Eight steps in Neuro-muscular Integration

In the early 1930's E. B. Holt was a lecturer at Princeton. He had retired from Harvard several years earlier and moved up to Maine to live the isolated philosopher's life. A friend named Herb Langfeld, who had been a Harvard colleague, convinced Holt out of retirement to come teach at Princeton, where he was much beloved by the students (undergraduate and graduate, including J. Gibson). While in Maine, Holt had deepened his interest in physiological psychology, and was desperately trying to tackle the biggest questions regarding how a physical body could do mental processes. This lead to his long, and difficult book Animal Drives and the Learning Process. It also lead to a chapter in a festschrift for Beritoff, a Russian physiologist, entitled "Eight steps in neuro-muscular integration." These works are great early examples of epigenetic thinking about behavioral development, it is contemporary with Kuo's earliest work, and anticipates Schneirla, Lehrman, and Gottlieb by decades.

It is not necessary... to assume the existence of any "inherited" pathways as a basis for reflex conditioning or learning. The very first and simplest reflex paths are learned, that is, conditioned (prenatally) according to the reflex-circle principle; and the earliest muscular contractions, as required for the starting of reflex-circles, are the early random movements of the foetus.... and this fact leads one to question whether the important role so universally ascribed to "heredity", at this point, is anything more than an old myth. (Holt, 1936, p. 27)

I want to summarize the festschrift paper here, both for its historic value, and as a step towards adding this information to the chapter I am working on with Andrew and Sabrina. We are working (slowly) on a chapter for a book on "Neuropragmatism". This is an emerging movement, not yet well formed, and even if it putters out rapidly, I have a feeling the initial book will be widely read. The short neuropragmatism conference that I attended reaffirmed my belief that large numbers of neuroscientists are becoming disillusioned with cognitive psychology and looking for alternatives. Andrew, Sabrina, and I will claim that the most valuable thing that could come out of an integration of pragmatism with neuroscience would be a better way to talk about the role of the brain in psychological processes, and we will be offering some initial ideas along those lines. Note that any vocabulary acceptable to pragmatists would likely be acceptable to behaviorists and ecological psychologists, as those moden movements grow out of the pragmatist tradition.

Full disclosure ahead of time: Holt's paper was not as rich as I had hoped, though I think there are still aspects of it worth pulling forward.

Eight steps in Neuro-muscular Integration

(A Summary)

The first thing worth noting about Holt's approach should be obvious from the title: He is not interested in how the brain "controls" the body, he is interested in how the body as a whole becomes more integrated in its responses to the environment. (One would not be silly to think about the dynamic systems implications of J. J. Gibson's work on perceptual systems combined with E. Gibson's work on perceptual learning, both had classes from Holt as graduate students.) It is also worth noting that the "steps" that Holt lists might be better described as "levels". During the course of development, there is some sense in which you need to get some organization of the first type before you can get organization of the second type, etc., which is why Holt called them "steps". However, once some of that development has already occurred, it would probably be more natural to think about them as levels of organization (fractal levels of organization?) within a complex system.

Holt's introduction criticizes Pavlov and his followers for their "singular disregard for the generally accepted facts of physiology", a criticism many would later level against Skinner and his followers. Holt also suggests that much work in physiology has recently allowed better (mechanistic and developmental) explanations for Pavlov's findings, but that the physiological studies were carried out by such different methods, in such different contexts, that the connection has not yet been made. Holt then describes a sensory-motor loop, whereby a movement creates new sensations, which eventually leads to stimulation of the neurons which caused the motions, leading to further neural development of the perception-action pathway. With this, he hopes to connect Bok's notion of the "reflex-circle" with Sherrington's notion of the "circular-reflex." (Though the names suggest an obvious similarity, one comes out of physiology, and the other out of psychology, one in German, one in English, etc., so it is not trivial to be synthesizing them.)

Be forewarned, this is a bit odd, but stick with me. The bit not to miss is that we are talking about similar things happening at different levels of organization (which unfold in steps during the course of development). Thus all "higher" steps are not intended to stand on their own; they assume the existence of all "lower" steps, and build from there.

