Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

6. A general approach to behaviour change

How to use the worksheets

El Paso, Texas, USA

This paper will address two major issues: (1) a general process for developing effective instructional programmes - the approach I will describe is known as criterion referenced instruction; (2) a planning process for identifying what performance discrepancies need to be addressed in an intervention project and how to formulate effective interventions.


The way we think about instruction and teaching is usually determined by the way we were educated. Consequently, most of us think in terms of traditional, subject-matter-oriented instruction that principally uses textbooks and lectures for delivering informational communications to students. Let me describe another approach to instruction that works much more efficiently.

In the CRI approach, instruction is broken down into small units called modules. Each module provides the information, practice, and testing required to learn a precisely identified skill or set of skills. Students typically proceed through the modules at their own individual rates. Each student must master one module before proceeding to the next; that is, all students must meet the same standard on each module. Information is typically presented by means of print or audio/visual devices, as appropriate to the learning and communication skills of the students. Lectures are rarely used, simply because they do not provide as effective or efficient learning.

Students cannot control the presentation rate or sequence in a live lecture as well as they can with other media. As a student, you cannot rerun parts of lectures for a few minutes to think about a point. You can ask for a different explanation if you have trouble understanding a point, but this problem can be averted in other forms of instruction by using well-designed and tested materials, by providing optional resources, and by providing access to an instructor whenever necessary. Furthermore, the inspirational effects of live lectures generally cannot match the inventive effects of instruction with modules. In CRI instruction the student (a) receives frequent evidence of his own learning progress; (b) is never asked to learn something he is not yet ready to learn; and (c) has frequent one-to-one interactions with his instructor when they review his performance together on the criterion tests at the end of each module. Student and instructor interactions in CRI are frequent, close, and directly concerned with each individual student. Instructors are trained to use a very positive approach in their interactions with students and to avoid destructive criticism and degrading remarks.

Most importantly, the CRI approach is oriented towards human performance in real-life situations. It does more than give people information - it teaches them to apply information and skills when needed in real life. It teaches students to perform effectively in later criterion situations because the design of the instruction is referenced to the latter.


CRI is both more effective and more efficient than traditional instruction. It is superior in various ways. First, there are benefits that accrue from allowing students to progress at their own individual rates. In a traditional programme of instruction, all students finish at the same time. Our experience with self-paced instruction is that on average students finish in 30 to 40 per cent less time than in a traditional programme. Some may take longer, but most will take much less time (see line S-P in figure la). Such time savings usually lead to greatly reduced training costs.

Second, there are benefits that accrue from requiring students to master one skill before moving on to the next one. In a traditional programme of instruction, most students make grades in the C range, with considerably fewer making grades in the B and A ranges. Similarly, some will make Ds and Fs. The result is the typical bell-shaped curve (see line C in figure lb). Our experience with a mastery progression programme is that most students make As and Bs and very few make lower grades (see line M in figure lb). The studies from which these results are obtained used the same tests for assigning grades in both the traditional and mastery progression programmes.

Benefits also accrue from referencing instructional design to the criterion situations in which students will later perform. After students finish a traditional programme of instruction and start doing the job for which they were trained, they typically exhibit growth function during their initial time on the job; that is, their job performance improves dramatically as they gain experience. Eventually, their performance tapers off and exhibits very little improvement after they start working (see the lowest curve in figure 1c). Our experience with students after they finish CRI programmes is that they show very little improvement after they start working (see curve R-1 in fugure 1c). However, this failure to improve substantially comes about because their initial job performance is almost as high as that of a highly experienced worker. In other words, there is much less room for improvement.

Sometimes the analysis required to develop a CRI programme creates new ways of doing work that produces graduates who are able to perform better than anyone else ever has, regardless of how much experience they might have had (see curve R-2 in figure lc).


The development of a CRI programme of instruction uses a strategy that (a) requires the use of very precise language for specifying what is to be done on the job and what is to be learned and (b) proceeds through a series of developmental stages one after the other in a prescribed order. The stages for developing CRI can be specified by a set of seven questions. Each stage consists of developing the answer to the question for that stage and each answer is based on the answer to the previous question. The questions that follow are in order and are followed by a brief description of the form taken by each answer.

Fig. 1. Cost-benefit effects of CRI.

