Surely you noticed the furious debate arose around an article by Daryl Bem the suggestive title Feeling the future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect .
The author (who, incidentally, deals with "parapsychology" more as a hobby, as its main field of research, the psychology of sexual orientation) claims to provide evidence supporting it the possibility of precognition.
That way, it seems one of the many esoteric things aggirantisi the border between science and Jacob. But things are a bit 'more complex. First, because the author is not a charlatan by trade, but cha has a respectable history of science. Then for the newspaper, which with its impact factor of 5 + and a recognized authority, is not certainly the place where you think you can find quackery. And finally, perhaps most importantly, the idea behind the study.
One of the most common criticism of studies on psi (we use this term, which is common in the scientific, rather than the slightly derogatory term for "parapsychology") is to use statistical methods and experimental paradigms uncommon, or extremely complex with the consequence of being fairly reproducible. See about the history of the Ganzfeld experiments .
The strategy followed by Bem is exactly opposed. Bem has chosen nine classic paradigms of modern psychology, there was slightly modified (I explain after how) and analyzed them using standard techniques, offers a full range of applicants for the software used to perform the experiments.
Bem wants to study the phenomenon that is called "priming." That is how much influence they have on the choices of an individual signal preconscious (subliminal). The basic experiment (repeated a thousand times by psychologists of any kind) works like this: a person is faced with two options (usually in front of a computer, two windows, one right and one left) and must make a choice. Before you make the choice shows a picture for a very short time (that does not reach the consciousness of the subject) behind one of the windows. If the image is of a certain type (images erotic work very well), the probability that the subliminal image window is chosen is higher, and this thing can be detected statistically by repeating the experiment several times.
This experiment was changed by Bem in this way: the subliminal image was displayed after the subject had made the choice, but in fact the experiments he carried out are 9, but they are all more or less variants of this experiment classic. He found that in 8 out of 9 experiments, the image shown after he had a subliminal effect on the choice before easing to a level that would be considered statistically the most convincing scholars.
Some objections come immediately to mind:
1) Often these experiments are affected by the fact that the investigator interacts directly with the subject. To eliminate this problem, the experiment was designed such that the instructions were a computer.
2) Obviously, the number of subjects is important. Bem has used approximately 100 subjects per experiment, a fairly convincing.
3) It is also important as the choice of computer display of the window where the image is randomly subliminal. Bem has chosen to use as the random number generator CD Marsaglia. If I remember correctly, in an experiment has even used a random number generator true.
4) The problem of sub-groups: among other things, Bem shows that what he calls "psychological performance", ie the efficiency with which subjects guessed their prediction, correlates with certain personality traits commonly associated with capacity psi: an open mind, be outgoing, etc ... In some forums, Bem has been accused of forming subgroups of his sample to build its evidence. I do not think so: the results are independent of these subgroups. Only then, and additionally, he shows that this "psychological performance" correlates with other factors.
5) Some mathematical Dutch have polemicized with the type of statistical tests used by Bem, saying that it is necessary to use more refined tests. To prevent this objection, Bem has used statistical tests considered standard in the field of psychology. To the critics of Wagenmakers et al. is a bit 'double-edged sword, because, if correct, would invalidate more or less all the research in the social sciences (not only) made in the last 50 years.
0 comments:
Post a Comment