[ExI] heaves a long broken psi

[Damien Broderick]

On 1/24/2010 8:43 AM, BillK wrote:

> If you give the brain a random list, it is most unlikely that the
> human will guess correctly at the expected chance (random) level.
> Either because the test was too short. Because the expected chance
> level is only achieved over long durations tests to remove random
> fluctuations.
> Or the human was making up patterns of guesses (it has to - that's the
> way it works) and the patterns don't match a randomized list. They
> will be better or worse.

Hardly anyone uses iterated guessing protocols any more (remote viewing 
protocols are far more informative, because they contain high entropy 
gradient targets and are better fitted to the way the mind looks for 
meaningful structures), but there is a huge data base of just such 
experiments compiled at Princeton and elsewhere. The existence of 
preference patterns in unmotivated calls is one of the first things 
established in such experimental runs, and the most interesting aspect 
to look for is deviations from such individual or population biases.

I looked back at a lot of data accumulated between the 1930s and 1950s, 
comparing calls against the background preference patterns, looking for 
the proportion of guesses when a particular option is target and 
comparing that vote with the proportion it got when not a target. That 
is, the comparison is not made against "expected chance (random) level" 
but against an internal control that tracks non-random preferences. 
Since the target list is random and the choices are made blindly, there 
is no apparent way in which "when target" scores can be significantly 
deviant from "when not-target" scores--yet they are, to a significant 
extent.

> I call it random. That's why the tests are not repeatable.

I see. First you explain that psi tests are bound to give results that 
deviate from chance because the mind produces patterned or skewed or 
non-random streams of calls, and these inevitably match better or worse 
than m.c.e. against a short randomized list, which easily accounts for 
the non-chance results of parapsychologists--and then you explain that 
this is why such tests can never be replicated by non-parapsychologists! 
If what you say were valid, anyone who tries this should get results at 
the same level, including the most ruthless skeptics, because it's just 
an artifact, right? But of course it is only those gullible 
parapsychologists who do so. So we're driven back to John-Clark-type 
"they just made it all up" or "they all cheated" theories.

Damien Broderick