[Martin Taylor 2007.08.31.13.15]
[From Rick Marken (2007.08.31.0915)]
Rick,
It seems that Martin Taylor has a good point in suggesting that "an
alternative hypothesis" to your assumption and assertions of differing
purposes "is that the priming changed the perceptual functions and thereby
the perceptual signals, and not the reference inputs" (Martin Taylor,
2007.08.21.23.11). Neural priming resulting in increased neural network
sensitivity to inputs of possible nonsense words could have produced the
differing results.
Yes, of course. But, as I said, the perceptual functions would have to
have been purposefully changed. That is, in order to configure a
perceptual function in a particular way there has to be a
specification (reference) for the goal configuration, some indication
of the current state of the existing configuration, a way of comparing
current to goal configuration and a way of acting to make the current
configuration more like the goal configuration.
Why? Other than ideological certainty? Who says that there is a "goal configuration" or any representation anywhere of the "existing configuration"?
Does Hebbian learning go out the window because EVERYTHING that happens in an organism must be purposeful? Do you dismiss the neurological findings on the changes in synaptic connections with use or disuse because the investigators haven't dealt with the purposeful control systems that effect the changes?
Saying that "priming changed the perceptual functions" is a
cause-effect explanation of what seems to me to be a clearly
purposeful (control) activity.
"Seems to me" is the key phrase here. It doesn't seem that way to me. We know, for example, that a neuron is less likely to fire in the few milliseconds following a firing, and that the sensitivity continues to change over the following period. It's a change in neural sensitivity. Is that a clearly purposeful activity?
It is an S-R explanation inasmuch as it
implies that what a person (the experimenter in this case) says
(words) can cause a change of a particular kind in neural
configuration.
It probably can, but I'd rather say "influence" than "cause". "Cause" sounds too much as though there were no other influences. Anything the experimenter does, and anything else a person experiences, is likely to change future perceptions. That's called "learning" in some circles.
I just don't buy that; it assumes too much intelligence
on the part of neurons for my taste.
Intelligence? To know enough to realize that the sensitivity "ought to be" changed (in order to achieve -- what?) after a firing, or aftr a neighbouring neuron fires -- or fails to fire? Such sensitivity changes do happen, you know, whether or not the neurons seem to be intelligent enough to know how they ought to change. I guess we are back to believing in homunculi, but now there's one for every neuron, I suppose.
If what was changed was the
perceptual function, it was changed purposefully (the state of the
configuration of the perceptual function was controlled).
Unsupported assertions will get you everywhere -- in politics, but not in science.
> You as an external observer may be assigning differing purposes to subjects
in the differing circumstances when no such differing purposes exist.
Even if all that changed was the perceptual function of the same
control system, I would still call that a change of purpose (in the
PCT sense) because the controlled variable is different in the two
cases.
Yes, it would be, or rather the c_E_v (complex ENVIRONMENTAL variable) would be. The controlled variable (perception) would not be changed, as Bill P. pointed out [Bill Powers (2007.08.31.0605 MDT)].
> Martin's suggestion has merit and should perhaps be more carefully
considered.
I considered it about as carefully as I could and found it to be S-R.
You correctly found it to be S-R. But then you went a step further, and dismissed it for that reason.
Every single stage in the control loop is S-R. The link from input to perceptual signal is S-R; the link from perceptual and reference signals to the error signal is S-R; the link between the error signal and the output signal is S-R. I wouldn't dismiss perceptual control for that reason.
OK. Now I'll partially switch sides, and refer to a fascinating experiment by Tony Marcel (Marcel A. J., Conscious and preconscious recognition of polysemous words: locating the selective effects of prior verbal context. In R.S. Nickerson (Ed.) Attention and Performance (vol 8), Hillsdale, NJ., Lawrence Erlbaum Associates, 1980). I won't go into all the experimental details, but Marcel did take care of the most obvious criticisms. You may think of others he didn't consider, but I don't think there is any easy criticism that would account for his result.
The effects of priming are complex and time-dependent. I'll ignore all that and outline what Marcel did. If you show a word such as "Doctor" and follow it soon enough with a word that is masked and hard to see, the masked word is more likely to be correctly identified if it is associated (e.g. "Nurse") than if it is unrelated (e.g., "Bacon").
Now suppose the prime (instead of "Doctor") is a word with multiple meanings ("Bank"). Then the following (masked) word is more easily recognized if it associated with any of the meanings of the prime ("Cash", "Stream"). However, if the immediate prime ("Bank") is preceded by a word that primes one of its meanings and not the other, the effect changes: "Money->Bank" makes "Cash" easier to read correctly when it is masked, whereas "River->Bank" makes "Cash" harder to read than it would otherwise be.
So far, we have a whole lot of results of this kind, which you could interpret as purposeful. What Marcel did was rather ingenious. He sometimes masked the ambiguous word ("Bank" in the example) so that it could not be (consciously) read and reported. When he did this, not only did "Money->(masked)Bank" make the masked word "Cash" easier to read, but so did "River->(masked)Bank". To repeat: if the subject was able to read "Bank", and it was preprimed with "River", then "Cash" was harder to read, but if the subject was unable consciously to read "Bank", "Cash" was easier to read even when pre-primed by "River". By itself "River"->"Cash" would make the masked "Cash" harder to read than when shown alone. The masked intermediate "Bank" reversed this effect if it could not be correctly reported, but did not reverse it if "Bank" was correctly read.
The point of this for the present discussion is that there IS apparently an effect of conscious as opposed to non-conscious perception, and therefore quite probably an effect of purpose, in priming.
But then, as I said, the priming literature is quite complex. (As is the whole area of non-conscious perception; some deny it happens, but I think Marcel's experiment shows that it does happen).
Martin