LFR precognition experiment
- Thread starter T'ai Chi
- Start date
-
- Tags
- experiment precognition
Link to the website instead of the PDF - http://www.lfr.org/csl/new.html in case you don't like to download files.
What they are looking for are false positives. The question, and what will seperate real researchers from the woo-woos is how they will distingush a false positive from precognition.
Example:
I am testing Reaction Time to an auditory stimulous. I have a morse Key set up to a computer to capture a key press in response to a sound. Typically, you would expect a bit of a delay between the onset of the sound and the response. Sometimes a subject will respond prior to the sound, anticipating it, as it were. Lots of reasons for this and fairly common. Stepping on the gas before a light turns green, a sprinter taking off before the gun both come to mind.
Now if the probability of a stimulous occurring is 1 then any response prior to it correctly anticipates it (they are using GSR, presumably, but the same thing holds). The woo might interpret such responses as evidence of precognition. The deck is stacked of course and such an interpretation would be meaningless.
The question is does the response occur reliably prior to the stimulous with some fairly predictable temporal component.
If, for example, the response occurs, reliably, 3ms prior to a truely randomized stimulous we might have something.
I would trust that they would have trials with no stimulous whatsoever to serve as a control. These could be ramdomly scattered within the overall set of real stimulous trials and would indicate what kind of involantary respone occurs normally in that sort of situation.
I see two areas of problems.
1) Random means random. Control means control. Soundproof room, no other stimuli, reliable presentation of the test signal (like with ear phones).
2) Analysis means analysis, none of this Targ crap. As I pointed out any response is anticipatory. What is their operational definition of such a response that indicates precognition? Or, will they fish afterwards?
I can see a raft of post hoc rationalizations that we have seen with other research that has turned up bupkis. Such as "the precognitive temporal displacement (PTD) is highly variable and, by it's nature cannot be predicted with any degree of certainty.". This is, or would be, crap. They might also suggest that a response to a control trial is precognition of some trial in the future. They might sort thru the data looking for lag effects (response correlating with a trial 2 or 3 out) and defining that, post hoc, as evidence of precognition. PEAR, anyone?
They must replicate over a period of days. A significant effect one day that does not reappear on subsequent days cannot be explained away as some weird effect that is "characteristic of the precognitive".
If they engage in this sort of data torture, they might as well write their paper without the expense and bother of conducting an experiment.
If they look at the first set of trials compared with the last set and see a difference they might conjecture that precognition is susceptable to learning, in a classical way. This is also typical of woo research. Trying to gain legitimacy for a non-effect by linking it to accepted understanding though the existance of the effect is unprooven.
The net is that if these experiments are well designed and well run the results would be highly interesting. Beginning with a belief and then fitting the data to that will result in simply another sad woo experiment that will thrill the ignorant.
Example:
I am testing Reaction Time to an auditory stimulous. I have a morse Key set up to a computer to capture a key press in response to a sound. Typically, you would expect a bit of a delay between the onset of the sound and the response. Sometimes a subject will respond prior to the sound, anticipating it, as it were. Lots of reasons for this and fairly common. Stepping on the gas before a light turns green, a sprinter taking off before the gun both come to mind.
Now if the probability of a stimulous occurring is 1 then any response prior to it correctly anticipates it (they are using GSR, presumably, but the same thing holds). The woo might interpret such responses as evidence of precognition. The deck is stacked of course and such an interpretation would be meaningless.
The question is does the response occur reliably prior to the stimulous with some fairly predictable temporal component.
If, for example, the response occurs, reliably, 3ms prior to a truely randomized stimulous we might have something.
I would trust that they would have trials with no stimulous whatsoever to serve as a control. These could be ramdomly scattered within the overall set of real stimulous trials and would indicate what kind of involantary respone occurs normally in that sort of situation.
I see two areas of problems.
1) Random means random. Control means control. Soundproof room, no other stimuli, reliable presentation of the test signal (like with ear phones).