  1. What is done on the job? The answer to this question is a list of tasks that make up a job. Or, rather than a job, we could be concerned with a social role (such as being a parent) or a function that makes up one part of a job or role (such as feeding a child).
  2. How is each task done? The answer to this question is a description of how to perform each task. The description may be a list of step-by-step instructions or a flow chart for performing the task.
  3. What skills must be learned by the target population to perform the tasks? The answer to this question is twofold: (a) a description of the skills already possessed by the target population and (b) skill hierarchies for each task or set of related tasks that reach down to the entering level of the target population.
  4. When is learning complete? The answer to this question consists of a performance objective and a matching criterion test for each task and skill selected for the instructional programme.
  5. How are the tasks and skills to be sequenced? Typically, a course map is developed from the skill hierarchies. This map provides the students with choices in sequence wherever possible to fit their own interests.
  6. How is instruction to be delivered? The answer to this question consists of selecting the tasks and skills for each module, of developing the instructional content for each module, and of specifying and developing the appropriate media materials for each part of each module.
  7. How is student progress to be controlled? The answer to this question consists of records of student progress, instructions for administering the programme, positive incentives for insuring student learning, and the removal of obstacles to learning from the instructional environment.

The modules in a CRI programme do not just present information as do the lessons in typical subject-matter-oriented instruction. A module consists of six functional components:

  1. It communicates the performance objective to the learner in ways that he can understand. It tells him exactly what he must be able to do in order to demonstrate successful completion of the module, and how he will be tested.
  2. It communicates the contexts in which the performances will normally occur in real life. When will he do these things and why?
  3. It shows him what the performance looks like when it is done properly.
  4. It provides step-by-step instructions, if necessary, to guide the learner through the performance.
  5. It provides as much practice as the learner needs. This practice consists of the necessary performance situations, feedback for each practice trial, corrective guidance if needed, and positive consequences for improvement.
  6. It provides him with ways of knowing when he has practiced enough, including a criterion test.

Communications intended to change how people do things need to address all six functional components of a module in an integrated and effective manner.


When we become concerned with changing how people do something that makes up a major part of their daily life, we need to recognize that we are tampering with a way of behaving that has evolved over a period of time to best fit a particular environment within the beliefs and values held by those people. The social and physical environments in which they live may lead them to form some useful ways of viewing the world around them, some troublesome ways of doing so, or may simply not provide ways of viewing some aspects of their world at all. The ways in which people interact with the world around them, the habits, skills, and values that they have learned from previous experiences in the environment, the expectations and goals that they have set for themselves, and the responsiveness of the environment to their actions, all go towards making up an intricate and delicately balanced performance system. A small change in part of this system can have dramatic, far-reaching, and unexpected repercussions throughout the system.

A single habit that is held by many members of a population does not exist in isolation. It receives inputs from other habits and provides output to other habits. It is but one link in an extensive net of habits that form a performance system. If we are to be effective without doing more harm than good, then we must focus our planning efforts on the performance system rather than on single habits.

The various beliefs and values held or not held by our target population and the many conditions in their environment that impact on the activities we want to change can often form a very large and complex set of concerns to consider in planning. There are often many different ways in which we can intervene to try to change the ways in which the target population does something. In order to select the best combination of interventions, we need to describe all the significant characteristics of the performance system as it currently works and as we want it to work in the future. There are some useful ways of describing a system and some useful ways of identifying where to attempt interventions to change it.

The following describes a planning process that has some very critical behavioural science concerns built into it. This process begins with developing a description of the performance system as it exists and a description of the performance system that is desired. These two descriptions provide a basis for identifying the discrepancies between what exists and what is desired. These discrepancies become the focus of interventions.


Rather than just talking about systems analysis in general, the focus here is on the performance system we are trying to change - the family nutritional system. Initially, one needs to identify the major functions that make up this system, its overall goals, the components of the system (people, groups, or organizations), the processes or skills required to operate it, the resources it uses, and the environment in which it exists.

The family nutritional system consists of six major functions: (1) plan meals; (2) procure foods; (3) store foods; (4) prepare meals; (5) present meals; and (6) clean up. In order to describe each of these functions, we need to identify who does it, towards what end (that is, what is the goal?), the processes or skills used, the obstacles to performance that exist either in the environment or in the minds of the performers, and the immediate or motivational consequences of performance.

The first worksheet does several things:

  1. It states the categories of information that we need to identify about the system or about each function. These categories include such things as goals, components, resources, processes, obstacles, and consequences.
  2. It lists some examples of the kinds of things required in each category. These examples 1 was able to think up quickly with a very limited knowledge of societies other than my own. Please feel free to add to the list as you see fit to represent your own society and target population properly.
  3. It prompts you to describe the system or function as it actually exists right now and as you and other experts feel it ought to exist.
  4. It provides you with a basis for identifying the discrepancies between what is and what ought to be.

Before going to the details of the Family Nutrition Performance System Worksheet (Appendix 1), consider next how to change what people do and what they think.


