Information Exchange in the Time Domain


An important piece where Brodey reflects on the differences in approaching the study of people and their relations from the perspective of psychiatry and engineering. It introduces another important Brodey concept: time-graining.

THE EXPERIENCE on which this report is based is different from that of the average psychiatrist. I have had the enjoyable experience of reaching into the land of “the new technology” where tools are being built to use strategies that in man would be considered intelligent. I have been given the opportunity to think and work at the integration of man and machine, starting from the “man side” of the bridge and reaching toward those engineers and scientists whose technological skills are stronger on the machine side. This bridging is slowed by ignorance and by our own deficits as psychiatrists in finding ways of modeling our observations that would be more effective in communication. Our deficit, I believe, is due to the fear of science in which we psychiatrists lived—and I think justifiably so—in precomputer days, when there was no possibility of real time computation. 


Why would the psychiatrist, the astute observer, the skillful clinician, the man who knows people, live in fear of science? The answer to the question is that he was preserving his science from the popular misinterpretation of science as a system for divorcing the facts from their relations and “truth” from its context. He had been a victim, often hurt by the logical use of his words by those who did not understand this need for protection.

We have protected our science by coding it into handwaving and words which have meanings defined differently on each occasion. We have preserved our knowledge from the deadening oversimplification of experiments which imitated the rituals of scientific methodology without comprehending its substance. The experimenters heard our words as if they were simple in their way. One tried to explain: That word is meant that way only for that context at that time. These labels gained meaning as they provided context for each other. The physical scientist knew this relativism; the technologist imitator of the 1930s did not. 

My own training in science was a doctor’s training—a training in the ways of exploring, searching, or picking up curious relations— and of figuring out ways to check whether my observations were useful in terms of more or less rigorous testing. Rigor included striving for the level of accuracy meaningful for a purpose. Accuracy for its own sake, or objectivity, when it could only prove the truth or falseness of what I already knew, was not yet considered weighty enough to make a difference. Mostly I knew that the discipline I cared about did not significantly apply to the butchered pieces of phenomena that resulted when ordinary and simple measuring units were used. When complexity is denied, living things can no longer be studied with vigor and discernment: Procrustean “scientific methodology” produced valuable but nonetheless dead biology. 

In this paper I will attempt to show that it is now time for us clinicians to come out of hiding and to leave behind the fear that science means to destroy our discipline in its complexity in order to make it measurable


Join the ranks of scientists aware of the failure of the old scientific recipes. New tools allow us to reexamine the epistemological models that since Greek times were used to separate living phenomena into formal simplicities. These old traditions are hidden. They are built into our language of words, our grammar, our numbers, and in our out-of-awareness religions. Subject-object is a tradition of our culture —observer and observed are the same.

The prerelativism observer had the objectivity of God looking through a peephole from superspace. He was not affected by the system he observed in a way that meant the system changed by reason of being observed. All was true or false by some truthtable, statistical average. 

As we approach observation systems in which the observer is less in the loop, the simplicities of the old hard sciences retain their usefulness unless we are seeking extreme rigor in our measuring. The astronomy enthusiast does not affect the moon by his observations. 

The observer in the old “objective” position could not afford to get too intimate with what he was observing or what he chose to measure “objectively.” For he could not easily measure his own participation without being entangled in the richer network of the myriad signals he was receiving and returning as he talked to the phenomena, even as he listened to the reply. This way of talking to the phenomena is the way scientists actually search. Later they research their territory into the scientific format for describing and proving to each other the observed phenomena. The research is a way of finally formalizing their discovery into a code for communication—the language of hard science. The hard science formalism structure has power when the observer and observed are not interdependent. 


In the soft sciences we seek to comprehend the phenomena we are observing, knowing from the outset that observer and observed are strongly engaged in their data exchange dialogue. This dialogue does not mold into the questions that can then be significantly researched in the old style. So we had search, but little research; we were full of hypotheses, post hoc explanations, and ongoing manipulations, all discussed as predictions. It could not be otherwise: we lacked both notation and measuring apparatus which could measure dialogue. 

