Sequences (Was Re: Amazing!)

[From Erling Jorgensen (2003.06.03 12:30 EDT)]

I want to pick up on just one aspect of Bruce N's. post, a tangential reference he made to scheduling capabilities and sequence perceptions.

[Bruce Nevin (2003.06.02 17:28 EDT)]

That's what sequence perceptions are good for. But that suggests an ability to put arbitrary and unpredicted control tasks under sequence control. In the experiment of shifting from person to NPR and back, are you controlling a sequence? How could there be a generic sequence controller with ad hoc inputs? This 'scheduling' capability has puzzled me for some time. It seems a bit much to ask of our reorganization process.

I envision sequence perceptions arising in this way, and I believe it is only a small step for a reorganization process to arrive at such a result.

We already talk of the model "storing perceptions in memory" - albeit without a defined mechanism that I have seen - and making later use of stored perceptions as possible reference standards. It is also the case that many organisms, without necessarily having as complete a hierarchy of levels as perhaps humans do, accomplish different control tasks at different times. They do this, then they do that, without necessarily having a perception (yet) of first-this-then-that.

What would be required for sequence control to arise would be for a perception to be created (i.e., a new perceptual input function) out of the existing memories of controlling two or more results not at the same time. It is the combining of previously independent results into a new perception of temporal near-proximity (i.e., temporal order) that constitutes a sequence perception.

Let me see if this example works (reconstructed from when I had to do such obligatory behavioral experiments back at Oberlin College). A rat performs what researchers call in shorthand form "exploratory behavior" - (controlling what? Scents in the cage? The feel of nearby objects against the whiskers? Other controlled variables we haven't thought to look for?) In the process, they do things such as nuzzle a bar press, rear up on their hind legs to touch a string (which happens to be attached to a light), arch their spines toward a food dispenser, etc. At first, I am presuming that things like the arching is controlled (i.e., the coordinated pull of various back muscles), but not necessarily the "toward" Relationship-perception. That is, the "toward-ness" of the arching seems to vary all over the place as the rat moves around the cage. (Obviously, I am leaving out many additional controlled variables at lower levels, which are embedded in the implementing behaviors of th!
e rat - things such as stretch of tendons at the joints, or even the varying-but-controlled presence of the right ions in the cells of the muscles, etc., etc., and so on, and so forth.)

Now, if food starts to appear at the dispenser (thanks to the control-related actions of a behavioral researcher), the rat may start to control "toward-ness," a relationship that may start to become stabilized against the disturbances of whether the arching may need to go right or left from where the rat currently is to get to the food dispenser. If that same behavioral researcher then disturbs appearance of food by withholding it until the rat's nose comes close to a bar press on the wall of the cage, then the rat may start to control another Relationship-perception, that of proximity-to-bar-press.

If the researcher does not change the new contingencies of the environment too quickly (as many impatient undergrad psych students are wont to do), then the rat may start to control the two newly-learned relationships of proximity-to-bar-press and toward-the-food dispenser, in a new higher-level type of relationship. We could call this perception "then," which is a newly-constructed perception of the type known as a Sequence. The full perception becomes "proximity-to-bar-press then arching-toward-the-food-dispenser," which basically only requires some kind of memory function so previously controlled perceptions can be re-accessed and combined in some kind of stabilized new way. I do not think it would be such a big leap for random reorganizing to stumble upon that kind of new combination.

As I have described it above, it is a Sequence of Relationships that is constructed. I am guessing that such coordinated actions as the side-to-side arching of the back, involved in walking on four feet, already include sequences of muscle constriction and relaxation to make them possible. I think there was a series of articles and responses in Behavioral Brain Research a while back (which Bill may have contributed to), that dealt with this type of locomotion. If sequences are necessary at that level, then it would seem that the Sequence level of controlling perceptions is pretty basic for many creatures.

Returning to Bruce's comment that "it seems a bit much to ask of our reorganization process" - I'd say the following. The threads on attention and point-of-view have included suggestions such as noticing the innumerable Relationship-perceptions of, say, the distance between our right index finger and any object in your field of vision. It does not seem to be a problem to create such ad hoc perceptions on the fly, so to speak. In fact, these are essentially adopting different references for the same perception of distance-from-forefinger.