Words and Causes

A large part of learning to change how we ourselves think and act and how others think and act is concerned with learning to describe human activities and the possible causes of such activities in useful ways. Unfortunately, most of us are taught by the societies in which we live to think and talk about human activities in ways that are not only useful, but that may even be misleading or harmful. We need to learn to put these non-productive habits aside whenever we decide to get serious about solving human performance problems.

Let us consider some of these improper ways of thinking and talking about human performance. Almost all of us have learned at least some of these habits in the early years of our lives. We need to learn what they are so that we can try to counter them in our analyses.

What we believe to be the causes of what people do and think will guide how we analyse the problem of how to change what they do. For instance, if we believe that people act as they do because of strong forces within them that were created during their childhood, chances are that any change programme we try will be difficult, time consuming, and require very highly skilled change agents.

Often we assign causes to human actions simply by naming the action with an abstract word and use this name to explain the action. It leads to our saying such things as:

These kinds of statements are common in most societies. Yet they contribute nothing to human understanding of human problems. In each instance, the "cause" named in the statement is simply an abstract restatement of the activity or a very general "trait" name that is left undefined.

These ways of accounting for the causes of human activity can be very misleading. We think we have said something profound when in fact we have said nothing at all. These kinds of statements lead us to stop looking for more useful and more precise causes. For instance, consider some more useful ways of accounting for why people eat poor-quality food:

We also need to remember that many of these causes can happen together. That is a very important point. Human activities are almost always "caused" by many different factors. We need to identify all the likely causes for a problem we want to change and then select those that we can do something about with the resources we have.

When people are performing in one way and we want them to perform in a different way, then the difference between what they are doing and what we want them to do is called a performance discrepancy. Our change programme will attempt to reduce the discrepancy between what they are doing and what we believe they ought to do.

Many of the traditional ways of describing human performance discrepancies try to change too big a piece of behaviour all at once without breaking it down into the bits and pieces that make it up. Most often these descriptions attempt to change an abstract trait - for instance, they might try to raise people's nutrition consciousness. Nutrition consciousness can refer to eating a "balanced" diet, to eating "natural" foods, to eating enough food but not too much, to being able to recite the nutrients provided by various kinds of foods, to being able to recite minimal nutritional requirements in specific detail, and so on. Nutrition consciousness can occur in so many different ways that we hardly know where to begin. These big pieces of performance are often a useful place at which to begin an analysis, but certainly we should never end our analysis with something that can be interpreted in so many different ways. Personality and character traits are not causes of behaviour; rather, they are ways of describing behaviour. They are all right to use in casual discourse about human problems, but when we get serious about solving human problems we need to put trait terms aside and use a much more specific and precise language.

The Basic Change Process

We can improve our chances of changing human behaviour by doing several different things:

First, we must identify the human actions we want to change in very specific terms. What is it that people think and do now and what is it we want them to think and do?

Second, we must recognize that what we change is the frequency with which particular actions occur. We are not going to stop people from doing what they do now and cause them to start doing what we want them to do all at once. Change takes time. And during that time, what changes is how often things happen.

Third, we design treatments that make changes in the antecedents and consequences of the performances we want to change. To do so, we must first identify what those antecedents and consequences are now, and then decide what they ought to be in order to bring about the changes in performance that we desire.

Fourth, we select treatments that we can control. If you cannot control the antecedents and consequences you have selected, you may make things much worse and you will not be able to stop it.

Fifth, we introduce the treatments and monitor their effects. In order to monitor the performance effects of a treatment, there has to be something about that performance that we can observe. If there is, then we can note every time the performance occurs and determine if its frequency of occurrence is changing. If it does not change, or change is not fast enough, or if it changes in the wrong direction, then we have to modify our treatment in some way and go back to monitoring its effects.

Sixth, we have to monitor the performance system for other effects that we want to avoid or that we may not have anticipated. If some of these effects are undesirable, we have to be ready to stop or even reverse our previous treatment, or we may have to introduce additional treatments to deal with these other effects. Such additional treatments will also have to be monitored and modified, if necessary. So what we started as a relatively circumscribed change effort can mushroom as we attempt to control more extensive changes throughout the performance system. We need to anticipate as many of these more extensive effects as we can and take them into account in our planning efforts.

Changing People's Thoughts

As you may have noticed, I have spoken about changing both people's actions and their thoughts. Behavioural scientists are often represented as not being concerned with people's thoughts. That is not true. The problem is that we cannot monitor another person's thoughts and, consequently, we cannot tell what is happening. Here is one common way around this problem.

We can teach people to change their own thoughts. In this kind of programme, the change agent acts as a process consultant for the target person himself. There are two problems with this approach: (1) people do not always know what they think in various situations and (2) they must be convinced in advance that they need to change and be willing to try.

