can behavior be controlled?

Bruce Nevin [20190722.09:58 ET]

Boris criticizes the rest of us (and Rick in particular) for reporting the observation that behavior can be controlled. Let us set aside Boris’s misreading of this statement, as though Rick were saying that behavior is controlled (i.e. always, as a property of the control loop that is producing the observed behavior).

‘Behavior’, like all words, is an ambiguous word. The main meaning in general usage is ‘observable activity’. We might as well say ‘perceivable activity’, and when we phrase it that way it is obvious that behavioral outputs of a control loop may be either perceived or not perceived. When behavioral outputs are perceived (observed), like any perception they may be brought under control if we have means to affect them by our own behavioral outputs.

An example of controlling a perception of another person’s behavior is when the experimenter E in the rubber-band demo moves her end of the rubber bands in such a way that the movements of S’s end of the rubber bands moves in a path or configuration that E specifies. A policeman directing traffic, a parent stopping a toddler from falling into a swimming pool, a bully pushing others aside to get to the head of a line … the examples are endless.

As an example of controlling a perception of one’s own behavioral outputs during the course of controlling another perception, consider how we routinely correct a slip of the tongue, e.g. “the most highly played player – paid player.” Again, examples are endless–the students working with mirrors in the dance studio, the tennis coach, and recently here we had Martin’s description of observing his means of controlling with his right hand so that he could train his left hand to do likewise. Control of behavioral outputs is a very important form of learning, i.e., establishing appropriate connections so that control of the desired perception (eating with a spoon) cascades now without observation down through control of intermediating lower perceptions (which had been carefully observed as such during the learning process) to produce output actions analogous to those which had been carefully observed as such during the learning process.

So yes, behavioral outputs are not controlled by the control loop that produces them but they may be controlled by a different control loop with perceptual input from observing those outputs. Furthermore (as Martin’s example shows), any perception at any level may be a subject of observation by another control loop taking as its input some perception within the cascade of control through the hierarchy from the initiating perception to the control outputs. This may interfere with the control that is being observed because it requires attention to two disparate perceptions at once (attention is a limited resource) as well as because mere observation easily becomes control and control of a perception by two systems at once very easily creates disturbance and becomes conflict.

Note that this encompasses not only behavior defined as observable behavioral outputs affecting measurable variables in the environment but also behavior defined as a controlled variable in the perceptual hierarchy.