Copyright © 2007-2018 Russ Dewey
Priming Effects in Social Cognition
Priming research was originally conducted with words. Meyer and Schvaneveldt (1971) showed that people were quicker to identify a word when exposed to a related word earlier.
Suppose the experimental task is to decide if a string of letters appearing on the screen in front of you is a word or a non-word. The participant's instructions are to press button A if it is a word, button B if it is a non-word.
A person seeing the word APPLE will be slightly faster, a moment later, to identify TREE as a word rather than a non-word. The reduction in reaction time can be used as a measure of how closely the two words are associated.
APPLE, used as a prime, would result in a big reduction in reaction time for TREE. So the two would be judged as closely related in a semantic (word meaning) network such as the one below.
In a diagram like this, less distance means faster reaction times, if one word is used to prime another.
Another form of measurement involves degraded stimuli. A participant might see an unrecognizable blur on a screen.
Every second a few more pixels are filled in, until the word is clearly visible. The experimental task is to press a button as soon as the participant can recognize the word (and then report it, of course).
Example of a degraded stimulus
If you have been primed with FIRE you will be quicker to recognize the word SPARK as it gradually appears, pixel by pixel. You will also be more likely to use words related to fire in speech, or volunteer words related to fire or flames or burning as answers to a puzzle, or (probably) draw a picture with fire in it.
This is predictable because priming is a warm-up effect. Whatever you see or hear (or smell, or remember) things related to that will be warmed up. Those ingredients will be readily available if they must be used in a cognitive production. Priming warms up components for use in the cognitive system.
Priming effects expose what psychologists have always called associations. What is an association? It is something that brings to mind something else. It is a linkage between concepts, emotions, or motor responses. In any event, it is a linkage between cognitive elements.
Reaction times give evidence about closeness of associations. Closeness is a metaphor for a strong linkage. A close association might exist between a motor routine (at the back of the brain) and a word (headquartered in the left temporal area). Calling it a close association is the same as calling it a strong connection.
Closeness is strength of association or similarity in meaning between semantic elements. Reaction times are faster when things "go together" because they are already associated in the cognitive system.
Here is a simple example. Schvaneveldt and Meyer (1973) had participants indicate, as quickly as possible, if two stimuli were legitimate English words (like KNIFE-QUITE) or non-words (like SMUKE-REABL).
Subjects averaged about 85 milliseconds faster recognizing words that were closely associated (BREAD-BUTTER or NURSE-DOCTOR) as compared to the same words in un-associated pairs (BREAD-DOCTOR or NURSE-BUTTER). Reaction times revealed which items were closely associated.
Why were the reaction times faster for closely associated words? Things that "go together" (have previously-established associations) are easier to put together in a cognitive act. This seems to be what we mean by things being associated or related or close in meaning (nearby in semantic space).
Reaction time itself has a long history as a dependent variable in psychology. It was used in the first experimental psychology laboratory was opened in 1979 by Wilhelm Wundt. When cognitive psychology was revived in the mid-20th Century, reaction time was recognized as an ideal variable for research.
Time is measured using what statisticians call a ratio scale: a measurement scale with a true zero. This allows researchers to use powerful statistics.
Time is also easy to measure. The stopwatch was invented in 1776, a century before Wundt. Time could be measured with precision.
The combination of precision, ease, and an underlying ratio scale made reaction time an extremely useful measure in psychology. Michael Posner, a noted cognitive scientist, wrote a whole book about it: Chronometric Explorations of Mind (1978).
In the mid-1990s social psychologists realized priming effects could be used to study associations between attitudes as well as words. If you are primed with the word GOOD you might be faster to recognize the word BEER, but only if you like beer.
If beer was associated with misery in your life, or you were proud of abstaining from alcoholic beverages, GOOD would not prime the word BEER.
Priming effects like this are short-acting, but they are reliable and easy to measure. Reaction times depend on the fast processing system, so they are not influenced much by conscious intentions.
The Implicit Association Test (IAT)
The Implicit Association Test (IAT) became the most widely used test of implicit social cognition after its invention in the mid-1990s. Implicit here means about the same thing as unconscious or out of awareness or automatic. Implicit social cognition is about relatively uncontrolled, automatic responses in social cognition.
Why is The IAT so popular? It solves a number of important problems faced by researchers in social cognition. The Implicit Association Test requires no introspection. Even if a person cannot look within to describe an attitude, the IAT can reveal it, using reaction times.
