This study has now finished and we are no longer in need of participants. The study was a replication of research conducted by Daryl Bem at Cornell University in 2010. A prepublication copy of the original paper is available (hosted by Bem at http://dbem.ws), while related articles appeared in New Scientist magazine, Wired magazine and many other publications. A clear and concise summary is available from the British Psychological Society Research Digest (republished below) and Prof. Richard Wiseman from the University of Hertfordshire has produced a list of the replications of which he is aware. The New Scientist also published an article in response to the decision by the Journal of Personality and Social Psychology which declined to publish the non-significant replication under their Editorial policy.
Perhaps there's something in the drinking water at Cornell University. A
new study involving hundreds of Cornell undergrads has provided a
dramatic demonstration of numerous 'retroactive' psi effects - that is,
phenomena that are inexplicable according to current scientific
Rather than having the students read each others' minds or wear sliced ping-pong balls over their eyes, Daryl Bem has taken the unusual, yet elegantly simple, approach of testing a raft of classic psychological phenomena, backwards.
Take priming - the effect whereby a subliminal (i.e. too fast for conscious detection) presentation of a word or concept speeds subsequent reaction times for recognition of a related stimulus. Bem turned this around by having participants categorise pictures as negative or positive and then presenting them subliminally with a negative or positive word. That is, the primes came afterwards. Students were quicker, by an average of 16.5ms, to categorise negative pictures as negative when they were followed by a negative subliminal word (e.g. 'threatening'), almost as if that word were acting as a prime working backwards in time.
If psi abilities have really evolved, it makes sense that they should confer survival advantages by helping us find mates and avoid danger. In another experiment Bem had dozens of undergrads guess which set of curtains in a pair on a computer screen was concealing an erotic picture. Participants were accurate on 53.1 per cent of trials, compared with the 50 per cent accuracy you'd expect if they were simply guessing. This accuracy was increased to 57 per cent among students who scored higher on a measure of thrill-seeking. By contrast, no such psi effects were observed for neutral stimuli.
In another experiment participants looked at successive pairs of neutral mirror images and chose their favourite - the left or right. After each pair, an unpleasant picture was flashed subliminally on one side or the other. You guessed it, participants tended to favour the mirror image on the side of the screen opposite to where an unpleasant picture was about to appear.
The examples keep coming. The mere exposure effect is when subliminal presentation of a particular object, word or symbol causes us to favour that target afterwards. Bem turned this backwards so that participants chose between pairs of negative pictures, and then just one of them was flashed subliminally several times. Female participants tended to favour the negative images that went on to be flashed subliminally, as if the mere exposure effect were working backwards through time.
This backward mere exposure effect didn't work for male undergrads, perhaps because the images weren't arousing enough, so Bem replicated the experiment using more extreme negative images and erotic images. This time a 'backwards' mere exposure effect was found with men for unpleasant images. For positive imagery, mere exposure traditionally has a negative effect, as the stimuli are made to become more boring. Bem showed this effect could also happen from the future. Presented with pairs of erotic images, male undergrads showed less favour for the images that went on to be flashed subliminally multiple times. It's as if the participants knew which images were going to become boring before they had.
Finally, we all know that practice improves learning. Bem tested students' memory for word lists and then had them engage in extensive practice (e.g. typing out) for some of the words but not others. His finding? That memory performance was superior for words that the students went on to practice afterwards - a kind of reverse learning effect whereby your memory is improved now based on study you do later.
These reverse effects seem bizarre but they are backed up by some rigorous methodology. For example, Bem used two types of randomisation for the stimuli - one that's based on computer algorithms, which produce a kind of pseudo-randomisation in the sense that a given distribution of stimuli is decided in advance. And another form of randomisation based on hardware that produces true randomisation that unfolds over time as an experiment plays out. Also throughout his paper, Bem uses multiple forms of simple statistical test and he reports results for each, thus demonstrating that he hasn't simply cherry picked the approach that produces the right result. Across all nine experiments the mean effect size for the psi effects was 0.22 - this is small, but noteworthy given the nature of the results.
So what's going on? Bem doesn't proffer too many answers although he argues that his psi phenomena vary with subject variables, just like mainstream psychological effects do. For example, the phenomena were nearly always exaggerated in the more extravert, thrill-seeking participants. From a physics perspective, he believes the explanations may lie in quantum effects. 'Those who follow contemporary developments in modern physics ... will be aware that several features of quantum phenomena are themselves incompatible with our everyday conception of physical reality,' Bem argues. 'Many psi researchers see sufficiently compelling parallels between these phenomena and characteristics of psi to warrant considering them as potential candidates for theories of psi.'
