Professor Koen is noted for his cutting edge points-of-view and this session promises to be a very interesting discussion of a very timely topic—defining and teaching engineering ethics.

Billy V. Koen has been a mechanical engineering professor at The University of Texas/Austin since 1968. He served from 1988-1993 as Vice President of ASEE and has held 25 different positions and is an ASEE Fellow. He is the author of Discussion of the Method: Conducting the Engineer’s Approach to Problem Solving (Oxford University Press, March 2003) and Definition of the Engineering Method (ASEE, 1985). He has received Olmsted, Chester F. Carlson, Centennial Medallion, and W. Leighton Collins awards from ASEE. He received the Helen Plants Award in 1994 and in 1985, and the Benjamin Dasher Award in 1984 and again in 1985.

To provide some background material for this presentation, two of Professor Koen's most recent papers are available online. These papers were presented at the Nashville ASEE Conference in June 2003:

1. I know you have been concerned with the definition of engineering and engineering ethics for a long time. Perhaps a good way to begin would be for you to give an overall outline of your views. We can then define the basic terms and expand on your central themes throughout this forum.

First let me thank you, Rob, for asking me to participate in this forum. IEEE, in general, and you, in particular, should be congratulated on this innovative use of the Web. You and this forum will give me an opportunity to develop my ideas with the help of the most knowledgeable engineers, worldwide. I hope they will help by adding their own views and asking questions. I’ll give hyperlinks to amplify some ideas and provide references for those who might be interested.

And now to your first question.

I make the following specific claims about engineering ethics:

A. The engineering method and the use of heuristics is an absolute identity. Nothing the engineer does in his or her role as an engineer is outside of the direct influence of a heuristic.

B. The engineer's notion of 'best' is fundamentally different from the classical Platonic one.

C. An engineer should be judged against the state-of-the-art that represents best engineering practice at the time the design is made.

D. Ethics, ethical judgments, and ethical systems are all heuristics on a par with the technical heuristics found in engineering and must be totally integrated into the state-of-the-art to determine best practice.

E. How engineering is defined and how ethics is understood fundamentally affects how engineering ethics should be taught.

2. Based on the first item on your list, what do you feel is the basic nature of engineering?

No serious consideration of engineering ethics can proceed without a commitment as to what engineering is. It is hard to see how an engineer and his or her behavior could be judged as ethical or unethical without knowing what behavior we are talking about. For many years now, I have claimed that the basic nature of engineering is captured in the following definition:

The engineering method is the use of heuristics to cause the best change in a poorly understood situation within the available resources.

A first cut at a definition of the word heuristic in this view of engineering is:

A heuristic is anything that provides a plausible aid or direction in the solution of a problem but is in the final analysis unjustified, incapable of justification, and potentially fallible.

Both of these definitions have recently appeared in my book, Discussion of the Method: conducting the engineer's approach to problem solving, published by Oxford University press (which I’ll refer to during our forum as DOM) (see DOM reference below), and have been among my active areas of research for over 35 years. This definition of engineering is to be understood not as applying to engineering only from time to time but as being an absolute identity. Hence my claim:

All engineering is heuristic.

Representative engineering heuristics include heuristics that aid in the allocation of resources such as the admonition to “allocate resources to the weak link," others that control of the amount of risk to be assumed such as the admonition to “give yourself a chance to retreat," and still others that indicate the appropriate attitude for an engineer to take when encountering a problem such as the admonition to “work at the margin of solvable problems." In addition, the technical engineering heuristics include all of the orders of magnitude, graphical correlations, and mathematical equations used throughout engineering practice.

It is worth noting that engineers in some cultures do not have the word heuristic, but most have terms for its near engineering synonym rule of thumb. As examples, we find that in France, le pif (the nose); in Germany, Faustregel (the fist); in Japan, menoko kanjo; and in Russia, na paltsakh are used as equivalent terms for the heuristic or rule of thumb (DOM, page 34 (see reference list below)).

I'm interested in finding additional examples of the English term rule of thumb expressed in other languages. Is there a Chinese engineer, an Italian engineer, or an engineer from some other culture participating in this forum who would be willing to help me out?

Obviously all conceivable engineering heuristics do not apply to all designs at all times. It is convenient to use the symbol, sota| _{design; time} to represent the set of heuristics or state-of-the-art used in a specific design at a specific time. All other definitions of engineering method can be subsumed into this one as additional engineering heuristics (see DOM1 below).

References for Answer 2

• DOM - Koen, B.V., Discussion of the Method: Conducting the Engineer’s Approach to Problem Solving, Oxford University Press, March, 2003, ISBN 0-19-515599-8
• DOM1 - Koen, B.V., Discussion of the Method: Conducting the Engineer’s Approach to Problem Solving, Oxford University Press, March, 2003, ISBN 0-19-515599-8, pg. 28.

Rob, you have correctly identified the key component of the definition of engineering method that we will need to understand ethical engineering behavior.

For an engineer, best means optimum. Thus we hear engineers say “we have designed the optimum television set given the resources we have” to mean that “we have designed the best television set given the resources we have.” What follows is somewhat tedious, but it is important for an understanding of the engineer’s notion of best which is essential to engineering ethics.