The IAT is simple. The participant need only hit two buttons, usually two keys on a keyboard, while comparing words. The results are available immediately.
You can take Implicit Association Tests online to see what the test is like from a participant's perspective. Here is an outline of a typical IAT session, based on an example used by Nosek, Greenwald, and Banaji (2007).
First you are asked to rate a list of words as pleasant or unpleasant. Your instructions are to push the button on the right for pleasant words, the button on the left for unpleasant words. The words are easy to evaluate, like "wonderful" (push the button on the right) and "terrible" (push the button on the left).
Next, if we are interested in your reaction to gender, you might be asked to react to a series of male and female faces. You press one button for a female face, the other button for a male face. This goes on for a few minutes.
In the next phase of the experiment, you evaluate two dimensions at the same time. If the stimulus is either male or unpleasant, press the right button. If the stimulus is either female or pleasant, press the left button.
Then the dimensions are switched around. You press the right button if a stimulus is either a male face or a pleasant word and press the left button if a stimulus is either a female face or an unpleasant word.
Depending on which version of the Implicit Association Test you are taking, this might go on for 5 or 10 more lists. The experimenters are interested in your attitude toward males and females, so after the experiment finishes they will compare your reaction times for the conditions where you must detect male-or-pleasant versus male-or-unpleasant, and female-or-pleasant versus female-or-pleasant.
Usually there are big differences between reaction times in the different conditions. Assume for a moment you are somebody who finds females more pleasant than males. It will be easy for you to press the button quickly when looking for either a female face or a pleasant word. That is like judging whether BREAD-BUTTER are both words. They go together, so the judgment is swift.
However, if that same person is looking for "either a male face or a pleasant word" reaction times will be longer. The subject will see a face or a word and have to decide, "Am I looking at either a male face or a pleasant word?" before pushing the button. That slows reaction times.
Why does this happen? If males evoke unpleasant emotions, deep down, then instructions to look for "either male or pleasant" stimuli are pitting two different impulses against each other. This slows people down.
On the other hand (assuming a person prefers females) female faces and pleasant words might just "go together" like bread and butter. It feels like responding to one stimulus instead of two, so the decision is fast. Reaction times will be quick.
As soon as the IAT came out, social psychologists saw the potential in it. Here was a test that could test "forbidden" attitudes like sexism and racism. A person could be asked to separate names into African-American or Caucasian sounding categories, then separate words into pleasant and unpleasant categories.
A few minutes later, the participant is being asked to press one button if they see an African-American name OR a pleasant word. They press the other button if they see a Caucasian name OR an unpleasant word.
A person with an immediate negative reaction to African-American names (even if they consider themselves non-
Are the results of the IAT valid? Judging from many, many attempts to validate the test, it does indeed reflect attitudes that influence behavior. In fact, it is one of the most revealing tests ever invented for revealing prejudices and stereotypes.
McConnell and Leibold (2001) compared IAT results concerning racism with actual behavior of participants toward a black or white experimenter. Trained judges and the experimenters rated subjects for prejudicial behavior as they interacted with the experimenter before and after the test. The ratings correlated with attitudes revealed by the IAT test.
Goldring (2011) reported that over 700 published studies used the test. It "appears to do what it claims to do: it is resistant to social presentation concerns." It is "not easily 'fake-able.'" Most importantly, it predicts behavior across a range of domains.
The IAT is used most often to probe attitudes or traits a person might not want to admit. An example is alcohol identity probe we alluded to earlier. This is "the extent to which an individual perceives drinking alcohol to be a defining characteristic of his or her self-identity" (Gray, LaPlante, Bannon, Ambady, and Shaffer, 2011).
Not only does a specially designed IAT detect alcohol identity; it predicts behaviors (such as who will consume more alcohol after stress) that few other measures, including self-reports, can predict. That is what makes the IAT so special; it does not require introspection, it overcomes biased self-reports, and it produces robust and easy-to-measure effects that have predictive validity.
The IAT is useful not only when a person might be unwilling to disclose an attitude (such as liking the same sex). It can reveal attitudes a person is unable to disclose an attitude, because it is not conscious.
People might be unable to tell you what gender they find more trustworthy, male or female. Perhaps they never thought about it, or it is not obvious to them, or they claim to have no opinion. But a clear preference might be disclosed by the Implicit Association Test.
The test avoids self-presentation artifacts: disturbances in the data due to a person's awareness they are being evaluated (Greenwald, Nosek, and Banaji, 2003). The IAT can reveal things that a person does not realize ("lack of introspective access" does not stop it).
