Social Priming – Does It Work After All?

By Neuroskeptic | October 13, 2016 10:37 am

“Social priming” has been the punching-bag of psychology for the past few years.

untitledThe term “social priming” refers to the idea that subtle cues can exert large, unconscious influences on human behaviour. The classic example of a social priming effect was the “professor priming” study in which volunteers who completed a task in which they had to describe a typical professor, subsequently performed better on a general knowledge task. In other words, as the authors put it, “priming a stereotype or trait leads to complex overt behavior in line with this activated stereotype or trait.”

Although it was a booming research area for many years (that professor priming paper has 756 citations), social priming has lately come under heavy criticism. Many researchers have failed to replicate previously claimed effects. The field’s reputation also suffered when Diederik Stapel, a high-profile social priming researcher, was exposed as a fraud who faked data in dozens of studies.

But now a new paper claims that reports of priming’s death have been greatly exaggerated. It’s called Replicable effects of primes on human behavior, and it’s from psychologists B. Keith Payne and colleagues.

Are Payne et al. really saying that social priming can work, after all? Sort of. The authors don’t like the term “social priming”, and in fact one section of the paper is devoted to explaining why it’s an unhelpful phrase. Payne et al. prefer to describe their work as “behavioral priming”, but they also use this term to cover the kinds of studies that are often called “social priming”. They say that their procedure “captured the defining elements of previous priming studies, in that ostensibly irrelevant primes influenced behavioral responses, and appeared to do so unintentionally”. Overall, the authors imply that their results are relevant to the social priming debate. Whether that’s true is something I’ll address later.

So what did Payne et al. do? They present the results of six experiments. In each study, volunteers performed a simple gambling task: on each trial, two playing cards appeared on the screen, and the volunteer had to choose whether to “bet” or “pass”. They were told that if they bet, their chance of winning would be based on the combined value of the cards, in Blackjack style with face cards scored as 10, aces as 11.

The priming aspect of the study came in form of words which were flashed up on the screen for 300 milliseconds, just after the cards were shown. The words were randomly picked from a pool of six: bet, gamble, or wager, pass, fold, stay. Payne et al. found that merely viewing a word such as “gamble” encouraged volunteers to choose to bet, but it only worked if the cards were mediocre, not so good or so bad that the best response was obvious.

Here’s the results of Experiment 1:


The other experiments replicated the effect in different samples of volunteers – five samples in all – showing very similar results. One experiment (#4) confirmed that the priming effect still worked (albeit weakly) when the priming words were presented subliminally, in a ‘masked‘ presentation lasting just 12 milliseconds. As volunteers could not consciously perceive these masked words, this suggests that the influence of the primes on behaviour was outside of intentional control.

Overall, this is a really interesting series of experiments. I find the results convincing – but I’m not convinced that this study has much relevance to the world of “professor priming” and the like.

One major difference between Payne et al.’s method and classic social priming methods is time: the primes in this study preceded the decision prompts by just 100 milliseconds, but in social priming, it’s not uncommon to see claims of effects that persist for several minutes or more. Rapid priming effects seem intuitively just more plausible than slow ones. The authors acknowledge this point.

But in my opinion there’s a more fundamental difference between this gambling task and most of the “social priming” work: in Payne et al.’s task, the priming effect “makes sense” in a way that social priming doesn’t.

Roughly speaking, we might say that when someone is faced with two equally good options, even a small and irrelevant prompt can make them pick one over the other. To put it another way, if there are no good reasons to pick one option, the brain is susceptible to making the choice based on bad reasons, such as familiarity e.g. the fact that one option was “advertised” 100 milliseconds before. There are ways one could formalize this idea in psychological or neuroscience terms.

My (rather unscientific) point is that the Payne et al. priming effect is a lot easier to believe than professor priming. Why should thinking about some group automatically make us behave like that group? I just can’t see why it should, and it doesn’t fit with my everyday experience. I know I’m not alone in this; social priming just doesn’t “ring true” to many people and, rightly or wrongly, this contributes to the skepticism over the idea.

In summary, this is a lovely paper, but I don’t think it offers much comfort to the social priming field.

ResearchBlogging.orgPayne BK, Brown-Iannuzzi JL, & Loersch C (2016). Replicable effects of primes on human behavior. Journal of Experimental Psychology. General, 145 (10), 1269-1279 PMID: 27690509

CATEGORIZED UNDER: papers, select, Top Posts, Uncategorized
  • OWilson

    It’s an old debate, but effective behaviour modification can take place on many levels.

    From teachers, role models, counselling by experts, peer pressure (I want an Iphone), TV hero worship, and of course advertising.

    In it’s most extreme form it is called brainwashing.

    (It works!)