It was one of last year's most astonishing scientific stories: a leading psychology journal accepted a paper presenting evidence for precognition – an ability to perceive future events. What's more, mainstream psychologists had pored over a preprint of the paper and found no fatal flaw.
Bold scientific claims need to be replicated before gaining widespread acceptance, however, and now the Journal of Personality and Social Psychology, which published the paper in its March 2011 issue, has touched off controversy by rejecting the first attempts to repeat the work without sending them out for peer review.
The incident exposes a problem that may be biasing the entire body of psychological literature, argues Richard Wiseman of the University of Hertfordshire, UK, one of the authors of a manuscript that describes three failed attempts to repeat an experiment in the original paper. If failed replications languish unpublished, he says, "you don't know whether the effects that are published are genuine. It's a problem in psychology, and it's a particular problem in parapsychology."
The original paper described nine experiments conducted over eight years by Daryl Bem, a social psychologist at Cornell University in Ithaca, New York, with a long-standing interest in parapsychology. Bem's strategy was to take well-established psychological phenomena and reverse the sequence of events, so that the "cause" happened after the "effect", rather than before.
When news emerged that the Journal of Personality and Social Psychology had vetted the work and decided to publish it, believers in the paranormal were delighted, sceptics perplexed, and bystanders fascinated.
New Scientist's initial story about the paper was among the most widely read articles we published online last year. Bem ended up a minor celebrity, being interviewed on Comedy Central TV's The Colbert Report – where host Stephen Colbert homed in on experiments into "time-travelling porn", in which volunteers seemed to anticipate the position where erotic images would appear on a computer screen.
Frivolity aside, confirmation of Bem's findings would turn established ideas about time, cause and effect on their head. "We openly admit that the reported findings conflict with our own beliefs about causality and that we find them extremely puzzling," the Journal of Personality and Social Psychology's editors said in an accompanying comment.
Other researchers began to try to repeat the results as soon as news of Bem's findings began to spread. Wiseman set up a registry of such attempts, which has so far documented five.
Three of these were described in a paper from Wiseman, Christopher French of Goldsmiths, University of London, and Stuart Ritchie of the University of Edinburgh, UK, which was sent to the Journal of Personality and Social Psychology.
Each tried to repeat an experiment of Bem's which was based on a well-tested experiment for memory priming. In Bem's back-to-front version, participants were shown a list of words and then asked to recall words from it. Later Bem showed them words randomly selected from the same list, and it turned out that they had been better at recalling these words in the prior test. The subsequent display seemed to have influenced their earlier memory.
In the conventional psychological experiment on which Bem's experiment was based, people are shown particular words, and then are given a list of words that include the ones they have previously experienced. The participants are next asked to recall as many words as possible from the list, allowing the experimenters to quantify the effect of the prior priming on the recall of those words.
In contrast to Bem's results, Wiseman, French and Ritchie failed to find that the subsequent typing facilitated the volunteers' earlier recall. But Eliot Smith of Indiana University in Bloomington, the Journal of Personality and Social Psychology editor who handled the submitted paper, declined to send it out to review. "This journal does not publish replication studies, whether successful or unsuccessful," he wrote.
Smith defends the decision, noting that he made the same ruling on another paper that, by contrast, supported Bem's findings. "We don't want to be the Journal of Bem Replication," he says, pointing out that other high-profile journals have similar policies of publishing only the best original research.
"I certainly agree that it's desirable that replications are published," Smith told New Scientist. "The question is where. There are hundreds of journals in psychology."
Bem stressed the importance of replication in his original paper. However, he argues that firm answers will come only when it is possible to conduct a meta-analysis of multiple attempts at replication. "I understand the journal's position," he says. "It almost never publishes a single study."
Wiseman is unconvinced, however, arguing that a meta-analysis may miss the whole picture if journals are reluctant to publish replication studies: "My feeling is that the whole system is out of date and comes from a time when journal space was limited." He argues that journals could publish only abstracts of replication studies in print, and provide the full manuscript online.
Content last modified: 09 Aug 2011
Goldsmiths, University of London, New Cross, London, SE14 6NW, UK
Telephone: + 44 (0)20 7919 7171
Goldsmiths has charitable status