Finally, the IAT is easily adapted to a broad variety of situations. A researcher merely sets up a few lists of words, and it is ready to go. The results are easy to analyze.
All of this means the IAT test is well suited to online demonstration, and the creators of it realized this. They set up an online version of the IAT that became very popular.
An Implicit Self?
The IAT can detect attitudes a person is either unable or unwilling to disclose. "Unable" would apply to attitudes that are unconscious, including attitudes toward the self.
Some researchers think the IAR can reveal an implicit self quite different from the public self. For example, a person present a honest, trustworthy persona. But the IAT might show that "down deep" the person associates dishonesty and deceit more closely to self-related words.
Greenwald and Farmham (2000) wrote:
The distinction between explicit and implicit operation of the self is especially interesting if it turns out that the self functions differently in these two modes. Accordingly, it is useful to be able to measure self-esteem and self-concept in ways that can distinguish the self's implicit and explicit operations.
By saying, "the self functions differently in these two modes," Greenwald and Farnham are perhaps mirroring Freud's suggestion that the conscious part of the mind protects itself against challenges to its own image.
Freud said defense mechanisms protect "pretenses of the ego." Pretenses are things we pretend about ourselves. But they cannot fool the IAT.
100 years before the IAT, and before he met Freud, Carl Jung realized that his word association technique revealed what Jung called psychic lesions and complexes. He instructed his patients to respond immediately to each word in the free association test, giving the first response that came to mind. But sometimes that did not happen.
Jung's patients might choke or go silent in response to certain significant words, or come out with odd associations. Jung felt this pointed to a mental conflict. The same basic logic underlies the IAT test. Conflicting associations, things that "do not go together," are slower to retrieve.
The IAT is one huge tradition of research in social cognition. It took advantage of the fact that word priming is reliable and easily measured and takes place before people consciously influence the process.
A different type of priming became widely used in social psychology research. This goes under the labels of social priming, goal priming, behavior priming, or prime-to-behavior effects. When people in social psychology use the isolated word priming, this is the type they are usually talking about.
In social or behavioral priming, a priming stimulus affects later behavior in some way. We saw an example of social or behavioral priming earlier in this chapter, on the page about helpful behavior.
Bryan and Test (1967) showed that if people drove by a scene in which a man was changing a woman's tire, they were more likely to help a woman standing next to a car with a flat tire, a quarter mile further down the road. This was called "modeling" in the Bryan and Test research, but it was the same effect as behavioral priming.
The priming stimulus in the Byron and Test research was incidental. It was plainly visible, but nobody called it to people's attention. (Nobody was riding down the freeway in L.A. with random people to make sure they saw the helpful behavior).
Psychologists just counted the percentage of motorists who stopped when the model was there, or not. The experiment worked; it showed people were more likely to give aid when they passed by an example of somebody being helpful, a quarter mile earlier. At freeway speed that would have been about 15 seconds earlier.
John Bargh demonstrated behavioral priming effects in over 200 laboratory experiments over 30 years. In most cases, participants were unaware they were being primed. The priming stimuli were incidental. The stimuli were easy to perceive, but they were not called to the person's attention.
For example, Bargh, Gollwitzer, Lee-Chai, Bardollar, and Trotschel (2001) had students do word search puzzles that contained either achievement words (win, success) or control words (building, staple). The students who looked for achievement words performed better on subsequent search puzzles, as if their achievement motivation had been activated the seeing words related to achievement. (Spoiler alert: this research suffered a replication failure over ten years later.)
A classic social priming study (also the most admired, ridiculed, and criticized) came from Bargh, Chen, and Burrows (1996). They described three experiments in the original report, but the second one got all the attention. It became famous as the "old person" or "elderly stereotype" study.
The research was intended to show how activating a stereotype could influence behavior. Participants worked at a scrambled sentence task, which they were told was part of a language proficiency experiment.
The scrambled sentences (for one group) contained words relevant to stereotypes of elder people. The scrambled sentences contained words intended to evoke an elderly person stereotype: Florida, old, lonely, grey, selfishly, careful, sentimental, wise, stubborn, courteous, bingo, withdraw, forgetful, retired, wrinkle, rigid, traditional, bitter, obedient, conservative, knits, dependent, ancient, helpless, gullible, cautious, and alone.
After completing the task, participants walked down a corridor. A second experimenter using a hidden camera recorded the amount of time a participant took to walk down the corridor. "The main hypothesis was that participants who had been primed with the elderly stereotype would walk more slowly." That was indeed what the experimenters observed.
