Errorless shaping

What you will take from this lesson

You will be able to describe shaping as a planned change in one observable part of behavior.

This lesson will be theory-heavy, but don’t worry; we will break down each definition into teachable units!

Definition of shaping: 

We will go through some of the definitions of shaping you can find in the literature:

•  B.F. Skinner:

In Science and Human Behavior B.F. Skinner described shaping by creating an analogy:

“Operant conditioning shapes behavior as a sculptor shapes a lump of clay…The final product seems to have a special unity or integrity of design, but we cannot find a point at which this suddenly appears. In the same sense, an operant is not something which appears full grown in the behavior of the organism. It is the result of a continuous shaping process.”

• J.Cooper, T. Heron, W. Heward:

“Using differential reinforcement to produce a series of gradually changing response classes; each response class is a successive approximation toward a terminal behavior. Members of an existing response class are selected for differential reinforcement because they more closely resemble the terminal behavior.”

• Kay Laurance  

Kay discriminates between shaping and micro-shaping.

“Shaping often termed shaping through successive approximations, to a goal behavior. Shapi g can lead to a very high level of error and frustration and the learner will need good mental stamina and motivation to keep trialing. Error is calculated as any unsuccessful action.

Micro-shaping is at the opposite end of the spectrum when the opportunity for error is minimized by control of the environment, usually the placement of the reinforcer. This enables the subsequent behavior to have a chance of 95% success”

• Jesus Rosales-Ruiz and Mary Hunter

“Think of shaping as a straw rather than a funnel. With this mindset, the teacher starts narrowly by reinforcing a particular behavior the learner can already do. The component is then extended to create a new behavior or combined in a novel way with other behaviors. […] Rather than moving from broad to narrow, the teacher’s criteria for reinforcement and the amount of variability of the learner’s behavior both stay narrow the whole time”

The common ground of these definitions would be:

“Shaping a procedure relying on differential reinforcement of successive approximations to achieve a desired form of behavior.“

Where many (like even Kay) would see a space for interpretation in this definition, a space for use of extinction I would argue otherwise. For me, the crucial part of this definition is “successive approximations“. Successive means leading towards the goal, which means they need to be successful. If we also apply what B.F.Skinner taught us about errors:

“Errors are not necessary for learning to occur. Errors are not a function of learning or vice versa, nor are they blamed on the learner. Errors are a function of poor analysis of behavior, a poorly designed shaping program, moving too fast from step to step in the program, and the lack of the prerequisite behavior necessary for success in the program.”

We immediately begin to see why I treat errorless shaping as a design direction: make useful responses easy to find and do not rely on repeated failure.

Errorless?

What does it look like?

One practical way to begin is to aim for a very high rate of accessible success. Kay Laurence’s micro-shaping example uses 95% as a guide. I do not treat that number as a universal rule; the learner’s behavior tells us whether the step fits. A clear target can still help a novice trainer see the size of each approximation.

Another approach is to look at each training loop and refuse to repeat the same error contingency. If a mistake occurs, change something in the setup before asking again. This asks more flexibility from the trainer, but the principle is the same: keep useful responses easy to find and let the learner’s next repetition inform the plan.

It’s essential to remember that as a teacher, your role is to create an environment that facilitates smooth learning and progression through the approximations. Setting up the stage for the learner and arranging the training environment appropriately are key factors in achieving an errorless approach.

Scientific parts of the definition broken down:

Differential reinforcement:

• a procedure where only behaviors that meet a specific criterion are reinforced, while others undergo extinction;
• example: a teacher reacts only when a child raises his/her hand but does not react when the child starts screaming.

Successive approximations:

• it is often highly unlikely that the desired behavior in its final form will be offered in the first stages of shaping the behavior;
• some behaviors do not belong to the natural repertoire of the given organism;
• without establishing successive approximations, i.e., consecutive stages of the behavior, it is impossible to achieve it.

Extinction:

A procedure where the reinforcement of a previously reinforced behavior is discontinued (this refers to both positive and negative reinforcement), which lowers the frequency of the behavior. We will talk more about extinction in a separate lesson this week. 

Resurgence:

The reappearance of previously extinguished behavior during the process of extinction of another (recently reinforced) behavior. 

Read more here

Antecedent Arrangements:

“They are antecedent manipulations of environmental events prior to the occurrence of the behavior. They could affect problem behavior or the behavior we target to increase as a replacement.”

From a shaping perspective, the antecedent arrangement is “setting the stage”. It’s the first thing you got to think about when creating a shaping plan. What can I do to increase my learner’s chances of success? What objects to choose? What location in the training room? Should I use any environmental boundaries? 

This is an example of choosing an incorrect object for teaching behavior. Look how many frustration behaviors got in since there was too much extinction involved.

What can we shape? 

J.Cooper, T. Heron, W. Heward: “The end product of shaping – a terminal behavior – can be claimed when the topography, frequency, latency, duration, or amplitude/magnitude of the target behavior reaches a predetermined criterion level.”

In other words, we can shape:

• topography: refining motor patterns of sit, heelwork, throwing a ball 
• frequency: increasing number of steps in heelwork during one minute, increasing number of pushups during thirty seconds session
• latency: increasing the time between barking, decreasing the time between scent cue and freeze response in nosework
• duration: increasing the time of down-stay, increasing the time of plank
• amplitude/magnitude: increasing the height of the jump, increasing the strength of paws targeting the station in flyball

Within topography:

It means that the form of the behavior remains constant, but differential reinforcement of successive approximations is applied to another measurable dimension of behavior, like for example the accuracy of treat tossing. 

Across topography: 

More specifically across different topographies. It means that selected members of a response class are reinforced. Teaching someone to speak would be a perfect example. We shape through different topographies: lip movement, speech sounds, etc. but they all lead to a terminal goal; speaking. 

One important takeaway

One crucial takeaway from this lesson and the entire course is that there is no single correct way to shape behaviors. The goal of this course is to equip you with the necessary skills and prerequisites to find a tailored series of steps for each learner, behavior, and environment, regardless of your previous experience. There is always another approach, another progression to achieve the desired behavior. However, there is no universal guidebook that outlines how to shape specific behaviors like X, Y, or Z. While the laws of behavior are universal, the practical application must be adjusted to the individual learner in front of you.

Later in Week 2, in Stairways to Heaven or Legoland, we will turn this into a practical, modular approach to shaping rather than a climb up one fixed ladder. As you finish this lesson, write down one question you want that lesson to help you answer.

Want to go one step further?

Treat a high success rate as a design aim, not as a scientific pass mark. Here, errorless shaping means making useful responses easy to find and avoiding repeated failure as the teaching engine.

Choose one observable response dimension at a time instead of asking for a vague improvement.

Try this

Pick one behavior and one dimension only: form, latency, duration, frequency, distance, or force. Write the current version and the next small version in two short sentences.

Check your understanding

1. Can you raise form and duration in the same criterion?
2. Does errorless shaping promise zero errors?

Answers

1. You can, but you will not know which change caused a problem. Start with one.
2. No. It is an error-minimizing design direction.