to be correct than people who aren't confident? To find out, go back to the original data file for each of your subjects. What you need to do is create a table that will allow you to look at both the choice (which polygon have you seen before) and the degree-of-confidence (how confident are you that this judgment is correct?) for each trial. You should make a table like the one printed below:
Trial Number | Choice (C or E) | Degree (1, 2, or 3) |
---|---|---|
1 | ||
2 | ||
3 | ||
4 | ||
5 | ||
6 | ||
7 | ||
8 | ||
9 | ||
10 |
For each trial, you will have both the person's choice rating and their degree or confidence rating. You now need to calculate the average degree of confidence rating for the times when the person's choice was correct (labeled C in the Response column of your Excel spreadsheet) as well as when it was incorrect. If people were actually more likely to be correct when they were highly confident, then we would expect a higher confidence rating when their answer was correct than when it was incorrect. You will need to do this for each of your 4 subjects. When you are finished, make a table like the one below and fill in all of the relevant numbers.
Subject | Response (correct or error) | Average Confidence |
---|---|---|
1 | C | (Average over correct trials) |
E | (Average over incorrect trials) | |
2 | C | |
E | ||
3 | C | |
E | ||
4 | C | |
E |
Enter the table above into a StatView file (best would be one row per subject with one column for confidence over correct trials and one column for confidence over incorrect trials). You should also get the data from your classmates and enter it into your spreadsheet to increase the sample size. When you have all of the data from the class, run a paired t-test to see whether the mean confidence ratings are actually higher when people are correct than when they are incorrect. You should now do the very same thing for your preference measures. This will allow us to test whether people have stronger preferences when they select polygons they've seen subliminally compared to when they select other polygons.
Subliminal presentations also seem capable of influencing behavior. For example, people with a competitive disposition were more likely to compete in a game after subliminal exposures to competitive words (Neuberg, 1988). In an even more impressive study, participants behaved more aggressively towards another student after their concept of hostility had been subliminally activated ( Bargh, Chen, & Burrows, 1996; Chen & Bargh, 1997). These findings raise many questions regarding the limits of subliminal influence and, more generally, of the extent to which our thoughts, judgments, and behaviors are triggered automatically by the environment.
Discussion: Consider the questions below, and also any questions you had while reading the assigned papers. Submit by e-mail written summaries of your discussion.
The experiments included in this module can be adapted to investigate a variety of other questions as well. You might want to think about these when designing your own independent projects. For example:
When designing your own independent project, you may want to alter the experiments already included in this module. Suggestions for alterations are included in Appendix A. Feel free, however, to design a completely new experiment if you are interested.
SuperLab files are organized hierarchically into blocks, trials, and events. Events indicate the stimuli that are actually presented on the screen. Series of events are linked together into trials, and series of trials are linked together into blocks. The images actually presented are not created in SuperLab, but rather are created in a separate drawing program and saved as a PICT file. These files are then linked to events in SuperLab. Before you start modifying the existing programs, you should become familiar with SuperLab by reading Chapter 2 from the SuperLab manual. Don't worry, it's very short (only 10 pages) and will make you familiar enough with the program to follow the instructions below.
Semantic Priming Experiment: This experiment is organized into two different blocks. The "First" block contains the introduction and practice trials. The "Second" block contains the rest of the experiment. Each of the trials is labeled descriptively. The trials containing the subliminal presentations are labeled by a number (One, Two, etc.) and the words in the Lexical Decision Task are labeled by a number (1,2, etc.) and a description of the word presented (related, unrelated, nonsense, sec. nonsense). The trials containing the subliminal presentations include many different events.
Each time a word is presented subliminally, four events must occur. First, a delay containing only a blank screen is presented for 1.5 seconds. These events are simply labeled "Delay", followed by a unique number. SuperLab doesn't allow you to use the same name for different events, so we add a number to the label. Second, a fixation point or dot must be presented in the middle of the screen. These events are labeled "Dot", and are presented at random intervals ranging from 2-5 seconds. Third, the word is presented for roughly 80 milliseconds depending on the monitor you're using. These events are labeled by the name of the word and also their position on the screen. The screen is divided into quadrants by drawing a horizontal and vertical line through the center. The quadrants are labeled from 1-4: Upper left=1 Upper right=2 Lower left=3 Lower right=4. Fourth, and finally, a mask is presented over the critical word for roughly 80 milliseconds. This event is labeled "Mask" followed by the position of the mask.