"No participant expressed any knowledge of the relevance of the words in the scrambled sentence task to the elderly stereotype." So the priming effect was unconscious. The Bargh, Chen, and Burrows research impressed people. In the years that followed, many similar studies showed subtle effects of behavioral priming.
Can priming effects be overridden by conscious thinking? We might have reason to think so. The gist of the ELM theory (Elaboration Likelihood Model) was that conscious thinking overrides automatic emotional responses to advertisements and similar attempts at persuasion.
Webb, Sheeran, Gollwitzer, and Trotschel (2012) tested people's ability to consciously override a priming effect. They exploited a famous, previously-verified effect called the Supermodel Effect.
Showing people pictures of supermodels led participants to produce lower scores on the Wechsler Adult Intelligence Scale (WAIS). If they were shown pictures of professors instead, they scored a higher intelligence rating.
To override that effect, one group was told to form this goal: "If I see a difficult question, then I will use all of the knowledge at my disposal to answer it!" This eliminated the supermodels vs. professors priming effect. This seemed consistent with earlier research showing that slower, cognitive processes could override quick judgments based on stereotypes.
The 2012 Priming Brouhaha
In 2012, social priming research was going strong. One scholar said there were over 10,000 published studies on social priming between 2000 and 2010. I do not know if that is literally true, but it is plausible.
But there were clouds on the horizon. The year before, two Dutch researchers (Diederik Stapel and Dirk Smeesters) were found to have falsified social priming experiments, in separate investigations.
Stapel ended up retracting 58 papers. Dirk Smeesters retracted 7 papers. In both cases, most were social priming studies.
This led to heightened scrutiny of social priming research. One immediate impulse was to check older studies that seemed amazing or unlikely. Were they genuine? Could they be replicated?
In 2012, one of the most famous examples, the elderly stereotype priming study from Bargh, Chen, and Burrows (1996), ran into replication problems. This led to a "brouhaha" over social priming effects (Srivastava, 2012). (A "brouhaha" is "a noisy and overexcited reaction or response.")
By itself, a replication failure is not surprising, especially in an area of research with tens of thousands of experiments. However, the "old person" priming study was up to that point the single most famous example of behavioral priming. (One author opened a social cognition textbook with the elderly stereotype priming study.)
The failed replication came from Doyen, Klien, Pichon, and Cleeremans (2012). They were trying to replicate the elderly stereotype study of Bargh, Chen, and Burrows (1996).
In order to rule out experimenter effects, they made a crucial change. The experimenters (psychology students) were not told what to expect as an outcome for the study.
The replication failed. Doyen et al. ran two groups the next time. In one, student experimenters knew the expected outcome of priming. In the other, they did not. When students expected it, the "old person" effect came back.
This strongly suggested the original result might have been due to an experimenter effect. The influences of expectancy are powerful. Perhaps they were at work in the original Bargh research, undermining the meaningfulness of the experiment.
Bargh wrote a rebuttal to the Doyen et al. research group titled, "Priming effects replicate just fine, thanks." In it he cited two partial replications of the original study and a meta-analysis of "25 years of priming research" showing "statistically reliable effect sizes."
Bargh pointed out a key difference between the original study and the attempted replication. In the original study, not one of the participants noticed the elderly person primes.
In the replication Doyen et al. used many more elderly stereotype words, and their subjects apparently noticed. "Debriefing was suggestive of awareness of the primes," Doyen et al. wrote.
That is important. Awareness can negate the effect of priming. In fact, several other studies showed that obvious primes, noticed by subjects, had no effect.
This is an example of a now familiar theme. Slower, conscious cognitive processes can resist or overcome faster, stereotype-based responses.
Perhaps that happened in the Doyen et al. replication attempt. If participants noticed they were being primed with numerous old-age stereotypical words, that might have eliminated the effect.
In retrospect, Doyen et al. should have used the exact same lists as Bargh, Chen, and Burrows, to make the replication precise. Perhaps their participants would not have noticed they were being primed.
Meanwhile, Bargh's defensive reaction, and his claim that the elderly stereotype had already been replicated, drew a reaction of its own. Bargh cited two existing replications of the elderly person experiment, arguing that it "replicates just fine, thanks."
Statistician Andrew Gelman (2016) pointed out that both replications cited by Bargh actually failed to replicate the original effect, except when it interacted with other variables. Those interactions with other variables had not been hypothesized ahead of time.
