Your Source for Learning
Technology, Strategy, and News
ARTICLES       RSS feed RSS feed

Last Things First: The Power of Backward Chaining

by Bill Brandon

October 13, 2003


by Bill Brandon

October 13, 2003

Backward chaining is an old solution that improves remembering, reduces forgetting, and provides a built-in opportunity for learners to practice what they have learned. It is simple and effective, and you can learn how to use it by reading this article!

Everyone has heard of the “learning curve,” but simply producing e-Learning that efficiently helps learners up that slippery slope isn’t good enough. To be effective, e-Learning must also defeat the learning curve’s evil twin, the “forgetting curve.” And it’s not like the forgetting curve waits politely to start its work until after the e-Learning is over — forgetting starts as soon as learning begins.

In many situations in workplace settings and in education, we develop e-Learning to teach people how to successfully complete long sequences of actions required for specific tasks. Usually the process is to teach how to start the task sequence and then we proceed right along to the end of the chain of events. But have you noticed when you do this that:

  • People learn the first steps better than the last steps?
  • People have trouble remembering the last elements, and they get things “mixed up” in the middle of the sequence?
  • People make more mistakes the further they go in the e-Learning?
  • People get bored by the level of detail presented, discouraged because they keep “screwing up,” or want to quit in frustration because they aren’t able to make progress quickly?

When a learning progression is long, it becomes much more difficult for learners to add new steps. Many activities have to be completed before a new step can be added. It takes the learner a long time to be able to produce the final result. The forgetting curve does its best (worst?) work here. Sooner or later the cumulative memory load, forgetting, and mistakes interfere with learning, and e-Learning becomes less efficient and effective.

There is another, better, way to order many of your teaching progressions to avoid these problems and to banish the evil twin.

In this article, I will introduce you to this better alternative, backward chaining. You will read about instructional sequencing principles and effects, and you will see one very simple example. The simple example will demonstrate basic task analysis and “chunking” the results in a fashion that reduces forgetting. I will give you some guidelines so that you will be able to tell when you have a task that can be taught more successfully using backward chaining. And, most importantly, I’ll explain the series of steps that many designers use to assemble backward-chained instruction.

Forward and backward learning progressions compared

We are used to seeing learning progressions presented in “chronological” or task order: the sequence starts at the beginning and proceeds to the finish. This is the standard for classroom delivery, textbooks, and videos. This task order has become the default for e-Learning, too. Most people, including Subject Matter Experts (SMEs) just assume that the logical start-to-finish progression is the best way to go about any teaching or learning task.

If an e-Learning progression is set up start-to-finish, in task order, it is using “forward chaining.” If set up to teach the last part of the skill first, the progression is called “backward chaining.” (I wrote at length about learning progressions in the December 16, 2002 issue of Learning Solutions Magazine, “How to Build Composite Learning Progressions Using Approximations.” You may find it useful to review that article if the topics of task analysis and learning progressions are new to you.)

Problems with forward chaining

In addition to the problems with forgetting already mentioned, forward chaining sparks some undesirable learner behaviors as the skill being taught gets more complex. For example, learners may create a “mental checklist” so that they can be sure their technique is correct and that they are doing all the steps. Other learners may create mental roadmaps or come up with mnemonics to help them remember everything they are supposed to do, in the right order. In fact what any of these practices do is to make cognitive control part of what should be “covert” or unconscious behavior as the learner executes the skill. This almost always makes the skill performance less effective because of the memory load involved, and because of the time it takes to use the mental checklist. And of course, if the checklist, roadmap, or mnemonic are badly formed or incorrectly recalled, the performance may turn into a disaster.

Still more problems with forward chaining spring from increased learner anxiety associated with this strategy. Negative self-appraisals increase as the number of errors increase. The effect of this negative self-talk is serious and real. What an evaluator sees is that skills are executed well during the start of instruction, but they deteriorate as the sequence continues. I believe that, in some cases, this may be a significant factor in the high rate of abandonment noted by others in e-Learning.

What is backward chaining?

Backward chaining is formally defined as “the strategy of teaching tasks in reverse of the order in which they are done on the job.” In many situations, this is more effective than using the default sequence (forward chaining) and avoids common training problems. Systematic use of backward chaining will result in learning designs that:

  • Keep instructional input or presentation to a minimum, reducing demands on the learner’s short-term or working memory
  • Facilitate transfer of procedural information to long-term memory
  • Keep learners involved and challenged
  • Enable learners to successfully complete a task early in the progression
  • Can be repurposed for use in any medium or method of presentation

Backward chaining may be ideal for applications where the final product is the result of a linear, heavily cumulative sequence of tasks. These are tasks that tend to be done the same way every time and relate to what Ruth Clark and others refer to as “near-transfer.” Such tasks are common in the use of most computer software, in many processes at work, and in many educational situations from kindergarten through graduate school.

Additional criteria for when to use backward chaining appear later in this article. But let’s get down to cases for a minute.

A basic example of backward chaining.

Anyone want a cookie?

When my daughters were small, they loved to bake cookies. They learned how in a way that was totally opposite to the way everyone learns in school, yet it was totally natural. In fact, I’ll bet that readers with children taught their kids quite a few things in exactly the way we taught our girls to make chocolate chip cookies from scratch.

In school, you always start with the beginning. The textbooks always show a sequence of pictures that begins with the first step. We taught Valerie and Jessie to do things backwards, at least as compared to the school standard.