First let us approach the question of notation. We have as yet no way to symbolize this dialogue, no notation that can make this phenomenon translucent. 

Now I am using the word dialogue with a very definite meaning. To illustrate, I refer to the network of systems which allows us, for example, to look into each other’s eyes and to counterpoint each other’s changing as we change, so that we evolve a sense of being known even as we together metabolize and unfold the next moment and its surprises. 

Yes, a table of people, or a roomful, or a person, or a cell, or a city, metabolizes what was a moment ago merely “noise”—unpatterned background—into information. The surprises of the future becoming present grow as we grow with them. It is this growth into the unknown, and the rich power that systems have of organizing themselves and each other so as to metabolize the unknown, and finally to excrete the obsolescent, that is in the range of phenomena that applies to our problem and that needs modeling. This is the right arena. 

In viewing man as metabolizing information we see him as a creature who has choices as to how he will pattern what he perceives and the way he will locate his perceptors—so as to adapt to the unknown and stay within his homeostatic limits. 

As Ashby stated in the United States Air Force-sponsored Bionics Meeting held in Dayton in 1966, man does not operate within data processing limits that we know. In order to manage the millions of choices he can make at any moment, he clumps these into patterns, or programs. But remember, the question vital to this paper is not what these patterns are or where they come from but how man grows them moment to moment. How does he grow his capacity to metabolize into information what a moment ago he did not even recognize as meaningful? What units might be applicable to the description of the unfolding of a family? I do not know the answer. I know that it is not to be defined by timeless logical conundrums that legitimize the denial of the edge of change as it happens. 

Having raised this question, let me remind you of the new developments which now direct us towards binding techniques of dealing with the level of complexity and interdependence familiar to the clinician. It is now becoming possible for man-machine communications systems to simulate complex man-man communication. In fact, in order to make full use of our artificial intelligence devices—the new generation of computers—we must tackle the dialogue problem directly. In work in which I am a participant at the computer laboratories at Massachusetts Institute of Technology and the National Aeronautics and Space Administration in Cambridge, Massachusetts, this problem is critical.

In the biocybernetics laboratory we are seeking to develop techniques for maintaining control of the relatively closed information ecology of several men isolated in a spaceship during a deep space mission. The first problem is to develop nontrivial questions which are relevant to the purpose of the search. Let us proceed. When observer and observed are in the same loop, each changing as a function of the other’s changing, what can we specify as characteristic of the system? What can we specify as characteristic of that loop? What framework can we use to structure our notation system? What notation will allow us to tell of this framework so that we do not have to show the whole story in metaphor or handwaving each time we tell it? What we need is a structural framework for our model that will answer the following questions: What kind of model is appropriate for formalizing the behavior within the system that is relevant to our questions? What graining or texture shall we use to capture the features of the phenomenon we study so that we might at this beginning stage of study be able to reorganize and not go looking for a mountain with a microscope, or trying to measure significant movement of the human hand in a microsecond? We psychiatrists are experts in formalizing the dialogue in literary language. W hat can we say about information exchange characteristic of a dialogue? 


We realize first the power of a dialogue in driving its components. Two or more people speaking and contacting each other—en rapport —can become driven by their conversation so that they, the components, behave quite differently than if the loop of information exchange that generated between them were growing in a different way. It is in the nature of conversation that it would not develop were this not so. The spontaneous enthusiasm of two people close enough to the same “wavelength” may propagate to include a larger group, and the build of excitement may organize those who ordinarily do not vibrate to the “subject” context of the dialogue. For them the dialogue field created polarizes the neighboring noise into meaning, and context is engulfed and metabolized. What is the generalization of the above familiar situation? Complex fields of interdependent subsystems (humans or cells) do not vary randomly or continuously, but by reason of becoming interdependent they begin to lock in and out of step and organize. If each person in this audience were to stamp his boot now in time with his pulse so that the others could hear his rhythm, we would all soon succumb to a common beat or have to fight against rhythming together—being “entrained,” as the engineers put it.