Most of us believe that we know what we think in all situations. For the most part, this means that we can put words or images to our thoughts. However, the words are not the thought. Often we do not have words or images to put to our thoughts - we simply act on them. For instance, a person who believes he must never make or admit to a mistake in anything he does, does not say to himself in every situation: "I must not make a mistake." He simply acts as if he believes he must never make a mistake.

If we are willing to change and if we can actually observe our own thoughts in certain situations, then we can change our own thoughts. Or we can teach others and guide them in how to change their thoughts.

Obviously, this is not a practical approach for managing change treatments in large scale programmes. But it can be used in designed change treatments and in evaluating their probable effect in small experimental efforts before applying them on a large scale. Before we commit ourselves to a large programme intended to change how people think, we need to assess the effectiveness of the treatments in small experimental efforts.


The most significant treatments for changing how often people think or act in certain ways in particular situations focus on the consequences of their thoughts and actions. There are four kinds of consequences to consider. The first two increase the frequency with which an action or thought occurs in a situation.

1. Positive Consequences

These are conditions that are desired by our change targets, things that please them, things that they want. A positive consequence can be a physical reward such as money or material objects. Or it can be a special privilege, or recognition or praise from some significant person. Or it can be the good thoughts a person has about himself after doing something right.

A particularly powerful kind of positive consequence is called positive self-imaging. It is often used by people who are changing inappropriate behaviours to more appropriate ones. After each instance of having performed appropriately, the individual images his own long-term achievement of the goal being sought. For instance, an overweight person who eats a lowcalorie meal may image himself as a lean individual receiving attention from an attractive woman.

2. Stopping Unpleasant Conditions

A baby cries. The crying annoys the mother. The mother feeds the baby and the crying stops. As a result, the mother is more likely to feed the baby immediately after it starts crying in the future. This kind of consequence can be effective with people if there is already some unpleasant condition in their environment that we can remove temporarily when they perform as desired. However, we want to avoid introducing some unpleasant condition just so we can turn it off when our change targets perform appropriately. The danger in using this consequence is that it can cause people to avoid the performance situation rather than perform appropriately in it. This can be a very troublesome consequence to try to use with people. Not only can it lead to avoidance, but it can also lead to hostility and aggression.

The last two consequences decrease the frequency with which a thought or action occurs.

3. Punishment

This is an unpleasant consequence that happens immediately following a response. It is most often used to decrease the frequency of undesired performances, but it often exists naturally in an environment as a consequence of desired performances. It is common to find that the reason people do not do the things we want them to do is because they are punished in some way for doing it. Others may criticize them. For instance, a young mother may be criticized by her mother or mother-in-law for breast-feeding her baby. Sometimes doing what we want them to do requires greater effort or causes them to miss doing something they would rather do. What we or others say to our change targets can also be punishing. Although societies use punishment extensively for controlling people, its effects can be very uncertain. Punishment can also lead our change targets to avoid the performance situation or to be hostile and aggressive. Punishment is a very troublesome consequence to use.

4. No-consequence

One way we can decrease the frequency of an undesirable behaviour is to remove the natural consequence that typically follows it. We get rid of the positive consequence or the stopping-of-the-unpleasant-condition that has kept this behaviour going in the past and replace it with no-consequence. When we first introduce no-consequence, the frequency of the undesired behaviour will actually go up before it goes down. But if we just grit our teeth and hang on, it will come down in time. No-consequence can also be a significant reason why people do not do what we want them to do. If we do not provide a meaningful consequence for their doing what we want them to do, they are not likely to do it.

These four kinds of consequences are the basic events that cause a performance system to function as it does and the basic tools available to us in changing the performance system.

Knowledge about the various kinds of consequences that can exist is particularly useful in identifying the consequences that actually control the performances we are interested in changing in their natural environments. Human behaviour is principally controlled by the intrinsic outcome of the behaviour itself, by the things that other people say and do as a result of the behaviour, and by the things that people say and do to themselves. If people are not behaving in some way that appears reasonable and proper, then we must examine the real-life situations to determine what consequences actually are operating in the real situations.

Applying Consequences

There are several important rules we have to keep in mind both in identifying the consequences already in a performance system and in applying our own consequences in an effective manner.