The first thing they learned was how to safely take cookies out of the oven, wait until they were cool, and put them in the cookie jar (with a small “cookie toll” being collected along the way). The next thing they learned was how to safely put the sheet of cookies into the oven, how to set the timer, and how to tell when the cookies were ready. Then they took the cookies out and put them in the jar (with the usual toll). Each time we made cookies, the girls added a step closer to the start, until eventually they were able to start by gathering the ingredients and tools, and to complete the entire process under the watchful eye of one parent or the other.

This example, simple though it is, illustrates the basic way in which backward chaining works. To understand it better, let’s upgrade the cookie-baking to a course given in an educational setting.

The cookie curriculum. As part of the development of the cookie curriculum, a team of experts analyzed the task of making cookies. (See Figure 1.) They decided that all of the many details involved in making cookies could be reduced to seven essential steps, which can be further combined into three groups. The groupings they arrived at are discrete “chunks” that can be taught in a single module. The chunks are of such a size that the learner can easily feel connected to the actual task itself, and never lose sight of the fact that the reason for learning all of this is to make cookies. The resulting chart actually provides a basic domain theory; that is, it would apply to the making of many, if not all, types of cookies. In this case, the outcome is chocolate chip cookies.


outline of cookie production procedures

FIGURE 1 The experts’ view of cookie production


You can see that there are at least two logical orders in which to teach this whole task. You can start with preparing the ingredients and conclude with the baking. Or you can start with the baking and conclude with preparation of ingredients. For the purposes of this article we’ll ignore the other possibilities (start with arranging the workstation and concluding either with ingredients or with baking). The alternatives are sometimes useful, but in this case they would just complicate the problem of the next step in designing the cookie curriculum, which is sequencing the instruction.

The designer decided to use backward chaining as the sequence for the basic course (the complete curriculum also includes a course on decorating cookies, for example). The progression the designer created is summarized in Figure 2. It is important to recognize that in both Modules 2 and 3, the learner will complete the task by baking a batch of cookies. It is also possible for the designer to decide to cover “Measurement” as a short introduction to Module 3, rather than all the way at the beginning of the course. Measurement could also be addressed in another course, or as a course of its own. To keep the progression simple for this example, I elected to show all pre-requisites grouped at the beginning of the course. Usually, you would prefer to address important skills like measurement closer to the point in the progression where they will actually be used.


FIGURE 2 Overview of the cookie curriculum


The final step the designer performed was to set up the exercises in each of the modules. Figure 3 shows one way to do this, using learning theory borrowed from a number of sources. Without getting lost in the details that will be covered later, notice that Module 3 includes four exercises. In the first exercise, learners are provided with pre-computed ingredient amounts, which they then mix. At the end of the fourth exercise, the learners set up their work station and bake the cookies (this has been summarized as “Produce a batch of cookies”). Subsequent classes of would-be pastry chefs in Modules 1 and 2 can use all those mixed ingredients from the first three exercises. Presumably it is no problem to dispose of all the cookies produced.


FIGURE 3 Exercise design for Module 3 of the cookie curriculum


How backward chaining fixes learning problems

The biggest advantage to backward chaining from the learner’s point of view is that it offers immediate satisfaction. The learner completes the activity in every exercise or module. The step sizes minimize mistakes and this makes for higher probability of success. Unlike forward chaining, backward chaining steadily increases skill strength instead of letting interference degrade the skill elements that have already been taught.

Interference does not occur because the learner is always “working into” and practicing elements that have already been done with success. This allows the learner to put undivided attention on the new content.

In addition, the learner feels less tension or anxiety because the task and its steps are “chunked” in a way that keeps them simple. The learner always has a clear understanding about what to do next. Finally, because the first thing the learner experiences is a successful outcome, it is easier for the learner to visualize and anticipate success. This increases learner confidence and further raises the probability of success.

Studies done years ago showed that backward chaining is superior in developing speed, accuracy, fluency, and skill maintenance. Subjectively, my experience with backward chaining is that learning takes place faster, learners develop greater confidence, and performance is generally better on the job. In addition, if a learner quits early, the chances are better that he or she stopped because the backward progression arrived at the level of task elements they had already mastered. In my opinion, backward chaining should at least be considered seriously whenever designing e-Learning to teach progressions or sequences of skills and behavior.

Applications of backward chaining.

Backward chaining has a long history of use in teaching a broad range of skills. It has been used to teach children to tie their shoes and it has been used to teach graduate students to perform analysis of variance and other complex statistical computations. It has been used to teach basic skills in writing in primary and secondary schools, job skills to factory workers, and complex emergency response procedures to military personnel.

When to use backward chaining. Earlier, I mentioned that backward chaining is best suited to near-transfer training: situations in which the task is basically done the same way every time. Backward chaining may not be as suitable for use when learners are required to use judgment to vary the steps of a procedure. However, these statements are also true for forward chaining.

There are four specific situations when backward chaining is preferable to forward chaining:

  • When completion of the task provides natural reinforcement for the learner;
  • When “escaping” from instruction would motivate the learner;
  • When the learner has mastered less than half of the steps in the task chain OR when the learner is close to already having acquired the steps near the end of the chain;
  • When the learners are less patient or less inclined to be cooperative.

Topics Covered

Appreciate this!
Google Plusone Twitter LinkedIn Pinterest Facebook Email Print

Login or subscribe to comment

Thanks for this great and informative article.

I clicked on the URL for how-to-build-composite-learning-progressions-using-approximations but it tells me :"article not found"

May I trouble you to check the URL?

Thanks a billion
Related Articles