Figure 1—Sharpness vs. Knob Setting

Recalling the technical terms used in optimization theory and borrowing the figures given in DOM, page 17 (see references below) will help us define best. Consider a television set with one control. We will assume that turning this knob to a higher number will produce a better picture but at the same time worsen the sound; turning the control to a lower number, on the other hand, will worsen the picture but improve the sound. It is relatively easy to adjust the knob for your personal preference as you balance the relative importance of picture and sound. The quality of the picture (say, sharpness) and the quality of the sound (say, fidelity) will be called the criteria for the problem. It is against these two criteria that a judgment is made as to whether or not the control setting is best. In general, the criteria are conflicting in that an improvement in one implies a worsening of the other. The criteria taken together make up the optimization space or axes system of the problem. Figure 1 is a graph that shows how the sharpness of the picture might change with the position of the knob. The dotted line indicates that a setting of four corresponds to a sharpness of eight. The sharpness of the picture is higher for a setting of six than for one of two.

Figure 2—Fidelity vs. Knob Setting

A different return function, Figure 2, exists for the sound. Turning the knob to a higher setting decreases the fidelity of the sound. When the control knob is set to the value four, the value of the fidelity is also four as can be seen on the graph. To represent pictorially what a person does instinctively when she selects her preferred setting, the two return functions must be combined or superimposed on the same graph; they must be put on the same basis; or, technically, they must be made commensurate. This requires that a common measure of goodness be found for picture and sound. What the engineer seeks is the relative importance of sound and picture to the owner of the television set. A ten-percent increase in the sharpness of the picture is worth what percentage decrease in the fidelity of the sound? This relative importance is expressed by the weighting coefficients of the two conflicting criteria. Another term used for the set of weighting coefficients in the value system of the problem.

Figure 3—Objective Function vs. Setting

Let us assume that sharpness and fidelity are equally desirable. That is, the relative weights in the two cases are equal. The resulting combined graph is given in Figure 3. The bottom two curves are the return functions for the conflicting criteria that were shown in Figure 1 and in Figure 2. The upper dotted curve is the sum of these two under the assumption that an improvement in picture and sound are equally desirable. Verify that this combined dashed curve with a value of twelve at setting four is equal to the sum of our previous reading of eight and four at the same point. This dashed curve is the common measure of goodness for the problem. It is sometimes called the objective function or measure of system effectiveness (MSE). We want the largest or maximum value of the objective function. This number is called the optimum or best value. For our television set, the optimum setting corresponds to a control setting of the knob or independent variable of four.

Obviously, in a more realistic case, there will be many more criteria to be considered and often the operation will not be carried out mathematically. Engineers seldom do the suggested analysis explicitly; but they always do it implicitly. To use my terms, they do it heuristically.

We can say mathematically that:

where: MSE is the objective function or measure of system effectiveness, t is the independent variable or knob setting, and f(t) are the individual criteria and return functions. The weighting coefficients give the relative weights of the criteria and commensurate the functions.

With the engineer’s notions of the sota (or state-of-the-art) and best in hand, it is easy to understand how we should evaluate an engineer’s performance.

The rule of judgment in engineering is to evaluate the engineer or an engineering design against the sota that defines best engineering practice at the time the design was made.(see DOM2 reference below)

Using the nomenclature given in an earlier question, this is equivalent to insisting that the engineer be evaluated against sota|_{best eng prac; time}. Anything other than this is a call for a clairvoyant engineer. It is simply unfair to evaluate an engineer against information that is available now but was not available in the past when the design decisions had to be made. And to the critic who argues that the engineer should always anticipate future uses of engineering products, I insist that the ability to project any reasonable use of a product into the future is simply one more heuristic in the sota|_{best eng prac; time}.

This standard is a relative one based on the best thinking of the engineering community (viewed in its broadest sense) as opposed to an absolute one. It has the advantage of permitting action in the face of uncertainty instead of being paralyzed by the fear of being wrong.

I do not deny, at this point, that the possibility exists of a future absolute standard. Time permitting, I will return to this important consideration at the end of our discussion. In answer to the unvoiced question “but how is this standard to be established?" Once again, I'm sure you can anticipate my answer: Heuristically, of course.

This proposed basis for judging the ethical behavior of an engineer has an advantage of allowing us to develop a strategy for solving some of the thornier problems that arise in engineering ethics. An example will make this clear.

Consider the problem of the whistleblower. The first observation that must be made is that the whistleblower is looking for the best solution in a different optimization space from the one being used by company management. As we saw earlier, completely different optimization spaces or optimization spaces that only partly overlap produce different answers for the optimum. The response to the whistleblower is completely equivalent to the resolution of all standard engineering design problems. There is a trade-off between squelching a whistleblower and losing the potential benefit of the heuristics important to him or her and squandering precious project resources to appease the whistleblower whose axis system is missing heuristics that are essential to the success of the project. The answer to this dilemma, of course, is to develop sota|_{best eng prac} that includes specific heuristics to resolve this trade-off.

Just as no serious consideration of engineering ethics can proceed without a commitment as to what engineering is, no serious consideration of engineering ethics can proceed without a commitment as to what ethics is.

Ethics, ethical systems, even the very existence of the notion of ethics—indeed all of ethics—must be taken as heuristics.

Let us be truthful—an admirable, but often neglected ethical virtue, when it comes to talking about ethics, by the way. Ethics is a mess. Philosophers have been studying the matter for at least 2500 years and they seem little closer to producing an absolute notion of what it is or even if it is. Is the conscientious engineer trying to do the right thing to believe the intuitionist, the empiricist, the rationalist, the hedonistic, the instrumentalist, the situationalist, the pragmatist, or the emotivist (if this last is the correct word for Ayer’s view of the world)?