  1. The consequence should immediately follow the performance. The longer the delay between the performance and the consequence, the less effective the consequence will be. We are talking about a matter of seconds.
  2. The consequence should be appropriate to the target population. Because we think something is a positive consequence does not mean our target population will. Or, because we think something is punishing it does not mean they will. This is one reason why change efforts need to be monitored. Sometimes we find it is necessary to change the consequences we originally devised because they did not have the effect we intended. If we have any doubts at all, it will not hurt to check out our consequences in a few small experiments when we design our programme.
  3. Initially, the new consequences we have selected to use in our programme should be applied as often as possible when the target performance occurs. If possible, we want the consequence to follow immediately every occurrence of the target performance. Then we can gradually reduce the ratio of the consequence to the performance as the programme begins to take effect. This is another reason for monitoring the programme. We need to make sure that we do not reduce the consequences too quickly. Performances that are maintained by low-ratio consequences tend to be very stable and resistant to change. By gradually reducing this ratio, we make it more likely that the changes we introduce will last and become a permanent part of the performance system. On the other hand, if the performance we are trying to change is maintained naturally in the performance system by low-ratio consequences, we can expect that change will come very slowly and be very difficult to bring about.

Punishment is used around the world and throughout history as the principal means of controlling human actions and thoughts. However, most such punishment is not applied immediately and does not break up the ongoing response that we are trying to change. Sending someone to prison is legal punishment, but it is not behavioural punishment. The effects of legal punishment on the frequency of an undesired action are very uncertain, particularly if the undesired action is hostile or aggressive in nature. Legal punishment may actually aggravate that kind of problem. Behavioural punishment and legal punishment should not be confused. They are not necessarily the same thing.

Antecedents and Contexts

Human performances occur in response to some event or signal in the environment. This, of course, is the basic idea expressed in the stimulus response paradigm of behavioural science. It suggests two very important considerations to observe when we try to change people's habits. First, it suggests that one way of changing what people do is to prevent or change the signals or stimuli that elicit or call up the response. If a person eats too much rich food, then one way of reducing the frequency of this performance is to remove most of the rich food from the person's environment. This kind of treatment does not change the basic habit, but it does control it.

The basic S R paradigm has another important implication that is being supported by recent findings on human memory. If we want to change how people think or act in certain situations, we must build associations between the responses they are to make and the appropriate signals in the situations in which the responses are to occur. Most of our educational practices around the world violate this principle to at least some extent. Students learn to recite facts and principles, but do not learn to apply those facts and principles in real situations. They can answer verbal questions on tests correctly, but they cannot perform effectively in real situations. Knowing something in one situation, such as a test, does not mean that it will be remembered in another situation when the information is needed. Cognitive scientists are learning that it is much more difficult to retrieve information from human memory in appropriate situations than it is to put that information into memory in the first place. If people are to make effective use of the information they learn, then learning must include practice applying the information in the same or very similar situations. This practice assures that we build the proper associations to retrieve information from memory when it is needed.

Sometimes when we want someone to perform in a new way in an old situation, we may have to add a signal to the situation to call up the new response. For instance, one way in which pregnant women in the United States reassure (the new response) themselves and others of the importance of their condition in their daily activities (the old situations) is to wear a T-shirt that displays the word "BABY" in large letters and an arrow pointing down at their abdomen. It becomes a signal to the woman and those around her to behave in ways appropriate to her condition. It says to her and to others, "Something very important is going on in here." Similar T-shirts read "PRESIDENT 2024." Or one might read "NATIONAL RESOURCE UNDER DEVELOPMENT." Or, in Sri Lanka, where the people are apparently more economically minded, such T-shirts might read, "GROWING RETIREMENT FUND - 18% RETURN."


One of the most powerful ways of changing undesirable thoughts and actions is to teach new thoughts and actions that prevent the old undesirable ones. We call this counter-conditioning. For instance, rather than punish someone for undesirable actions in some situation, we teach a better way of acting in that situation and reward the individual for doing so. This approach gives more stable change in both thoughts and actions with fewer undesirable side-effects than does the use of punishment.

If people are doing something we desire them not to do, it is not enough for our interventions simply to diminish the frequency of the undesired habits. Habits are not broken. They are replaced by other habits. The new habit may be no more than saying to oneself, "I will not respond (as I used to)" to a given situation. Even this is a new action (or thought) that replaces the old action. Or the new action may be an extensive chain of thoughts and actions that prevent the old actions from occurring. We can either control the selection of the new action or let it develop from the target's own predilections. Since we cannot be certain how things will work out on their own, it is probably better to control the selection of the new action wherever possible.

The consequence that we use in building a new response must be more attractive than the natural consequence that follows the old response. If we want to speed things up, we can either prevent the natural consequence for the old response, if that is possible, or introduce a punishment as a new consequence for the old response. Again, however, we always need to be careful about using punishment. Its side-effects can be very damaging to the long-range goals of a change programme.

Contents - Previous - Next