---------

Step 1: Circular reflexes.
A circular reflex is the loop from a motor action to the perceived consequence. In the simplest example, the nerves tense a muscle, thereby squeezing proprioceptive cells. The activation of these proprioceptive cells will typically lead to muscle relaxation. This is what happens, for example, when you hold your posture.

Step 2: Reciprocal innervation.
At this level antagonistic muscle groups are integrated together so that the neurons that flex one muscle are linked to the neurons that relax the other.

Step 3: Adient reflexes; "Instincts"
At this level, movements produces sensations, and this will lead to "cannalization" of the connection between the given sense-organ and the muscles that produced the movements. For examples: A human fetus makes lots of random movements, and when the fingers flex they touch the palm. The resulting reflex then makes the fingers flex in response to the stimulus of the palm being touched: i.e., the new born baby's grasping reflex when you press on its palm. To be "adient" is to approach, and these sensory-motor loops cause the organism to approach the cause of a stimulation.

While this principle most literally holds in the case of touch, there is also speculation about how similar processes could lead to auditory imitation (as a loop between the throat moving a certain way and hearing a certain sound).

Further, as the flexing or relaxing of several muscles that can potentially result in hand movement, integration at this level can begin to explain the coordination of more elaborate muscle groups.

When a [sensory] surface is stimulated, all muscles which by their anatomical position can assist in moving this surface towards the stimulus "adience", will do so (muscular "synergy").... When it is recalled that muscular synergy is not wholly determined by anatomy, that the special posture from which a movement starts (together with the direction etc. of the movement) determines which particular muscles are the synergic ones, in each case, it becomes rather obvious that the reflex-circle principle is the only conceivable means by which synergic coordination could be established. (p. 30)

Step 4: Concatenation of reflexes
 At this level, reflexes are chained together to form longer behaviors. Holt specifically invokes in this context William James's notion of Habits (a thing someone does as a matter of routine). Holt adds the bizarre teaser that:

For the psychologist, the [feedback at this level] is the mechanism par excellence of the "association" of "images" and of "ideas".

What on earth is he getting at? With the scare quotes in mind, this is my take: At this level organisms are responding to complex patterns of stimulation with complex (from a neuro-muscular standpoint) behaviors. The complex patterns of stimulation are what psychologists are trying to get at with the notion of the "retinal image", and complex behaviors are (Holt likely feels he has established elsewhere) what psychologists are trying to get at with the notion of "ideas". Thus perception-action loops at this level are connection "images" with "ideas" - scare quotes required.

Step 5: Movements of exploration
Here, for the first time, Holt explicitly narrates the build up of the phenomenon in question from the previous four "principles". He explains how such loops, along with a bit of random movement, create the exploratory behavior one sees, for example, when a ball is placed in a toddler's hands. If these experiences repeat, then new patters of action are set up, such that the same "exploratory" movements can happen even in the absences of the ball. (Hold your hands out in front of you and pretend to be touching the outside of a ball... that's what he's talking about.)

Now, readers of this blog would likely see that even the above principles have implications for psychology, but here Holt wants to make sure his reader gets the point, and in the most radical way possible:

The reader may have noticed that we have been considering phenomena which seem, more and more, to lie on the borderland between physiology and psychology. And it may be well to state here, that in the opinion of the writer the motor exploration of the contour of an object is the "perception" of that contour; and the motor re-creation of the contour (in the absence of the object) is an "image" or "idea" of that contour.

Wow. Descriptive mentalist, non-magical psychology at its most brazen. Alas, further defense of these ideas would be out of place in this particular article.

Step 6: Interconditioning of simultaneous explorations
In this step exploratory movements between different senses come to be connected, such that the exploratory movements of the eyes arouse (at least a little) the exploratory movements of the hands, etc. Thus, something that stimulates a "visual" exploration necessarily also activates the parts of the body that move a person closer to the object being explored, around the object, etc.