The lexical decision task contains a short paragraph of instructions, and a second series of instructions to begin the next round of subliminal presentations. Each word contains a short delay followed by a word or nonsense string of letters. These stimuli remain on the screen until a key is pressed. The labels for these events are descriptive and should be easy to identify.
Mere Exposure Experiment: The mere exposure experiment is also organized into two blocks: Awareness and Preference. The Awareness block contains all the trials and events related to the awareness items, while the Preference block contains everything related to the preference items. The order of these blocks determines which questions are presented first. The trials are labeled with descriptive titles. The subliminal presentations are labeled by numbers (One, Two, etc.) corresponding to the numbered pair (Pair #1, Pair #2, etc.). The trials containing the choice and degree measures are labeled by the number of the trial (1,2, etc.) and the type of judgment made. The choice measures are labeled by either "Aware" or "Prefer", depending on the question being asked. The degree measures are labeled by the number of the trial, the word "Degree", and the type of judgment made (Aware or Pref.).
The events are a bit more complicated because a variety of events are associated with each subliminal presentation. First, a delay containing a blank screen is presented for 2-5 seconds. These events are titled "Delay", followed by the length of the delay. These lengths were randomly determined before the program was designed. Second, the polygon is presented for approximately 13 milliseconds, or 1 screen refresh. These are labeled with the word "Poly" followed by the identification number of the polygon. The polygon is followed by two masks, labeled Mask One and Mask Two. Each of these masks is presented for approximately 36 milliseconds. Finally, a screen containing small circles are presented.
Participants indicate whether they think there are an even or odd number of circles. This event is labeled by "Circles" or "Circ." and the number of circles presented. For each number ranging from 7-20, two configurations of circles were created. Thus, some labels include the word "Two" to indicate the second configuration.
Both experiments can be modified with the same procedures since they were both created in SuperLab. There are several common modifications that I will explain here. More specialized or extensive modifications not covered below may require help from the instructor or SuperLab manual. Always alter a copy of the experiments rather than the experiment file that the class is using as the template.
After All Modifications: Depending on the modifications you make, you may need to go back and make sure all of the events are linked to the appropriate trials. As a rule of thumb, you should check for this after all your modifications.
Altering Presentation times:
One thing you might want to do is decrease or increase the duration at which stimuli are presented. This is fairly easy to do.
Please note that your exact presentations times are somewhat restricted by the computer monitor. The 75 Hz monitors we are using 'redraw' themselves every 13.3 milliseconds and you can present stimuli only in intervals of 13.3 milliseconds. If you'd like to present something for exactly 50 milliseconds, for example, you can't. The closest you can get is 53 ms. (4 screen refreshes).
Altering Existing Stimuli, or Creating New Ones: You can do one of two things do change the stimuli that are presented: either alter the stimuli that already exist or create your own. In either case, you should make a copy of the folder containing the SuperLab experiment file and the stimulus files, and work on that. To alter the already existing stimuli, simply find the file containing the picture you'd like to change, open it into a drawing program (such as GraphicConverter, which is on all the Macintosh clones) and have at it. Save the picture as a PICT file without changing the name and you're set. If you change the name you'll have to link this new file to the intended event in SuperLab.
If you'd like to create a new stimulus, you'll need to use a drawing program to create it. GraphicConverter is good for making simple pictures or converting premade pictures (such as ones you find on the World Wide Web) into PICT files, which SuperLab can display. In SuperLab, create a new event in the location where you'd like it to appear. Give the event a descriptive title. Next, select "Open File" from the Picture menu and select the name of your new file. Select how long you'd like it to be presented and that's it. Just make sure it's linked with the appropriate trial.