Behavior of interconnected objects or systems develops harmonies which allow the labeling of significant and meaningful groupings within the assemblage, or the interdependence breaks down and there is no dialogue. The dialogue of an interdependent system is maintained by growing and evolving, for simple repetition does not easily persist when there is adequate complexity.

But how can we conceptualize the dialogue so as to retain its feature of ongoing growing, of evolving not by stopping change but by playing against the kind of changing that is evolving. How can we do this? W e psychiatrists are the experts. We join the dialogue—by playing the timing game. Can we formalize the way we time into the loop of an internal or interpersonal conversation? This is an important aspect of therapy; we change the dialogue’s evolution by the way we time our changing as the patient changes. A mother who does not enter into a dialogue with her child cannot transmit the fine texture of information the child needs if he is to be skillful in evolving. The mother and child who can time to each other in a responsive way may each give to the other myriad data. But if each displays at the moment what the other cannot receive, the dialogue does not develop so as to facilitate the next message exchange. The predicament is selfreinforcing. Thus two children may, because of this timing problem, live in worlds which are different in information content by many orders of magnitude. This timing is an important aspect of the information exchange system of a mother and a child.

The loop as defined in cybernetics or general systems is never timeless. The simplicity we thought we lost in giving up (i ) the historic, simple dependent-interdependent model of relations and (2) separation of the system into components as if the components contained all the information about their relations, we now trade off against data we can produce by grouping the interdependent phenomena of dialogue in terms of their timing.


But here we must begin, as you would expect, to introduce new words. First, let me introduce time graining. By this I refer to the kind of time lattice work, or texture, or grain (as on a camera film) with which one considers a system. This concept has been presented in a paper called “The Clock Manifesto” presented at the New York Academy of Sciences Conference on Time in 1966, and published in its Annals [1].

This paper deals with our unwillingness to consider clocks as existing for our use—and for our resourceful design. Clocks are considered to be divine, like the stars they imitate. You are all familiar with size graining and the need to distinguish the perspective of a mountain and a flea, for each perspective delivers a different system to our notice. This is equally true of time graining. History 7 is commonly clocked into a lattice of important event time. The time distance between two important events, wars, for example, is considered the same. This time graining is not an equidistance in the astronomical sense, though it may be in the information sense—the wars occupying more space in the national memoirs than the poorly remembered periods of peace. This clock is marked by wars into equal amounts (a familiar story to the clinician). If this story still sounds strange to your ear let me remind you that it was not long ago that the clock used by common people was sunup to sundown as an equal span regardless of the season.

Time graining implies a necessary constraint in the discussion of complex information exchange. We must specify the kind of time regularity we will use in conceptualizing our loop phenomenon. No system which is complex and active will be without periods of relative stability and instability when it is more or less controllable. Phenomena are often quite naturally grouped by similarities in their time textures. Dialogue between systems of vastly different time textures is likely to be sparse, and thus these can more easily be modeled in the hard science mode. Complex systems have their own simplicities. Now you will see why we are not out to kill complexity but to use it for our study of the system.

The self-organizing power of our engagement with a system which already has some stability depends upon our playing jujitsu with it, by rolling with its changings just enough to find its moments of lessened stability, where a new piece of information may stream through many different levels—rippling throughout the system to produce a readiness to use the next ripple we create at just the right time in an even more significant way. You will recognize here the mechanics of innovation. When we can enter the dialogue we can examine each organism’s way of evolving and evolve our own skill in adapting

I am saying that each system has a clock and that if we wish to enter the dialogue this is to be measured in terms of the control periodicity natural to the system rather than mean Greenwich time. Only heroic change can produce innovation by a creation that is not evolutionary—and heroic change may kill the system’s growth, slowing the adaptation that creates information out of the system’s context. The mother crudely told not to spank her child may kick him; so it is, when change does not grow from inside. Modern control theory is quite different from the power control of the nineteenth century; it now conceptualizes control in terms of skillful use of information to make minor interventions count.