  • PsyoSkeptic

    Payne’s study here is poor as a scholarly work. It’s no more social priming than the early priming experiments of Neely were. That effect is something very well established across many decades. This attempt to make a small additional generalization of traditional semantic priming into something grand that could save social priming is just more of Payne’s crafty ability to spin wool into gold.

    (it’s also a misapplication of process dissociation procedure)

    • Neuroskeptic

      The study itself is good, but it’s being oversold as something it’s not, in my view.

      • PsyoSkeptic

        I concur. When I say Payne is being a poor scholar I mean he doesn’t properly place the science in the framework of priming or psychological science. He even references some of the correct past work in passing but doesn’t use it correctly. The experiments, while bering designed with a lot of experimenter degrees of freedom, through replication come out being believable enough.

        • WT

          I agree with PsychoSkeptic. What has been lost is a proper understand on “social” priming effects in the context of the priming research summarized by Neely and others. The reason why the Professor priming and other effects were initially seen as so persuasive is because skeptical cognitive psychologists could no longer say to Bargh and others “this is just semantic/associative priming what’s new here?” – the effects generalized to real world phenomena or so it seemed at the time…

  • Gregory Vollmar

    Actually the social priming studies are varied. It is because the “social” is “conducting” in various environment and what kind of atmosphere. It often gives out the poor conclusions from the limited studies. So there is zillions of social activities. It has to be studied at every moment multiplied with various atmospheres and environments as goes for the customs and rituals of cultural activities. For example, playing poker in this very room where Payne’s research occurred, it is limited and expected. What if the same players in an old western saloon? Will they react differently?

  • D Samuel Schwarzkopf

    Great post! At first I thought “I have to read this study” but your post is very comprehensive so you save me the trouble of reading this 😉

    This sounds like a reasonable study. First of all, it’s great to see they used a within-subject design. This has long been one of my main gripes with the social priming (I’ve also been told not to use that term but I was never told what the alternative ought to be…). I can understand why these researchers are a little worried about order effects but this is not an insurmountable problem and I’d at least like to see someone attempt a within-subject design. If these theories were correct, it *should* really work as within-subject design.

    I think you are also completely right that this isn’t the same kind of priming as professors making you smart or words related to the elderly making you walk slowly. The fast priming in this betting task seems a lot more plausible. We *know* that priming exists, certainly in visual search and also semantic priming etc. I would be fairly confident that it also goes beyond such very simple lab designs. As an anecdotal example, yesterday I read this article by one of the women accusing Donald Trump of sexual harassment – in that article the word ‘therapist’ appeared several times. After having read most of the grueling account I actually read that as ‘the rapist’. There has got to be some activation of schemata going on here that is not that dissimilar from the manipulations in many social priming experiments…

    The bit that I disbelieve about social priming is not the basic concept that some stimuli, even subliminal ones, can affect behavior. Rather it is that the effect size of the dependent variables used at extremely implausible. In both of Bargh’s elderly priming experiments the absolute effect size is a difference of 1 second in walking speed and the relative effect size is around d=1. This is simply not realistic. As I always say, humans are dirty test tubes. There are so many factors involved in determining somebody’s walking speed as they leave the lab that this is quite simply not possible.

    • Neuroskeptic

      Good points!

      Regarding the term “social priming”, I take the view that it’s a useful term. We know what it means: we know it when we see it. It’s not a precise term, and it’s probably a misnomer (there’s little “social” about it really) but it serves its purpose in most cases.

  • Jon

    What about stereotype threat? Isn’t this a form of social priming that, as far as I’m aware, is pretty replicable? Recent paper shows performance on tests depends on how the test is framed. Is all social priming research flawed?

    • D Samuel Schwarzkopf

      I think there were some meta-analyses casting doubt on stereotype threat effects. Not too familiar with this though…

      • Neuroskeptic

        Even if stereotype threat is real I don’t think it’s a form of social priming in the sense of “professor priming”. It could be in theory but AFAIK there are more straightforward explanations e.g. that some people get anxious, stressed or demotivated when tests are framed in particular ways, and that stereotypes can contribute to these feelings.

        • D Samuel Schwarzkopf

          Oh yeah I’d certainly agree with that. Stereotype threat is something I find mechanistically much more plausible that professor/elderly/turning-a-wheel priming.

      • Haven Monahan

        This meta-analysis of gender stereotype threat found a small effect, which, however, wasn’t significantly different from zero after trim-and-fill imputation of “missing” studies.

        • Brad Warner

          Importantly, stereotype effects don’t appear to generalize outside of manufactured lab studies. Effects don’t appear in real world high stakes testing situations.