There seem to be as many different ethical systems as there are ethicists. Indeed there may be even more, for surely there is a prolific ethicist somewhere who has come up with more than one ethical system. Restricting ourselves to some of the current literature of engineering ethics alone, we find textbooks discussing value ethics (see Value reference below), duty ethics (see Duty reference below), and rights ethics (see Rights reference below) to mention only a few of the better-known ethical systems. This variety alone should make one suspicious that all of ethics is heuristic.

Unfortunately the notion of ethics is far more tenuous. In trying to determine what is to be taken as ethical behavior, an engineer gets a different answer depending upon the chosen system. For example, the author who wrote the book on value ethics cited above reminds us that

After this selection [of principles and methods], a specific range of right action appears . . . Different sets of principles and methods yield different ranges that often overlap only partially (see Value reference below).

One of the characteristics of the heuristic is that two or more heuristics may contradict each other. And to give a final reason why we should consider ethics as a heuristic, ethicists have had to invent the notion of prima facie rules when it became obvious that many of their rules of conduct were not absolute but could be overridden by other rules in special cases.

In her book on ethics, Caroline Whitbeck (see WHITBECK reference below) argues that we should relate ethical problems to design problems by analogy. Specifically, she says:

Engineers encounter difficult ethical problems in their practice and in research. In many ways, these problems are like design problems. They are complex and often ill-defined; resolving them involves an iterative process of analysis and synthesis; and there can be more than one acceptable solution.

Although I agree with this sentiment, my position has always been considerably stronger. I feel that ethical problems are not just like design problems in some aspects so much as they are identical to and part of design problems in all aspects.

Furthermore I contend that appropriate ethical heuristics must be totally integrated into the technical heuristics to make up the basis of best engineering practice. In my terminology, sota|_{best ethics prac; time} is a fundamental, essential part of the sota|_{best eng prac; time} . This sota must contain all appropriate ethical, as well as technical, considerations. Therefore, in determining best engineering practice in a specific instance appropriate weights must be given to the traditional theories of ethical behavior such as virtue ethics, duty ethics, rights ethics, etc. completely on a par with the technical aspects of the problem.

It is easy to show that ethical heuristics do not stand apart from the technical heuristics to be tacked on at the end of the engineering design process to placate ABET. In his classic book, Science and Human Values, Dr. Bronowski (see SCIANDHUM reference below) argues that “the practice of science compels the practitioner to form for himself a fundamental set of universal values.” He then gives examples—one of which is the necessity of the virtue truth in t he operation of science—as in the quotation “the sanction of truth is an exact boundary which encloses him [the scientist], in a way in which it does not constrain the poet or the painter.” Likewise, I would argue that it is hard to see how engineering design could ever be practiced without the classical virtue of truth applied throughout the entire process. The state-of-the-art of the engineer must be permeated with a sense of the heuristic truth to exist at all.

Other reasons to believe that ethics is inherent in the overall state-of-the-art are less esoteric. In practically every engineering design project there are fundamental trade-offs between the technical considerations and such obvious ethical considerations such as human life, happiness, and well-being. One well-known example will make this clear. According to Professor Machemail of the University of Texas at Austin, in 2004 a good rule of thumb for the cost of constructing a controlled access highway is approximately $5 million per mile. The rule of thumb for an average highway without controlled access is, perhaps, $1 million per mile. The optimization space for the decision as to which kind of highway to build in a specific instance contains a large number of technical considerations to be sure, but it also contains essential trade-offs between them and the number of lives saved, convenience for the traveling public, severity of accidents, and so forth. These latter trade-offs are an integral part of the optimization space.

I hope that participants in this forum will send in their examples of trade-offs between specific ethical heuristics and technical ones. This would be very helpful to me in my future research.

The ethical considerations cannot be just "add ons" at the end of the design process, but must be treated simultaneously with the technical axes. We saw earlier that, theoretically, the engineer cannot find the optimum of the various parts of the overall measure of system effectiveness (MSE) and then hope to combine them to obtain the overall optimum. In the present context this means that since the ethical considerations must be completely integrated into the state-of-the-art that represents best engineering practice, we cannot find the optimum solution from the point of view of the technical heuristics and then find the optimum solution from the point of view of the ethical heuristics and then by some procedure combine them to determine the overall optimum solution for the project. The two spaces are intimately interwoven and must be treated together.

Although the ethical axes are the ones under consideration in this forum, it is worth noting in passing that they are not the only axes that are often neglected in engineering design. For instance, different cultures place different weights on the aesthetic axes. Since it is undeniable that the human environment affects the health and happiness of humans, I have always argued that aesthetic heuristics must also be integrated into the sota| best eng. prac; time and given the appropriate weights. Likewise those of politics, cultural differences, and gender diversity must be included.

In answer to your question, “What is the relationship between engineering and ethics?”, I must respond: total integration to make up sota|_{best eng prac; time} against which we intend to judge an engineer’s performance.

Past development of the technical heuristics and the ethical ones have not been equivalent, however. We expect the engineer or professor to be knowledgeable of the state-of-the-art in the technical areas and to reflect best engineering practice in the chosen technical area. If he or she uses the Colburn relation to calculate the heat transfer coefficient, the clear implication is that its use is consistent with the view of, say, 90% of the interested parties (in this case, the practicing engineers and engineering professors) in a similar situation. We expect engineers to keep up with the literature, to attend appropriate conferences, and to perform research in the chosen field. Typically this is not the situation with the heuristics of engineering ethics. We cannot assert with equivalent confidence that at least 90% of the interested parties (practicing engineers, members of society, engineering ethicists, administrators, and whistleblowers) would agree on the ethical behavior taught in ethics courses. Engineering professors ordinarily do not keep up with the ethics literature, do not attend conferences on ethics, or perform research in the area of ethics. Students will leave unsatisfied when it comes to knowing how to become consistent with the ethical aspects in the state-of-the-art of best engineering practice in ethical areas.

This indecision and equivocation is a plausible reason for the feeling among too many students, professors, and practicing engineers that the commitment of colleges of engineering, departments of engineering, and societies of engineering to engineering ethics is, at best, superficial. In one extreme case, a competent, well-prepared, enthusiastic professor taught an ethics course that was labeled by the students as “The course from hell.”—a truly ironical comment for a course in ethics.

There is always uncertainty as to what is to be taken as best engineering practice in technical matters; similarly, there is always uncertainty as to what is to be taken as best engineering practice in ethical matters. To the extent to which it is practicable to judge an engineer's technical ability against best engineering practice in the technical arena, it must be feasible to evaluate the engineer's performance ethically against best engineering practice containing the equally important ethical components. If the axes of the optimization space are ethically chosen; if the commensuration of these axes is ethically performed; and if the resulting optimization is ethically implemented, it is hard to see how an engineer could be criticized as unethical. My position may now be succinctly stated as follows;

A sufficient basis for practically judging the ethical behavior of an engineer is the optimum in a multi-variant space which contains both the technical and ethical heuristics that represent best engineering practice appropriate to the problem under consideration at the time the design is made.

What we most desperately need now is a well articulated sota|_{best ethic prac; time} based on the best thinking of the engineering community. (see NASHPROF reference below) Armed with this, the engineering professor can go to work to develop ethical engineers for the future.

References for Answer 6

Simple. We need only argue by analogy with the way the technical axes are determined. The technical knowledge that created the Golden Gate Bridge did not spring full-blown from the Minerva’s breast. From the first log that chanced to fall across a stream and allow the safe passage of a human without getting his feet wet - sota|_{best eth prac; time of the cavemen} - to what is surely one of the most famous bridges in the world - sota|_{best eth prac; the present day} - was a long process of successive approximations, iteration, correction, and feedback—all good engineering heuristics for making improvements in the state-of-the-art. And so it should be with the ethical engineering heuristics.

At the present time, engineers are fortunate. Most people seem to understand that engineering is a risk taking endeavor and they do not criticize engineers too harshly when an engineering product fails. The public now holds that engineers are basically ethical. This advantage is not enjoyed as much by other professions such as lawyers, car salesman, and, more recently, businessmen and news anchors. In addition, engineers enjoy several natural additional advantages. First, most engineering societies have well-established codes of ethics that can serve as a first cut at establishing what is to be taken as ethical behavior.

A second natural advantage in establishing best ethical practice, often overlooked in books on engineering ethics, is that engineering ethics is a subset—a relatively small subset—of all ethics. This makes our task simpler because we do not have to consider all esoteric ethical theories in detail to have a satisfactory approximation to best practice. For example, different religions disagree in important respects as to what constitutes ethical conduct and society is currently struggling for a coherent policy on abortion, euthanasia, and stem cell research. By and large engineers are not called upon to make professional pronouncements in these areas. Although a course in ethics in the philosophy department must confront these issues and examine the theoretical bases of competing claims, this is not the purpose or obligation of a course in the engineering department. The requirement to teach only ethical conduct with respect to engineering behavior greatly simplifies our task.

A third natural advantage of the engineering profession is that engineers are typically asked to “play down the center the field” ethically speaking. The job of a college of engineering is not to teach students to skirt common ethical standards, but to observe the obvious ones as determined by the profession. With tongue-in-cheek I might add that it is not clear that the state-of-the-art for the ethical practice of law enjoys this same advantage.

The fourth and final natural advantage in determining best ethical practice is that the engineering profession already has a well articulated strategy or set of heuristics to modify the technical heuristics in the state-of-the-art by evolution. It can easily be applied to the evolution of engineering heuristics as well.

Engineers must not bask in the high esteem to which the profession is currently held and allow their Enron to happen.

All we need to do now is to modify our currently acceptable sota by “making small changes,” “allocating resources to the weak link,” and “giving ourselves a chance to retreat” as we develop more sophisticated ethical heuristics. The Accreditation Board for Engineering and Technology (ABET) has mandated this process when it established Evaluation Criterion 3: Outcome #f signaling a renewed interest in instruction in ethics at colleges of engineering in the United States. (ABET reference below) Outcome #f specifically states that “Engineering programs must demonstrate that their graduates have an understanding of professional and ethical responsibility.” IEEE has opened a conversation on engineering ethics by establishing this forum. And, finally, the National Academy of Engineering (NAE) is looking for ways to improve the knowledge of the American engineer in ethics. The state-of-the-art that represents best engineering practice in ethical matters in engineering is in good shape at the present time and we can have every expectation that it will evolve and improve in a satisfactory way into the future based on the model of the engineering heuristics that guide the evolution of technical heuristics. (see CHALLPROF reference below)

To use a good engineering heuristic: determining a procedure that encourages the engineering community to develop the sota of best ethical practice is the “weak link” in creating ethical engineers where we should allocate our efforts for maximum effect. Should we connect large groups of the engineering community over the Internet during Engineer’s Week? Should we engage the engineering community in voting on specific engineering ethical situations as is done on some web sites? What heuristics should we use to resolve this issue? I'm confident we're up to the task in this crucial design problem. Care to brainstorm?

References for Answer 7

Designing a strategy to produce more ethical engineers is no different from designing a bridge. The engineering method applies to both. A good heuristic in both domains is first to define the problem we are trying to solve. What is our goal; what is our problem?

We do not want students just to be “sensitive” to ethical problems in some unspecified sense; we do not want them just to be able to identify historical ethical systems; we do not want them just to be consistent with some vague, ill-defined definition of good or truth; and we do not want them just to propose alternative interpretations of engineering situations from the points-of-view of different ethicists.

All of these results may be welcome as by-products of our efforts, but our central focus is on action. Ideally we want one engineer, at one moment, working on one project, to make one judgment—then the next—the next—and the next—such that a consensus of creditable engineers, managers, laypersons, judges, commentators, and whistleblowers will all label them as ethical. Stated more succinctly, we want the engineer’s actions to be consistent with sota|_{best eng practice}.

The strategy to teach a student about engineering ethics is quite different from the one to teach a student to be an ethical engineer. Many, if not most, traditional engineering ethics courses and books on engineering ethics emphasize the former and teach about engineering ethics by reviewing some of the traditional ethics literature from antiquity to the present day and by examining a variety of arbitrarily selected, poorly authenticated Case Studies.

Instead of this approach we will take as our goal, educational objective, or outcome: when the course is completed, the student will be able to identify ethical behavior based on current best engineering practice with the expectation that they will then act ethically. In this way, we achieve our ultimate goal of developing engineers who behave ethically.

Your question asked how theoretically we should teach ethical behavior. As is true in all good design projects, let us take a short digression to survey the relevant theory.

Both identifying what is ethical and acting on that knowledge are human behaviors. They are activities an individual does. Theoretically if we are to claim that an individual is behaving ethically, we should be able to identify specific actions the individual takes that we identify as ethical. The analysis proposed here is based on a theory of learning or modifying behavior developed by B. F. Skinner called behaviorism. (see SKINNER below)

Behaviorism (or reinforcement theory as it is sometimes called) is based on the Thorndike’s Law of Effect. This law asserts that behavior is modified by its consequences. Therefore, to modify an engineer’s behavior to make it more ethical (or to maintain the ethical behavior we now see), we must reinforce (reward) good behavior and ignore bad behavior. Theory predicts and experiment confirms that this strategy increases the probability that the individual will exhibit the desired behavior in the future.

Theoretically, then, we want an engineer to be able to identify ethical behavior and, secondly, to act on this knowledge. In the analysis to follow, we will need two common words generalization and discrimination used in their technical, behavioral connotations. A specific example will help make them clear.

The theory is unambiguous and unequivocal. To create an ethical engineer we must reward the engineer when he or she is presented with and acts appropriately in a situation that is identified as ethical. The engineer will then generalize to other situations that bear some resemblance to the one now recognized. If a new situation that is similar is subsequently labeled as unethical, the student learns to discriminate ethical behavior from unethical behavior. For instance, all of us can distinguish between a dog and a horse. We learned this behavior so long ago that most of us have forgotten how we learned it. Our parents must have rewarded us with parental approval and said “good” when we first identified a dog. Theory predicts that the child will try to generalize this knowledge to get the reward in future, sometimes inappropriate situations. To quote the well-known learning psychologist, Dr. Keller, (see KELLER below):

In everyday life, examples of generalization are so common that they go unnoticed. They are most obvious, perhaps, in children, where they are often amusing. Parents smile at the child who calls out “Doggie!” at the sight of a horse, a cow, or some other four-legged creature; or they may laugh to hear a child say that soda-water “tastes like my foot’s asleep.”

What we are seeing is the first step in concept formation called generalization. The second step is to make distinctions between the dog and the horse by discrimination. For this step to take place, we must present examples of dogs and horses randomly and reinforce the correct response. If you want a behaving organism to learn a concept you must present a large number of exemplars with and without this concept and reward generalization and discrimination.

Note well several important characteristics of concept formation. We will need them later. (1) The samples we present when we are training must be dogs or horses. If we suddenly present a red herring instead of a dog or a horse under the guise of teaching the difference between dogs and horses, the essential generalization will be weakened. (2) It will not do only to present the child with dogs. You cannot learn to distinguish between dogs and horses if you never see a horse. (3) The parent must know for herself or himself the difference between a dog and a horse. (4) A time stamp must tell when the learning took place. The definition of a horse has changed over time. In the distant past, horses were the size of the modern dog. It would be a pity for a budding paleontologist not to recognize a horse from the past because our faulty training failed to tell her or him when our definition of a horse was appropriate. If our training set does not have these characteristics, the child is left flailing about trying desperately to understand what is expected of her or him. We will return to these characteristics momentarily. See any good book on learning theory for a more detailed look at this theory. (see LEARNTHE below)

References for Answer 8

• SKINNER - .Skinner, B.F., The Behavior of Organisms, D. Appleton-Century Com, McGraw-Hill, 1961.
• KELLER - Keller, Fred, Learning: reinforcement theory, second edition, Random House, New York, 1969.
• LEARNTHE - Holland, James and Skinner. B. F., Analysis of Behavior, McGraw-Hill, 1961.