There are optimal time grainings for different purposes. This becomes important, for example, when one is trying to decide how responsive one should be in picking up the change in the system. Some people are set to pick up change in moments of very brief duration— they notice a flickering eyelid which is out of timing with their anticipation; others listening with longer moments will note how today’s traffic movement is different from yesterday’s. The choice of time grainings useful for a particular purpose with a particular system is limited. The time grain constraints are imposed by size and conduction delay in the communication network which integrates the complex system so that it can respond to the innovation. They are also a function of that different time delay for data processing, storing, and retrieving in memory. A psychiatrist may jump from responding to one time system, perhaps of years, speaking in words called generalities while still speaking with his hands in a time system of a few seconds, change being built together with response into these brief period dialogues. His jumping is itself a major communication—his jumping a message channel.

If one is too far outside the time graining of the system in one’s choice of responding, communication is minimized and is said to be “simplified,” for it is now without context. Consider a cheer leader who does not play against the timing of the crowd and only gains attention when she “works for it.” Contrast her with a leader who gains the crowd’s attention when it is just emerging from a hushed end run. The cheer leader moves at first in, then out of phase, and then to a faster pace, guiding her action in terms of the change in the total system (including the change within herself that occurs as she is changed by her involvement). She is active in creating the build— a nonlinear build—a build whose growth organizes context. She is able to transmit enormous amounts of information with significant effect. Similar time graining is important in making the decision as to when the display back to the user of a computer will have the most effective meaning in terms of enriching the man-computer dialogue so that it organizes what was peripheral context for the problems at hand. The possible effect of devices that can present data in terms of actively predicting user needs is an important area. Mancomputer systems that will be able to use context as you and I do will require more knowledge of the timing which facilitates information exchange.


Another new word I would like to discuss is tim e driving. This is what the cheer leader did when she timed her action slightly ahead of where it was expected and called for. People easily lock step if they meet others who are sufficiently similar, and as I discussed earlier in the paper, the lock step may be one with continuous acceleration far beyond the usual threshold. The escalation in the system may time drive its components far past the point they wished to go. This happens most obviously in war when struggles intended to defend may go wild and destroy both attacker and defender without regard for their purpose. 

The content or information exchange of the dialogue must not be separated from its time characteristic; these are different aspects of the same loop which act as context for each other.

The same dialogue may have profoundly different effects, depending upon its time driving characteristics—its timing or phasing or feedback.

Let us review. In discussing time graining we decided that for any phenomenon we intend to alter there is a set of particular periodicity ranges within which we must set the periodicity of our observing, if we are to influence that phenomenon by our return communication. Thus, Wiener points out that it is possible to influence the frequency of fireflies’ lighting by flashing a light at about the fireflies’ own periodicity—just then, blinking a little faster speeds up the fireflies’ lighting. This illustration is like the cheer leader mentioned before who catches the tempo of the crowd, or the actor that of his audience, to gain a more significant communication position. But now let us move on from these homely examples to state the situation a little more formally. For each of these loops there are limited time ranges that will lend themselves to entering a dialogue with the phenomena. Flashing the blinker once a week or in each millionth of a second will not allow us to affect the fireflies’ rhythm.

If we (1) choose one loop of a system of multiple connected loops, (2) seek out the limits within which it is unstable, insofar as it is reactive to an appropriately time driving input-output signal, usually translated by its environment in such a way that some part of it becomes an input used for control, and put it back a little sooner, giving it a slightly out of phase time feedback, and (3) perturb the system while engaging it in a dialogue by distorting the time function of its feedback, we then can find (a) the points at which the time driving fades out and at which it continues no longer listening, (b) when the system shifts to another mode, and (c) what loops are driven together. Thus, we begin to see how loops are interconnected in a control sense by the way they change together as a function of the altered periodicity, and at what point thresholds are reached so that the system becomes a different one in a discontinuous way.

You will think I am being highly technical, but I am describing in my more formal way what you must recognize a psychiatrist does quite intentionally when he is trying to explore what circularities a patient or family will be driven to, when one circular system is repeatedly primed into action by a suggestive comment at just the moment when the circle would not otherwise be repeated until after a rest period—but still is available to react if stimulated strongly. The same suggestive comment at a different time would not have the same meaning. The psychiatrist uses this device less intentionally when he places therapeutic comments just before the patient would have said it, the therapist predicting from context, or from his recollection of the beginning facial or lip expression being followed by this comment many times before. The patient gets the sense of rapport; they are timing together; he understands; more information can be exchanged, and more associations come that fulfill the prediction.

Such time driving, natural to humans in their dialogue exchanges, is frequently out of awareness, for our namings have not included such timing procedures.

When we consider the problems of education, time driving is particularly interesting. Actively unlearning the obsolescent is as important in education as presenting new data. In helping the child excrete informational organizations previously useful at earlier stages of his learning evolution, and taking care not to allow the newly presented information to crystallize too much that is unworked into words prematurely, one can present the obsolescent expectation just before the child is ready to present it but with an irrelevance associated, such as varied unexpected words from new contexts. Psychiatrists use this technique for helping people unlearn patterns they consider to be no longer useful.

Time driving is a powerful way of building and enhancing the exchange of information. A mother who presents her infant with food at a time when he has signaled that he does not quite want it, or after he is just a little too hungry, or just when he has signaled most clearly that he wants it and has the message out, is able to use the food not only for nourishing the child’s tummy needs but also to build their communication. Being skillful in using timing is important to the earliest learning dialogue. The mother who knows the time graining which is appropriate to mapping her child’s meaningful moments into her own can enter into a dialogue with him which, as it grows, organizes their communication so that just a glance between them is rich and carries with it a confirmation of their acknowledging together the meaning of a gesture just made. 

But I have spoken to this point of time graining and time driving in the manner made necessary by our subject-predicate language, as if there is a “doer” of the action and “one done to”—a receiver. The loop concept has been subverted—if A and B are in a true dialogue their changings are as entwined as two lovers. Now there may be moments when the action is more unilateral, but the information exchange is the freshening of changings responsively timed to each other’s changings. Who is father to the child fits the cause-effect model, but the evolution of the child and his family are a matter of dialogue.

An action spoken as a family builds a relation in which their shifting conversation carries their associations in an easy, timed flow is a million times richer than if mother’s arm and infant’s or father’s do not time to each other’s. Knowing the others’ time graining and mapping it onto one’s own is responding to how long a child’s look sees when a child stares. Does he see very briefly a hand and a foot, or is he drinking in a whole scene? And this ability to teach complexity unintentionally by responding to moments which can be shared is a function of the dialogue which transcends the participants and evolves them. Each tests against the dialogue, each knows himself as a mix with another, and each knows what they can grow with timing to the music they evolve. This is a way of conceptualizing interdependent ecologies: families of similar cells or families of porpoises. I am using man as an example because it is easier to comprehend this idea in our own context of psychiatry.


Having introduced the concepts time graining and time driving, now let me go one step further into the problem of information time. This is the periodicity of equivalence in information exchange quite unrelated to the familiar clock. Let me state that this concept is simple, but it is also commonly discussed only by handwaving or showing, as opposed to telling. It has no formal language that is available to the nonmathematician. Let me point out first the usefulness of this term in conceptualizing effective dialogue. But let me state, again, that formalization of the concept is just at its beginning. Let us begin as before with homely examples. Compare grandma’s time to a 2Vi-year-old child’s, and you will be aware that their times are very different. The child will move through his event space at a rate which far exceeds grandma’s. His very biological changing, moment to moment, his fluctuation in heart rhythm and cell division, means he is processing more novelty simply by reason of changing his own receptor system. An hour of Johnny is equivalent in novelty to a day of grandma. Novelty has been used as a measure of information. Information in America is formally defined as “a measure of novelty.” If equivalent periods of information processing were used as equivalences for building a clock (instead of rotations of the earth) we would perhaps have a more useful clock for measuring dialogue.