  • Doug Rohrer

    Great column. You are right that there is a large difference between the kinds of social priming phenomena that have received so much attention and the kinds of priming that are well established in the memory field. Here is how David Meyers put it a recent letter to Science:

    “It is important to emphasize that some basic behavior-priming effects are real, robust, and easily replicable even if others are much more problematic. For example, if an English reader is presented with a printed word like DOG, then on average, s/he will be at least 10 to 20% faster at recognizing and responding to a subsequent associated word like CAT when it is presented within a few seconds after the previous word. This psychological phenomenon, called semantic priming, has been demonstrated many times during past decades. …Viewed from a metaphorical perspective, what some social psychologists have done is essentially like trying to show that presenting the printed word DOG may incline English-reading adult male humans more toward visiting remote “cat houses” (slang for brothels) even after substantial amounts of time (several minutes or more) have elapsed since the original exposure to DOG.”

    With less flair, my colleagues and I wrote the following in our recent paper about money priming:

    “The large, robust reported by Vohs et al. (2006) and Caruso et al. (2013) are quite different from the well-established priming effects observed in perceptual and cognitive tasks (e.g., Meyer & Schvaneveldt, 1971). For example, in studies using the lexical decision paradigm, subjects are faster to indicate that a letter string is a word (DOCTOR…“yes”) if the word immediately follows a related word (NURSE) rather than an unrelated word (BREAD). Although these priming effects are reliable, the effect sizes are small (Cohen’s d < 0.20, e.g., Pashler, Coburn, & Harris, 2012). Semantic priming effects are also fleeting. For example, Becker, Moscovitch, Behrmann, and Joordens (1997) found that lexical decision priming effects disappeared if the prime and target were separated by more than 15 seconds. In brief, classic priming effects are small and transient even if the prime and measure are strongly associated (e.g., NURSE-DOCTOR), whereas money priming effects are reportedly large and relatively long-lasting even when the prime and measure are seemingly unrelated (e.g., a sentence related to money and the desire to be alone).

    Doug Rohrer

    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.

    Rohrer, D., Pashler, H., & Harris, C. (2015). Do subtle reminders of money change people’s political views? Journal of Experimental Psychology: General, 144, e73-e85.

  • Pingback: Lectuur op zaterdag: wil je het donkerder, met foutjes, virtueel of met pure magie? | X, Y of Einstein?()

  • Josh

    100-300ms is a long time. This isn’t subtle priming, this is experimenter nudging, i.e. a demand effect.

    • Neuroskeptic

      That was what I thought at first, but it also worked when the primes were masked (albeit barely significant).

      • Josh

        Oops, I missed that. But I am *very* skeptical of that result for 2 reasons: (1) I have played with masking, you need a very special display to get 12ms, and that is way too short, and they expect people to process a word subliminally!? Who do they cite that have shown replicable effects at 12 ms for anything? (2) Sounds like they have big SEs, and low power. Standard p-value inflation issues apply.

  • Pingback: Signal Conduction – 14 October 2016 | Cognitales()

  • Pingback: One of the most controversial ideas in psychology just got a (small) boost : Break The Limit()

  • Pingback: One of the most controversial ideas in psychology just got a (small) boost | Business Insider()

  • joseph2237

    Sounds like they are taking about subliminal suggestions used in early advertising. Made a search and many companies still them.
    Try this for a few laughs, Unbelievable where they are sneaking into their ads.

  • Pingback: One of the most controversial ideas in psychology just got a boost | Business Insider Singapore()

  • Pingback: LOL Something Matters - Fake News Matters()

  • Pingback: What Is "Social Priming"? - Neuroskeptic()

  • Pingback: What Is “Social Priming”? – Gentle Bloom()

  • Pingback: What Is “Social Priming”? – Neuroskeptic – KESIMPULAN()

  • Pingback: Science Finds Once Again That Violent Video Games Don’t Turn People Into Rage Monsters – Professional Interests()

  • Pingback: Science Finds Once Again That Violent Video Games Don’t Turn People Into Rage Monsters | The Hindu Patrika()

  • Pingback: Science Finds Once Again That Violent Video Games Don't Turn People Into Rage Monsters - Joornlist()

  • Pingback: La ciencia prueba (sí, otra vez) que los videojuegos violentos no hacen a las personas violentas()

  • Pingback: La ciencia prueba (sí, otra vez) que los videojuegos violentos no … – Agibilis()

  • Pingback: Science Finds Once Again That Violent Video Games Don’t Turn People Into Rage Monsters – CHEPA website()

  • Pingback: Science Finds Once Again That Violent Video Games Don’t Turn People Into Rage Monsters – Marketfources()

  • Pingback: Science Finds Once Again That Violent Video Games Don’t Turn People Into Rage Monsters – Smiilee()

  • Pingback: Science Finds Once Again That Violent Video Games Don’t Turn People Into Rage Monsters:from Lifehacker | this is what I'm reading()

  • Pingback: La ciencia descubre una vez más que los videojuegos violentos no convierten a las personas en monstruos rabiosos – iCrowdNewswire Spanish()



No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

@Neuro_Skeptic on Twitter


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar