Cybernetics is the science of communication and control in the animal and the machine. Many modern cyberneticians are constructivists. Cyberneticians and constructivists claim to deal with conversations and new ways to structure environments; to have ways of thinking which will change the world. Implying that cyberneticians and constructivists should teach in a way that restructures the traditional classroom. Or to put it more baldly, cyberneticians and constructivists should be consistent in their theory and practice.
I will examine the traditional language of grading and its effect on learning in the classroom. I will present my restructuring of grading in the cybernetic or constructivist mode. The grading design which results permits the emergence of conversations emerge which structure environments for learning.
As an instructor in the Cybernetic Systems Program at San Jose State University and in Engineering Management at Old Dominion University, I have seen the importance and the difficulties of facilitating learning. I would like to share some of my thinking with you in order to show how I use notions of cybernetics to structure environments for students learning.
I begin with a problem statement: How do I grade or test in a way that facilitates learning within the context of the university classroom? The answer to this question rests on another question: How can I answer to my university and colleagues with reasonable certainty that a student has learned the material for which I have responsibility? Implied within this latter question are these beliefs:
Current grading practices support a power structure. It does little to facilitate learning. It is based on a material notion of learning where the instructor "owns", knowledge, which she gives like an apple or a book to the student. It implies that knowledge is a thing and that it is always the same. It also implies that the student has the same relation to the "knowledge-thing" that the instructor has - both individuals are identical, and that they have identical notions of the "knowledge-thing". If, and only if, all these relations and people are assumed to be identical is it possible to "test" or grade by making the student give back the knowledge to the instructor in the same form.. Implicit in this model is the belief that the instructor as the perfect exemplar of what it is to "know" this thing. Thus, the student teacher-relationship is a power relation in which the teacher has the power and the student is the supplicant. The teacher can withhold the necessary knowledge or the necessary grade, placing the student in a state in which risk taking is dangerous.
Most instructors are honest people, aware that they are not perfect, that their knowledge is not perfect, and that the system as it currently exists is contradictory. Knowing these things they are put to great pains to deal with the contradictions.
Some instructors "grade up," giving better marks than the work warrants, or don't grade at all. Eventually this erodes the quality of education in the school and devalues degrees.
Some instructors who don't want to face the problems squarely. They either give fuzzy instructions and criteria that students are left without any direction, or resort to the "objective" true false test.
We have neither considered the effects of grading, nor created a way to talk about it. The problem is compounded by becoming invisible. I believe that cybernetics gives us a way to think about grading that will enable learning and will make the structure of classroom as something other than a "power trip".
To grade learning I must determine what is meant by the words "to learn". Learning is a way in which a system changes its structure in response to the experience of its environment. Or, learning is a response of a system to its environment and that this response takes place internally in the altering of the systems structure. If this is so, there is no way for an external observer to determine if a system has changed its structure. It is a black box. Therefore, I can not grade learning. Hence the despair when it comes to grading.
I've said that there is no way of knowing anything about the internal state of a student so I cannot grade. But, I do know when a person knows something - not perhaps in a "scientific" manner but in a colloquial manner.
For example, if you and I have a conversation then I know that you have learned the English language. Let me "unpack" this statement. In the course of conversation I am able to observe that you, as part of my environment, are behaving in a way which I associate with knowing the English language. In a like manner, my response to you in our conversation would imply that I knew the English language. This implies:
So, I can observe learning only as a result of interaction. Let us use conversation as a model of interaction.
In conversations the content is derived from the context. Participants in a conversation interact such that the way in which they use words becomes more and more clear. For example, if I say the word apple" I may be thinking "red", "mackintosh," while you may think "green", "granny smith". As we continue to talk and I refer to the redness, the mackintoshness, the tartness and the suitability for pies, over time you have a better picture of my meaning of "apple". The participants converge upon the use of the word "apple" within the context of their conversation.
Equally, the context arises around the content. As the participants converse, each interacts not only with the words being used, but also with the various verbal and non-verbal cues reflecting their own internal state, and the perceived state of the relationship. As the interaction continues, what emerges is a clearer and clearer sense of: what we mean to each other; if we are friends or enemies; that we seem to have similar assumptions or ways of thinking; whether we understand what the other is saying. A conversation is a process of mutually causal interactions, in which content is shaped and formed by the context, and context is judged and interpreted through the content.
In order to understand this better I want to focus on one conversational round. The speaker says the word "apple", drawing a distinction with which she wishes to indicate the experience or set of experiences which she associates with the word apple. If the speaker observes that the listener is not behaving in a way the speaker associates with understanding, the speaker will attempt to indicate with more words and gestures what she means by "apple", e.g., "red", "round", "sweet". In trying to orient the listener toward her meaning the speaker must examine her own set of experiences and associations surrounding the word "apple" in the hopes of bringing forth or indicating a similar set of experiences for the listener. By this examination the speaker has clarified for herself what she means by the word "apple".
The listener, by understanding or not understanding, has controlled the interaction such that the speaker has either continued speaking, or has retraced and reexamined the associations and experiences indicated by the word "apple". The speaker indicates her own state and the listener coordinates his behavior to a state similar to that of the speaker. Thus in any one round of the interaction it is the listener who controls the conversation.
The implications are these:
Now I have a hint on how to proceed. If I want students to learn, I want students to become the speakers in a conversation coordinated by the instructor as listener. It is as listener that the instructor perturbs the students' environment such that they learn.
Now, although this discussion seems to have handled the large and messy issue of learning it takes me no further on the original issue of grading. It also leaves me with an additional problem: how do we structure or perturb an environment such that students become speakers in a conversation which deals with a chosen subject, one that covers the things the university requires their knowing.
In order for students to become speakers, I need to structure an environment in which they perceive a reason to speak about the specified topic. That environment must also be such that students feel safe enough to speak. There are a number of things I can do which I believe will achieve that environment.
First, I can perturb the environment to encourage students to speak by simply not speaking utilizing the notion that the listener controls the conversation, the instructor becomes a listener. This means that as instructor in class, I do not speak except in reply or in conversational interaction that students have initiated. I have observed, both as a student and as someone responsible for structuring environments for students, that this perturbation is effective for initiating conversation.
The next thing I want to do is structure the environment in order to constrain the conversation to topics appropriate for the course. One way in which this can be done is to specify goals and objectives for the student to reach, supply them with the appropriate materials, and let them have their heads.
Given the above discussion on learning, I want to phrase the objectives behaviorally since it is impossible for me to know a student's internal state: I want the student to behave in ways which I associate with knowing the material for which I am responsible. And I want these behaviors to be iterative. This is important for a number of reasons.
If I want a student to change her internal state, I want her to serve as environment and observer of the interaction, and by so doing, to examine her meaning by speaking. The more opportunities the student has to interact in this way, the more will learning occur.
My internal meanings are unobservable to my students, so for me to clarify my meaning, and for them to clarify their meaning, we need repeated conversational rounds in which to coordinate our actions. In this way we can produce a mutual context in which we interact. In other words, for us to converge upon agreement successive conversational rounds are needed.
I want the goal to be a complete thing. It is absurd to try to complete 75% of a goal: I want to evaluate behaviors on a presence/absence basis. If I as a cybernetician believe that a system with feedback is self-correcting, how can I give a grade of 75%? Is a guided missile successful if it comes close? In my observation of the world of work, workers are not able to tell a boss that they have done the job when only 85% is done.
In everything said above, it seems that learning and conversation are processes. Part of the difficulty in grading is that a grade is static: grade is a product, a "thing" which a student obtains, rather than a process. It doesn't seem to be "feedback", since the notion of feedback implies some correction or change of the trajectory. Grades are not changed by a change in behavior.
If grades are a product then I have to ask, "What is the nature of the product and how is it produced?" Grades are not a consequence of behavior in the sense that a wood carving is produced as a consequence of a particular behavior on a piece of wood. Rather, grades are a symbolic assignment of value, much as pay or salary is a social and economic assignment of value. I would say, then, that grades are a symbolic representations of my valuation of the behaviors observed in interactions in the course of the class.
Symbolic systems are by their nature arbitrary. However, since I am the arbitrator, I can break out my observations of how various grades correspond to various behaviors, make those observations explicit, and attempt to act according to my observations. Since the valuation is my subjective assignment, it is incumbent on me to produce a grading scheme which I can make explicit - - to encourage students to suggest other valuation schemes and to hold myself accountable for the value it engenders. I believe that being explicit is the only way to take responsibility for the subjective nature of grading.
As I think about what I associate with various learning levels I perceive some qualitative difference in the way people interact with me, and their environments, which I associate with differing degrees of knowing. In common terms, there are behaviors which I associate with different degrees of understanding. There should be some way of designing behavioral objectives for evaluation to reflect this difference.
The next activity I associate with learning is more complex: out of that set of distinctions, I perceive that things are similar and different. In other words I learn how to distinguish among distinctions, I can compare and contrast distinctions.
The next activity I associate with learning is the most complex: I can begin to form theories and apply this set of learning to other contexts.
I have now assembled a number of guides for my grading scheme. I want:
The first, Systems, Society and Technology, is an introductory, upper-level, undergraduate course and is a pre-requisite for other courses in the Cybernetics Systems minor. It covers various systems and cybernetic concepts and requires the student use Peter Checkland's Soft Systems Methodology. The second San Jose course, Applied Systems Science, has Systems, Society and Technology as a pre-requisite and teaches ways in which workers have applied cybernetics and systems in the "real world". Because most of our students are business majors we taught Stafford Beer's Viable Systems Model and Russell Ackoff's The Art of Problem Solving. I also co-taught Engineering Economics with Christoph Berendes at Old Dominion University and used this design scheme. Other instructors at San Jose State have used the design scheme in their courses successfully.
In each of the two San Jose cybernetics courses the objectives were given out at the beginning of the term. I also gave out the finals in each class within the first two weeks. The final is the 'C' level requirement in Systems, Science and Technology. In Applied Systems Science the final and exercises in the text are the 'C' level objective. The point of giving out the final in the beginning is to demonstrate that everything that needs to be done is in the student's hands at the beginning of the course. I tell them that it is up to them to get the answers in any way that seems reasonable. I tell them that if they can pass the course at that moment they are welcome to do so. I also give out the readings which I think will cover the material and then let them have the course.
The 'B' objective in Systems, Science and Technology is to do a Checkland model for some system of your choice, do a model of the same system from the point of view of your major field of study, then compare and contrast the two models as to their approach, questions asked and results obtained. The 'B' objective for Applied Systems Science is similar: do a Viable Systems Model of some system of your choice, do a Checkland model of the same system, then compare and contrast the two models as to their approach, questions asked, and results obtained. The point of the compare-and-contrast assignments is to help the students interact with the system in two distinct ways, be aware of how she approaches the same system differently in different contexts and how changing contexts affects both the interaction and the perception of the system.
For the 'A' objective in both classes the student can choose either to apply the concepts covered in the course to a project assigned in consultation with the instructor or to attend four extracurricular events and write a short paper showing how systems and cybernetic concepts were illustrated in each. The point of the 'A' objective is fairly transparent. It requires the most initiative by the student and allows a highest degree of freedom since the only limiting factor is the use of class material. Because it is the least structured, it is the most difficult.
In the engineering economics course at Old Dominion we were faced with a new constraint. The students use this course to help prepare for the Engineer in Training (EIT) exam. This exam is a series of short-answer, multiple-choice questions dealing with a number of aspects of engineering education, one of which is engineering economics. Because of this constraint, and in order to preserve the conversational aspects of the design, we divided the course into a series of units. In order to get a 'C', students were required to pass a test, patterned on the EIT. Each test consisted of three questions on the subject of the unit. Two of the three questions had to be correct to complete the unit. Students took as many tests for each unit as needed. Review sessions were held before each "make up", run only on student questions.
The 'B' level objective was a straight-forward financial analysis of some decision, then pointing out the limitations of that analysis in terms of "real world" concerns. This allowed application of the compare-and-contrast assignment to a "conversation of financial analysis" and a "conversation of the real world".
The 'A' level objective in Engineering Economics was a complete analysis of some decision which the student had chosen from the "real world". This had to include research on alternatives, financial and non-financial analysis of each of the alternatives, and a recommendation with a stated rationale. It also had to point out the assumptions and limitations of the approach and the data. In addition, the student had to present the report in a real situation and then report on how the report was received both in form and content. In short, the student assumed the role of financial consultant, did an analysis, and evaluated the results.
Needless to say, I still don't know if my students learn (since learning still requires change in an internal state). What I can say is that after the first few class meetings, I have never had more interactive classes. Students were at first uncertain of how to proceed without direction, and it's difficult not to jump in and save them with a lecture. The first questions are all about class structure, about what they really need to do, and about what this or that objective means. This is all part of defining context and trust. Then as the class proceeds and the first real questions begin to surface, the conversation begins. The first term I used this design, I had the first draft of the final due half way through the term. This was too broad a range for good conversation, and the class decided to have different topics due each week. The next term I broke my six- question final into three parts with associated readings. This has helped focus the conversation and the readings. It also means that I don't have to be prepared for questions across the entire range (although this can still happen though since there is no reason for a person to address only the section which is due next).
A serendipitous consequence of this approach is as the class converses they ask questions that are not on the final, but are implied by the ideas they are studying. The conversation separate from grading because the student can pick her own grade by electing to do all three levels or only to do one or two. The uncertainty of what will be asked is gone since the final is part of the conversation. In the Engineering Economics course students felt free to question us on our method, and even to question the way in which university classes are structured in general. In short, conversations extended beyond the course and course content.
Over time, it became clear to us that the difficulty of a course like Engineering Economics is not in the material per se but in exposing students to "business conversation" and language. It pointed us in the direction of talking about what constitutes persuasion or proof in a business environment as contrasted to an engineering environment.
Finally, I'm still not listening as much as I would like in class. And students aren't speaking as much as I would like. But I become more and more convinced that we're on the right track because it is a conversation. I'm getting better at perturbing the environment by listening, so that people can become oriented toward their own learning by speaking.
Any paper or idea owes much to the conversations which have preceded it. Some of these conversations are face-to-face, personal, and some are interactions with the printed page. I have written this paper without citations because I have lost the individual threads in the process of making the braid and because Heinz vonFoerster challenged me to do so. However, I still want to acknowledge some of the people who have influenced my thinking and whose ideas are in this work uncited.
Maturana, Humberto R. 1978 Biology of Language: the Epistemology of Reality, in Psychology and Biology of Language and Thought: Essays in Honor of Eric Lenneberg, G. Miller and E. Lenneberg, eds. NY, Academic Press.
Pask, Gordon 1979 Consciousness, in Journal of Cybernetics 9:311-000.
von Foerster, Heinz 1981 Perception of the Future and the Future of Perception in Observing Systems Seaside, CA, Intersystems.
In addition, many people have structured my environment such that I have been able to change my structure, to learn. Some have done this by example and some by conversations. I want to thank:
Chris Berendes for ideas on grades as a product (1987);
Brookdale Community College which provided my first acquaintance with Behavioral Objectives (1970-1972);
Warren Shapiro who taught one of the best classes I ever had: the first week he assigned a reading the second week he asked if there were questions, and when there were none he went home. For the rest of the term we asked questions (1973);
and Heinz von Foerster who assigned the form of this paper, gave feedback and defined the constraints (1987).