The interactions were found after data was collected. This reinforced Gelman's suspicion that "the original finding was a classic garden-of-forking-paths power=.06 story of opportunistic data analysis."
"Garden-of-forking-paths" is a statistical fallacy that occurs when researchers gather data first, then look for multiple relationships in the data. As they go along, they perform different sorts of statistical analyses, depending on hints in the data.
For example, "We failed to replicate the elderly stereotype effect, but Hey, look at this, if you take the variable Need for Cognitive Control into account, subjects highest in that variable did show the effect." Both of the replications Bargh cited were like that.
If an experimenter did not originally propose an interaction with Need for Cognitive Control (or whatever) then finding it after-the-fact does not count as a replication of the original effect. That was Gelman's point. To those who understood the statistical logic involved, it was devastating.
The upshot was that no pure replication of the elderly stereotype effect existed. Meanwhile. another psychologist, Daniel Simons, called Bargh's irritated response "a case study of what NOT to do when someone fails to replicate one of your findings." It inflamed the controversy.
Simons suggested that it would have been better to admit "the original research might be a false positive." Bargh could also have discussed the ways a replication could fail (such as random variations and small design differences).
Bargh probably did put his finger on the reason the Doyen et al. replication failed. There were too many "elderly person" priming words, and subjects became aware of the manipulation.
This valid point was lost in the brouhaha over Bargh's criticism of the researchers, plus Bargh's gratuitous and inaccurate criticism of the online journal (PLoS ONE) where their results were posted (he dismissed it as not being a peer-
Before long, another research group, Pashler, Coburn, and Harris (2012), failed to replicate another Bargh study. (Keep in mind Bargh published over 200 studies on social priming over more than 30 years, so it is not surprising some of his findings cannot be replicated.) The Pashler, Coburn, and Harris (2012) study attempted to replicate a finding by Williams and Bargh (2008).
Williams and Bargh had reported that if participants plotted two points on graph paper, far away from each other, those subjects were more likely to say later they were "distant" from their relatives, compared to subjects who plotted two points close together. Pashler et al. tried to replicate that effect and failed.
"The results showed no hint of the priming effects reported by Williams and Bargh (2008)," Pashler, Coburn, and Harris reported. Unknown to Pashler et al., another team of experimenters was replicating the same Williams and Bargh study, at the same time in late 2012.
This replication attempt was large, running 124 subjects. They found "neither significant results nor any trends in line with the original Williams and Bargh results."
Then, in 2013, came a third blow: this time involving the "achievement related words" study above (originally published as Bargh, Gollwitzer, Lee-Chai, Bardollar, and Trotschel, 2001). This was the experiment where subjects solved word puzzles containing achievement words.
Participants exposed to achievement words were faster than other subjects to solve additional puzzles. The original interpretation was that their achievement motivation had been primed.
Harris, Coburn, Rohrer, and Pashler (2013) attempted to replicate that effect, but got no effect. In fact, their results went in the opposite of the predicted direction.
Bargh suggested a modification in their procedures: a wait of 5 minutes between the first set of word puzzles and the next group. Still there was no effect.
Behavioral priming was becoming harder to replicate over time, a bad sign. This controversy received a lot of publicity among researchers studying social cognition in 2012 and afterward.
This controversy had some healthy effects. It stimulated discussion about two crucial issues:
1) Replication: the necessity of replicating surprising findings (and controlling for experimenter effects in the process)
2) Data collection: the necessity of specifying ahead of time what hypothesis is being tested, how many participants will be involved, how much data will be collected, and how it will be analyzed.
In response to the first issue, social cognition researchers started putting more emphasis on replications. To counteract file drawer effects, a web site was started to publicize replication attempts whether or not they succeeded. Several journals announced their intention to publish negative results as well as positive results, if experiments were well constructed.
The Association for Psychological Science (APS) issued new guidelines to researchers in 2016 suggesting that, before they collect any data, researchers should record their intentions for data analysis in a secure data repository. This information should include the number of subjects they intended to run, when they would terminate data collection, and how they would analyze data.
If experimenters followed these guidelines, that would prevent Gelman's "garden of forking paths" problem, where data is sliced and diced after data collection is over to find a significant effect. By the new APS guidelines, this could not happen, because the statistical tests would be specified beforehand.
These cautions are fairly complex and high-level; most introductory students would not be expected to grasp them. The important take-home message for intro students is that the "brouhaha" over replication failures had a good effect, leading to reactions that should strengthen research in the future.