When all is said and done, practical instruction in ethics consists of providing guidance as to what an engineer should do in a particular instance to be judged as having behaved ethically. In DOM on page 57 a specific rule is given for implementing the engineering method. Paraphrasing this rule and making it specific to ethics, it becomes

The rule of engineering for acting ethically is in every instance to choose the ethical heuristic for use from what your personal sota takes to be the sota that represents best ethical engineering practice at the time you are required to choose.

In other words, when the engineer is forced to make a decision as to what to do in a concrete instance, our advice is to choose from the set of heuristics available to him or her the heuristic believed most likely to be judged ethical based on the best thinking of the engineering community. Of course, an engineer's personal sota may be so impoverished in the heuristics it contains, his or her knowledge of best engineering practice so poor, or the engineering community’s development of best engineering practice in the particular situation so incomplete that the engineer will ultimately fail in this task and be judged as having acted unethically. But this rule of engineering provides the best guidance for ethical conduct.

So at last, we arrive at the climax of our work in this forum. The practical task before the engineering professor seeking to develop ethical engineers may now be stated concisely:

Use the theoretically mandated strategies of behavioral analysis called generalization and discrimination to make the student’s personal state-of-the-art for best ethical practice as nearly congruent as possible with the engineering community’s state-of-the-art for best ethical practice.

Based on the authority of Plato, I have confidence that the engineering student will then follow this philosopher’s celebrated dictum "to know the good is to do the good." (Although in all honesty, I would replace Plato's esoteric Form of the Good as the standard with the more practical standard of best engineering practice.) It follows immediately from Plato’s analysis that any engineer who now behaves unethically does so out of ignorance. I have no other answer to the implicit question I was asked by the title of this forum, how do you define and teach engineering ethics?

All that remains of our time together is for me to give my best heuristics as to how to bring about this noblest of engineering goals.

Theoretically, the most effective way to teach generalization and discrimination is to put the engineer in the actual work environment and to reinforce selectively the desired behavior. This strategy has the disadvantage of being difficult to implement, of being very slow in achieving the appropriate goal, and of raising the difficult question of whether or not the standard used in reinforcement fairly represents the entire engineering community. As a result of these problems the engineering professor typically turns to the creation of case studies that try to capture important aspects of the engineer's work to study in the classroom. I do not want to be overly harsh with my colleagues, but too often the cases created, and there are a large number of them, are not cognizant of behavioral analysis and as a result are relatively ineffective in modifying student behavior. At best they teach about ethics not what it takes to be judged as having behaved ethically.

In most books on ethics and during most conversations about engineering ethics, it is just a matter of time before someone mentions the Challenger shuttle failure. The participants seem to vie with each other to be the first to introduce it. But what is the clear ethical issue involved? After a standard review of the situation, let us consider the four essential characteristics of a situation necessary for a child to learn the difference in a horse and a dog.

The various recounting of the Challenger shuttle disaster very in length in the telling, but this much always seems to be present. In 1986, a launch by NASA of the space shuttle Challenger exploded soon after take-off. A large investigation of the failure by a “blue-ribbon” committee, the Rogers Commission, took place. (see ROGER reference below) This committee found that, in essence, the engineers assigned to the project recommended that the flight be scrubbed. This advice was not followed by management. One engineer was told to “take off your hat as an engineer and put on your management hat” before you give your answer. As events unfolded, the engineers were correct in their technical assessment and the flight failed. Complicated analysis of the so-called, O-rings, and their failure is often included in accounts of the disaster to support the engineer's judgment.

We begin by stipulating several things:

• Certainly the launch of a space shuttle is a complex system with many trade-offs.
• Certainly no engineer, manager, or politician wants failure and people killed.
• Certainly we can learn from the failure of one organizational structure and create a different one for the future.
• Certainly the trade-off between listening and following the advice of the engineers and considering other criteria turned out to be wrong—but was it unethical?

Now to the essentials for concept formation:

A. The samples we present must be dogs or horses (with no red herrings).

The frequency with which this story is told in relation to engineering ethics certainly implies that someone somewhere must have done something unethical. Since engineering is a risk-taking endeavor, failure alone cannot imply unethical behavior. It is also unfair, certainly educationally unsound, and quite possibly unethical on the part of the person recounting the story if the judgment of unethical conduct is based on documents or knowledge not included in the description presented to the students. Based exclusively on what we have been told, one of the implied "targets" of unethical behavior seems most likely to be the statement “take off your hat as an engineer and put on your management hat.” I have never seen or heard this story repeated without the inclusion of this statement. This statement is neither a dog nor a horse but a red herring. Managers and engineers operate in different optimization spaces with different criteria and weighting coefficients. To deny the possibility that managers are privy to important criteria unknown to the engineers is inadmissible. Perhaps the manager had access to outside information unavailable to the engineers. Perhaps the manager felt the engineers had a poor track record in making similar judgments in the past. The optima of the manager and that of the engineers will not be the same except in highly unusual circumstances. In an ethics course the student will naturally try to infer a dog or a horse from the pre-offered red herring. That is, the student will try to infer ethical or unethical behavior from the pre-offered manager’s statement. Cases to teach ethical behavior must raise ethical issues. If they do not, students will flail about and become justifiably confused as to what an ethical issue is.

B. It will not do to only present the child with dogs.

To be theoretically sound in teaching discrimination, case studies must either appear in pairs that describe very similar situations that differ only in the characteristic that renders one ethical and the other unethical or there must be a single isolated case that differs from best engineering practice only in the characteristic that renders it unethical. And most importantly in the second situation, best engineering judgment against which the isolated case is to be compared must be known to the student before hand. It will not do to only present the child with dogs.

Based exclusively on the description given above of the Challenger case or a similar description, it would be legitimate for the student to infer that you should always follow the advice of the engineer. This is absurd. It is simply not the case that all differences of opinion between management and engineers must be decided in favor of the engineers, even in matters that have a technical component.

It is easy to demonstrate the importance of this concept. We now have a second shuttle failure, the Columbia, to present in conjunction with the Challenger case. According to the Associated Press in August 2003, “[Linda] Ham [shuttle mission management team leader] said they had to rely on an outside contractor, Boeing, for an analysis of the foam that hit on the shuttle’s wing, which predicted there’d be little damage.” (see NASA reference below) Here the engineer’s advice was sought and followed leading to failure. From the direct comparison of the two shuttle cases, we are able to conclude correctly that sometimes you should not listen to the advice of the engineers and sometimes you should. The ethical issue melts away. To teach discrimination the child should not be presented only with examples of dogs.

C. The parent must know the difference between a dog and a horse.

The central issue here is validating a case used in instruction as representative of best engineering practice. If the professor does not know the difference between a dog and a horse, how can this difference be taught? Presenting improperly validated cases is a fraud and itself clearly unethical.

A single professor constructing a case is clearly not a satisfactory arbiter of ethical engineering practice. We all have hidden, and not so hidden, agendas. A liberal leaning professor will unavoidably create or interpret ethical situations differently from a more conservative professor. In the difficult area of defining engineering ethics, even the best intended, but isolated, individual is wanting when it comes to determining a standard of behavior for the whole profession.

We only improve the situation at the margin by slightly increasing the number of individuals. Moving from one individual to five, to seven, or even to twelve and then claiming that we have found the ethical behavior of the entire profession is an extrapolation so large that any competent engineer would refuse to make it in an equivalent technical area.

What we need is a consensus of the largest possible number of credible engineers, engineering ethicists, ethicists, engineering professors, developers of codes of ethics, members of society, media, lawyers, and so forth to decide whether or not a specific case can be reasonably taken as best engineering ethical practice. And, as important, they must tell us what small changes would render the situation described as unethical.

We can expect that there will always be honest disagreements along the ethical axes just as there are honest disagreements along the technical axes. We can, however, have confident that the larger and more representative the sample of individuals is; the more ethical, scholarly, conscientious the group is; and the narrower the gap between dueling ethical and unethical case studies is the more justifiable the resultant cases will be for use in the classroom.

In the best situation we would like to label a case by saying that a national consensus of engineers, engineering ethicists, members of society, managers, and so forth felt at a specific time that this behavior was ethical. Lacking that, we might be interested in saying that at a specific time a consensus of nationally recognized professional engineers found this behavior to be ethical, but a consensus of credible ethicists of the same period found it to be unethical for the following reasons.

The problem of case validation sometimes presents an additional curious manifestation. If the case is not clearly labeled as ethical or unethical and why, sometimes the professor does not know the answer himself. The authors of many cases pride themselves in presenting conundrums that even a group of conscientious professors has difficulty cracking. If the professor expected to teach the ethics course doesn't know the answer to a problem, it is hard to see how we can expect students to learn the answer. If anything the professor is teaching or implying that there are no standards for ethical behavior or that they are too difficult to understand. While it is certainly true that resolving close ethical problems is important and the gap between ethical and unethical behavior needs to be narrowed, the place for that is in determining best engineering practice by the largest group of conscientious engineers instead of in the classroom.

What we cannot do when the professor does not have a clear notion as to whether an ethical situation is ethical or unethical is to ask the group of students to study the case, discuss it, and reach a consensus for themselves. I am reminded of the retort a colleague of mine gave when I suggested a similar solution to a difficult sociological question as a young professor. He asked me if I thought we would get any useful information if we asked a group of six year olds the distance to the moon. Before we call NASA with crucial information for the next moon launch, we should certainly know the capabilities of the group that supplies the information. If we cannot tell the difference between a dog and a horse, how can we expect to be able to teach someone else to do so? The time and place to resolve conundrums is when the profession determines the sota of best ethical practice, not in the classroom when students are struggling to bring their personal sota in conformity with the sota of best practice.

When a professor or a student is presented with an engineering case for consideration, the first step should always be to say "Excuse me, but just who says this represents ethical or unethical behavior?"

D. A time stamp must tell when the learning took place.

Ethical standards have certainly changed in the past and will change in the future. If the decision were to be made today as to whether or not to fly the Challenger, it would certainly be judged unethical to say "yes." A case study must be clearly labeled to tell when it was considered ethical or unethical. Our present glasses did not necessarily correct our hindsight to 20/20. Ethical engineering behavior seems destined to be judged against the current state-of-the-art instead of that of the past.

Based exclusively on what is usually presented to the students in a description of the Challenger shuttle failure, I see no ethical issues. Of course there may be unreported ethical issues hidden from view, but the student can only learn from the information he or she is presented. For a case study to teach ethics, it must present ethical issues.

I am suspicious that this story is so often told in conjunction with discussions of engineering ethics because the engineers turned out to be the heroes. And that is certainly self-serving and most probably unethical.

Few, if any, suites of engineering cases provide the stark contrast between ethical and unethical behavior in very similar situations that is theoretically required to teach generalization and discrimination. If a suite of cases does not do this, it could conceivably sensitize a student to engineering ethical issues, teach them to distinguish between rival ethical theories, teach how to interpret a situation differently depending upon different ethical views, but it most certainly does not teach a student how to behave ethically. Only a series of cases that teach a student to generalize and discriminate against some declared, albeit approximate, norm can do that. The theory is unyielding.

To repeat once again, to teach ethical behavior, what we desperately need is a suite of carefully constructed and validated cases of sota|_{best eng practice} as it concerns both ethical and unethical behavior in specific, but very similar, cases, confident that the practicing engineer will generalize and discriminate and then apply the results to a new situation when perplexed as to what should ethically be done next. Unfortunately, most libraries of engineering Case Studies do not meet that standard.

Just as creating the state-of-the-art that represents best engineering ethical practice was the “weak link” where the profession should allocate its resources to improve ethical behavior, creating theoretically correct case studies is the “weak link” where professors must allocate their resources in teaching engineering ethics.

References for Answer 9

• ROGER—Rogers Commission Report, Report of the Presidential Commission on the Space Shuttle Challenger Accident (Washington, DC: U.S. Government Printing Office, 1986).
• NASA—NASA’s Columbia Conversations, CBSNEWS.com, August 4, 2003

Actually I do—thank you for asking. Very early in this forum I asserted that All engineering is heuristic. That is, that everything an engineer does is based on a heuristic. Somewhat later I continued this theme by asserting that All ethics is heuristic. Once again, this was done to imply that all ethics, ethical judgments, and ethical systems were based on a heuristic. These two statements need generalization in order to represent faithfully my personal views. In my book, I claim that not just engineering, not just ethics, but All is heuristic. In other words, my position is that everything including the word heuristic, the claim that everything is a heuristic, any problem with the self-referential nature of the position that everything is a heuristic, Indeed, all concepts, definitions, etc. used by humans are only plausible aids or directions in the solution of problems that in the final analysis are unjustified, incapable of justification, and potentially fallible. This radical claim is the center piece of my book Discussion of the Method and challenges directly much of modern Western epistemology.

I mention this fact here because twice earlier I promised to reopen the question of an absolute ethical system versus a relative one. For me the concepts absolute, relative, “any contradiction you may find between them” and “the necessity to choose between one and the other” are all heuristics. A very famous mathematical proof called Gödel’s Proof (DOM pg.130) has shown us that the definitive parsing we used to believe existed between propositions that were either true or false is an illusion. Gödel added a "maybe" to the world of all mathematics. When we insist on parsing the notion of ethics into relative ethical systems and absolute ethical systems, I feel that there is a convincing possibility that this parsing and the uneasiness we feel when we search in vain for absolute rules of conduct is also an illusion. My overall worldview renders me deaf to the criticism that I advocate “ethical relativism."

How ironic must it be to conclude a forum on ethics with a plagiarized comment? Surely plagiarism is the archetypical example of unethical behavior. Some time ago, I was asked to prepare a short section on ethics and the engineering method for a soon-to-be published encyclopedia. A person whose name is unknown to me was assigned to edit my contribution and changed my last sentence to better summarize the views I had given. I modified this sentence somewhat, but he or she insisted that the sentence be included as it was edited. I contracted the overall editor for an attribution line but she said that it was the fate of editors, like ghost writers, that their work went without acknowledgment and became the property of the original author. Although in the future the following quotation will surely be attributed to me, in honor of my unknown collaborator, I unethically, but gratefully, plagiarize his or her summary in concluding:

When engineering is recognized as a pluralistic utilization of heuristics to bring about the best change in a limited resource situation that remains to be fully understood, then not only are ethical principles available as useful heuristics but the engineering method can itself become a reasonable description of ethical problem solving in general. (see ANON reference below)

Finally, let me thank you for paying me the compliment of having an interest in my ideas and inviting me to participate in this forum. I know I have learned a lot from my communications with all of you both online and off-line. Your participation will be invaluable to me in a book I’m working on concerning Engineering Ethics that will possibly see the light of day sometime in the distant future. I hope this experience has been equally valuable to you.

Thank you, again, Rob for your help, enthusiasm, dedication, and creativity in setting up this forum. I look forward to your next one.

References for Answer 10

• ANON--Koen, Billy V., “Engineering Method,” Encyclopedia of Science, Technology, and Ethics, Macmillian Reference U.S.A., New York (In Press, scheduled for Nov, 2004)

Well, thank you Profesor Koen. That was a terrific talk; I am sure that we will all profit from your expertise.

Applause...applause...applause...

Ladies and Gentlemen...we are ajourned...thank you for coming...the next Forum will be annouced well-before it is scheduled to begin.

This presentation was sponsored by the IEEE Education Society.