2) Analysis means analysis, none of this Targ crap. As I pointed out any response is anticipatory. What is their operational definition of such a response that indicates precognition? Or, will they fish afterwards?
I can see a raft of post hoc rationalizations that we have seen with other research that has turned up bupkis. Such as "the precognitive temporal displacement (PTD) is highly variable and, by it's nature cannot be predicted with any degree of certainty.". This is, or would be, crap. They might also suggest that a response to a control trial is precognition of some trial in the future. They might sort thru the data looking for lag effects (response correlating with a trial 2 or 3 out) and defining that, post hoc, as evidence of precognition. PEAR, anyone?
They must replicate over a period of days. A significant effect one day that does not reappear on subsequent days cannot be explained away as some weird effect that is "characteristic of the precognitive".
If they engage in this sort of data torture, they might as well write their paper without the expense and bother of conducting an experiment.
If they look at the first set of trials compared with the last set and see a difference they might conjecture that precognition is susceptable to learning, in a classical way. This is also typical of woo research. Trying to gain legitimacy for a non-effect by linking it to accepted understanding though the existance of the effect is unprooven.
The net is that if these experiments are well designed and well run the results would be highly interesting. Beginning with a belief and then fitting the data to that will result in simply another sad woo experiment that will thrill the ignorant.
Skeptical Greg
Agave Wine Connoisseur
Why the results might be a little bit biased:
Why tell the subject anything of substance?
Even more telling, would be the mixing of of different stimuli.. auditory, visual etc... And for some subjects to get no stimulus at all...
I'm still trying to weigh the benefits of slight advance warnings of startling sounds... I'll get back to you..
Why tell the subject anything of substance?
Even more telling, would be the mixing of of different stimuli.. auditory, visual etc... And for some subjects to get no stimulus at all...
I'm still trying to weigh the benefits of slight advance warnings of startling sounds... I'll get back to you..
deBergerac
New Doc on the Block
- Joined
- Oct 27, 2003
- Messages
- 452
Ed said:
Why are you surprised? You did not think your comment was thorough enough?
Jeff Corey
New York Skeptic
- Joined
- Aug 2, 2001
- Messages
- 13,714
Yes, but as Diogenes pointed out, telling the subjects of an experiment the results you expect before running them is an extremely bad idea, due to the demand characteristics of the situation.T'ai Chi said:
I wonder who these people are? An undergrad with one experimental psych course would be likely not to make that blunder.
Jeff Corey
New York Skeptic
- Joined
- Aug 2, 2001
- Messages
- 13,714
I emailed Ed May (whose Doctorate is not specified on his web page) and he said that they do tell their subjects in advance what to expect.
And they use a one sec 97 db blast of sound as the stimulus!
Have they never heard of sensitization?
Pavlov showed over a century ago that if you subject animals to repeated aversive stimuli, they get jumpy and oversensitive to any environmental stimulus.
I could never get this project passed by our IRB. The safe sound level is conservatively considered 85 db, and remember, this is a log scale.
And they use a one sec 97 db blast of sound as the stimulus!
Have they never heard of sensitization?
Pavlov showed over a century ago that if you subject animals to repeated aversive stimuli, they get jumpy and oversensitive to any environmental stimulus.
I could never get this project passed by our IRB. The safe sound level is conservatively considered 85 db, and remember, this is a log scale.
Paul C. Anagnostopoulos
Nap, interrupted.
- Joined
- Aug 3, 2001
- Messages
- 19,141
You know all those articles we constantly read that tell us how good the experimental protocols are for psi? How they are the best protocols going in the psychology arena? Crapola.
~~ Paul
~~ Paul
Jeff Corey
New York Skeptic
- Joined
- Aug 2, 2001
- Messages
- 13,714
Luckily we don't read those shoddy screeds in peer reviewed journals. (Unless the peers are not swift.)
I'm going to reprint Ed May's instructions and give them to my classes as an exercize. Not in futility, I hope.
I'm going to reprint Ed May's instructions and give them to my classes as an exercize. Not in futility, I hope.