Bargh, J. A. (2012, May 11) Priming effects replicate just fine, thanks. Psychology Today. Retrieved from: https://www.psychologytoday.com/
Bargh, J. A., Chen, M., & Burrows, L. (1996) Automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psyhcology, 71, 230-244.
Bargh, J. A., Gollwitzer, P. M., Lee-Chai, A., Barndollar, K., & Trötschel, R. (2001). The automated will: Nonconscious activation and pursuit of behavioral goals. Journal of Personality and Social Psychology, 81, 1014-1027. https://dx.doi.org/10.1037/0022-35184.108.40.2064
Doyen, S., Klein, O., Pichon, C-L., & Cleeremans, A. (2012) Behavioral priming: It's all in the mind, but whose mind? PLoS ONE, 7, e29081. doi:10.1371/journal.pone.0029081
Gelman, A. (2016) Priming effects replicate just fine, thanks. Statistical modeling, causal inference, and social science. [blog] Retrieved from: https://andrewgelman.com/
Goldring, J. (2011) The development of an Implicit Association Test for measuring forgiveness. University of Adelaide. PhD theis. Retrieved from: https://digital.library.adelaide.edu.au/
Gray, H. M., LaPlante, D. A., Bannon, B. L., Ambady, N., & Shaffer, H. J. (2011) Development and validation of the Alcohol Identity Implicit Associations Test (Al-IAT). Addictive Behaviors, 36, 919-926.
Greenwald, A. G. & Farnham, S. D. (2000) Using the Implicit Association Test to measure self-esteem and self-concept. Journal of Personality and Social Psychology, 79, 1022-1038. doi:10.1037/KK122-35220.127.116.112
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464-1480. doi:10.1037/
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the implicit association test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197-216.
Greenwald, A. G. & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27. doi:10.1037/0033-295x.102.1.4
Harris, C. R., Coburn, N., Rohrer, D., & Pashler, H. (2013) Two failures to replicate high-performance-goal priming effects. PLoS ONE, 8. doi:10.1371/journal.pone.0072467
McConnell, A. R. & Leibold, J. M. (2001) Relations among the Implicit Association Test, discriminatory behavior, and explicit measures of racial attitudes. Journal of Experimental Social Psychology, 37, 435-442. doi:10.1006/
Meyer, D. E. & Schvaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence of a dependence between retrieval operations. Journal of Experimental Psychology, 90, 227-234. doi:10.1037/h0031564
Molden, D. C. (2014) Understanding priming effects in social psychology: An overview and integration. Social Cognition, 32, Special Issue, pp. 243-249
Pashler, H., Coburn, N., & Harris, C. R. (2012) Priming of Social Distance? Failure to Replicate Effects on Social and Food Judgments. PLoS ONE, 7. doi:10.1371/journal.pone.0042510
Posner, M. I. (1978). Chronometric Explorations of Mind. Hillsdale, NJ: Lawrence Erlbaum Associates.
Schvaneveldt, R. W. & Meyer, D. E. (1973). Retrieval and comparison processes in semantic memory. In S. Kornblum (Ed.), Attention and performance IV, New York: Academic Press. (Pp. 395-409).
Segal, S. J. & Cofer, C. N. (1960) The effect of recency and recall on word association. American Psychologist, 15, 451.
Srivastava, S. (2012, March 12) Some reflections on the Bargh-Doyen elderly walking priming brouhaha. The Hardest Science. [blog] Retrieved from: https://hardsci.wordpress.com/
Webb, T. L., Sheeran, P., Gollwitzer, P. M., & Trötschel, R. (2012). Strategic control over the unhelpful effects of primed social categories and goals. Journal of Psychology, 220, 187-193.
Williams, L. E. & Bargh, J. A. (2008) Keeping one's distance: The influence of spatial distance cues on affect and evaluation. Psychological Science, 19, 302-308.
Yong, E. (2012) A failed replication draws a scathing personal attack from a psychology professor. Not Exactly Rocket Science blog at Discover Magazine. Retrieved from: https://blogs.discovermagazine.com/
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151-175.
Zajonc, R. B., Pietromonaco, P., & Bargh, J. A. (1982). Independence and interaction of affect and cognition. In M. S. Clark & S. T. Fiske (Eds.), Affect and cognition: The seventeenth annual Carnegie symposium on cognition. (Pp. 211-227). Hillsdale, NJ: Lawrence Erlbaum.
Write to Dr. Dewey at firstname.lastname@example.org.