I see no real difficulty with treating Sequence-perceptions the same way. The essence of a sequence is the perception of "then-ness," with place-holders for what lower-levels perceptions are put in the "before" and "after" slots. Once such a generic structure has been constructed, all kinds of different actual sequences can be controlled for. Reorganization is only needed for the initial construction of that type of perception. After that, varying reference levels dictate which specific sequences are perceived.

At any rate, Bruce, thanks for the thoughts that got these musings going.

All the best,
        Erling

[From (Bruce Nevin (2003.06.02 17:14 EDT)]

Erling Jorgensen (2003.06.03 12:30 EDT) --

  The threads on attention and point-of-view have included suggestions such as noticing the innumerable Relationship-perceptions of, say, the distance between our right index finger and any object in your field of vision. It does not seem to be a problem to create such ad hoc perceptions on the fly, so to speak. In fact, these are essentially adopting different references for the same perception of distance-from-forefinger.

Yes, I see. While doing a tracking task in which the position of the cursor is disturbed, I could shift control from the mark-cursor relationship to the relationship between the cursor and the edge of the screen, or between the cursor and a flyspeck on the screen, or between the cursor and an ant crawling on the wall visible behind and to the right of the screen. The inputs to the distance-relationship control system are arbitrary, although both must be in the visual field.

I see no real difficulty with treating Sequence-perceptions the same way. The essence of a sequence is the perception of "then-ness," with place-holders for what lower-levels perceptions are put in the "before" and "after" slots. Once such a generic structure has been constructed, all kinds of different actual sequences can be controlled for. Reorganization is only needed for the initial construction of that type of perception. After that, varying reference levels dictate which specific sequences are perceived.

In the sequence A, B, C, what is controlled is dependence of B upon A and of C upon B. B cannot occur until A has occurred, and C cannot occur until B has occurred. If I am controlling C, then I must control B first; but to control B I must control A first. In this way a sequence of any length can be constructed without the latching flip-flops proposed (as one possibility among many) in B:CP.

Dependence is more general and adaptable than sequence. Sequences can do much of what we might think are at the program level; dependence can do not only sequences, but also if-then, and, or, and possibly negation. Control of dependence is the only cognitive ability that is needed to account for syntax and the informational capacities of language (Harris 1991).

So is there a Dependance level instead of a Sequence level? Suppose dependence is a function of associative memory. That certainly admits the arbitrary inputs that we were talking about!

If relationship has inputs of the same type (e.g. both visual), this is not the case for dependence. That's OK. Associative memory is not limited to pairing perceptions of like kind.

How would a sequence-level control system differ from dependences stored in memory? Is there a way to test this experimentally?

         /Bruce Nevin

···

At 01:29 PM 6/3/2003, Erling Jorgensen wrote:

[From Fred Nickols (2003.06.02.2002))] --

Bruce Nevin (2003.06.02 17:14 EDT)]

Erling Jorgensen (2003.06.03 12:30 EDT) --

  The threads on attention and point-of-view have included suggestions
such as noticing the innumerable Relationship-perceptions of, say, the
distance between our right index finger and any object in your field of
vision. It does not seem to be a problem to create such ad hoc
perceptions on the fly, so to speak. In fact, these are essentially
adopting different references for the same perception of
distance-from-forefinger.

Yes, I see. While doing a tracking task in which the position of the cursor
is disturbed, I could shift control from the mark-cursor relationship to
the relationship between the cursor and the edge of the screen, or between
the cursor and a flyspeck on the screen, or between the cursor and an ant
crawling on the wall visible behind and to the right of the screen. The
inputs to the distance-relationship control system are arbitrary, although
both must be in the visual field.

A minor technical question. Is it truly the case that "both must be in the
visual field"? Might it not be the case that I could transfer the
reference signal to something as vaugely defined as "the way I feel (i.e.,
sense) the joystick"? Admittedly, I might not satisfy the requirements of
the experimenter but could I not keep perceptions (i.e., my sense or feel
of matters) aligned with my reference signal?

Fred Nickols
nickols@safe-t.net

···

At 01:29 PM 6/3/2003, Erling Jorgensen wrote: