At 11:39 PM 2/11/99 -0000, Bryan Moss wrote:
>> subjects make a
>> detectable GSR `presponse' to visual targets not shown to
>> them until a couple of seconds later. This is a
>> statistical effect, as usual, found in concatenated
>> results
>This is how I'm interpreting the experiment (please correct
>me if I'm wrong): When they saw activity that they
>considered a 'presponse' they put a little tick in the
>'presponse' box, and if they saw no such ativity they put a
>little tick in the 'no presponse' box, and then added them
>all up in the end.
Sorry, I meant `consolidated' not `concatenated' - that is, they combined the curves gathered over many trials and screened out the random variations and noise. The composite curves then showed this apparent prodromal effect.
>the only way I can see the data being
>pure is to show only one image to each person, otherwise
>each 'presponse' can be attributed to the previous response.
The details can be found in Radin's paper, which might be accessible at his web site.
The method, roughly, is to show a series of bland images, to accustom the S to the task and establish a baseline, then to flash the test image somewhere randomly between the fourth and tenth images, say. And of course, there are many Ss, and their runs are consolidated. The aggregated GSR scores for the test items show a small bur significant surge in advance of presentation, compared with the background scores for non-target images (I simplify). Yes, there can then be `Gambler's Fallacy' expectation effects with confounding consequences, but the experimenters have allowed for this in their statistical analysis. Obviously, though, this will be the point at which skepticism will bear down hardest (unless it is assumed from the outset that the entire endeavour must be bogus - which is why it is interesting to examine existing data from earlier, non-parapsychological experiments open to retrospective analysis of such effects.. as is being done even as we speak).
Damien Broderick