Recall that Holt has already established that one can "imagine" an object, by doing the things one would do in the presence of the object. While Holt is not explicit here, in addition to explaining how (as Gibson would put it) the activity of "whole perceptual systems" develop, he is also trying to explain phenomenon like when a song playing on the radio invokes a visual "image" of past events. The bodily system orienting towards that past event is the remembering. (And for more support of that last point, I can't help but recommend Francios Tonneau's chapter in the Holt book, chapter 3.)


Step 7: Cross-conditioning of reflexes
Here Holt feels the need to set up that "specific motor activities" occur on a background of "tonic postural reflexes". I'm not sure why that's here, but he explains it well. He then tries to apply the prior principle with the responded-to object not present. That is, we often see a variety of sensory-motor systems brought to bear as if an object were present. Holt offers the example of a child

seen to "pretend" to hold a doll tightly under one arm, to prattle it, and feed it bits of candy, even when there is no real doll to hold.

This sets up a dilemma. Because when the doll is present, we may easily explain the behaviors as responses to the doll (or if you prefer as the result of the continuous interaction with the doll). But in the case of an "imaginary" doll (scare quotes in Holt):

there must be other stimuli operating... and since these stimuli go wherever the child goes, they must be either internal organic rhythms, or else such ubiquitous stimuli as those of gravitation, general illumination, etc.

 This seems to start in a good direction, but then gets so amazingly general. In fact the 'imaginary' doll is not everywhere, and I'm not sure why Holt doesn't claim that identifiable non-doll stimuli are identifiable. The next few sentences are worth noting:

A child that is thus "cross-conditioned" exhibits in its behavior an apparent "self-determination" which is sustained by no visible sensory stimulation.

 The name "cross-conditioning" has been given to this process merely because the actual stimuli which sustain the motor performance have no obvious, intrinsic or necessary connection with its performance.

This last part was appealing, both in its connection with the appearance of self determination, and with its appreciation of the importance, for psychology, of non-obvious causes (e.g. the excellent epigenetics work such as Miller's 1997 paper). This seems better than the blanket statements about gravity, etc. There are specific causal relations to be discovered, but they are not connected to things that obviously have to do with the ostensible target object (e.g. a doll). Note that for these types of response to occur, the organism must first have a repetition composed of perception action loops at the lower levers, which then come under the control other variables.

Step 8: Simultaneously sustained reflexes
 Holt reminds us that when dealing with directed motions (which I suspect he intends to include all steps here)

The response seems to "refer" back to the stimulus, and it cannot be, as a movement, accurately described without reference to the [stimulus]

And he goes on to claim that when two or more stimuli are present, no matter how subtly, the resulting behavior of the organism is the algebraic sum of the individual effects. Recalling from the earlier steps that each reflex "leaves as a residual after-effect a mildly tonic posture," any current behavior occurs with respect to many stimuli. Holt admits that this might not be impressive as a final level of integration, at least at first blush. However, he believes it is key for understanding phenomenon like successful navigation of the environment, and intelligent behavior in general (i.e., behaviors that must occur with respect to several simultaneous aspects about the world).

Holt ends with some minor notes about some differences, with regards to the relative importance above steps, between different types of organisms. He also praises Beritoff's contributions regarding the complexity of reciprocal innervation.

-------

Again, my goal has been here to summarize this paper and try to draw out aspects that might be useful to bring forward.  The one key throughout is an attempt to show that perception-action loops exist at several levels of analysis, and that they can be a key concept for psychology.

-----------------
Holt, E. B. (1936). Eight Steps in Neuro-Muscular Integration. In Problems of Nervous Physiology and of Behavior (pp.25-36). Tiflis: Georgian Branch, Academy of Sciences, USSR.

Miller, D. B. (1997). The effects of nonobvious forms of experience on the development of instinctive behavior. In: C. Dent-Read & P. Zukow-Goldring (Eds.), Evolving Explanations of Development: Ecological Approaches to Organism-Environment Systems. (pp. 457-507). Washington, D.C.: American Psychological Association.