But to many this may seem an esoteric problem. The problem of giving up the hard science methodologies of the last century remains with the soft sciences. Physicists work with many clocks and build clocks to fit their purpose. That they be well defined is the only requisite. But let us return to the homely examples I mentioned. In communicating, Johnny and grandma both enter systems with different time grainings. If grandma is a warm person she can enter the child’s time and the child will enter grandma’s, but the communication in their dialogue will depend upon grandma’s and Johnny’s capacity to map each other’s time.

To present yet another homely example, in an “emergency time” one has the sense that during the emergency, as when an accident is impending, 3 minutes are the equivalent of 3 hours—and one’s ability to act in the 3 minutes may match 3 hours under ordinary circumstances.

When a creature who acts very quickly meets another who is very slow they will usually converge their pacings if they are to find a way of building a dialogue. The child who moves at a very fast pace and learns in a way that seems quick to one who moves slowly may be so surrounded by a pacing context of slow-moving objects, as in a traditional school, that he slows down to a pace which is no longer his natural pace. Such a child may not process ordinary information well enough to be a very good learner, or his “biological system” is not locked into his dialogue with those who teach him. Note here I am talking of the biological dialogue as the dialogue which includes the whole child and his whole manner, the dialogue of muscles sensing each others’ movements as their “together movements,” called the child, enable each next participation. If his behavior is not matched to his test time, the child is not synchronized and appears to be intrinsically awkward.

Information time means setting two time periods as equivalent when the same amount of data is processed. If our problem is the enhancement of dialogue, then timing as a function of information exchanged is one way of mapping two systems so that they may be synchronized in the sense of reaching critical change points in a way that allows optimal dialogue. It is the need to meet the requirements for control of man-machine dialogue that makes pursuit of the concepts of time graining, time driving, and information time important. The approach to complexity by using the time dimension as another dependent variable promises to bring the hard and soft sciences closer together.


It would have been only recently that this discussion of dialogue would have been another discussion of the importance of the holistic approach I have always thought of as holy. This is no longer true. The modeling of information exchange in the time domain I have presented is a primitive start in thinking through practical problems. The bottleneck in making computer resources available to man is the need to interface their communication so that it will grow in the style of a real conversation—as exploratory conversation, as learning conversation, where the associations and experience of each will be available to the other. The innovative use of computers depends upon solving this dialogue problem. Man must adapt to the advantages bestowed on him by his inventions—fire, wheel, atomic energy, and computers, among many—or die.

The computer already has begun to alter man’s science in a way that should please us who are used to complexity. But the dialogue of the soft sciences cannot be easily analyzed by the tools at our disposal, even yet, for we have no formal mathematics which can provide a calculus of relations. Our logic is as yet tautological and timeless. A statement of truth does not change because time passes. It does not decay or become obsolescent in a formal way though it does in the real world. Our power to develop the soft sciences as living sciences depends upon our learning to formalize. The computer’s ability to match human complexity, to synthesize the dialogue, requires only that we prime it with our beginning analysis of the features of dialogue.

As we learn to build computer programs that have a little more ability than those we know so far, to enter into a rich and complex dialogue with man, we begin to learn a little better how to understand what we have hidden in our art and our practice—the science of information exchange. Having to formulate the skill in timing shown to us during our psychiatric apprenticeship may assist us in developing a science which encompasses the dialogue. We need not be afraid of such a science. Knowledge of the dialogue is needed if our machines are to engage in rich and complex dialogue with us and our children. 


Brodey W.M. 1967. “The Clock Manifesto.” Annals of American Science 138 (1): 895–99.