11. Functional assessment | Behaviour Analysis: A Primer

11. Functional Assessment & Functional Analysis

Measurement is important in any science and in the previous chapter we looked at how to measure various dimensions of behaviour. Given that the occurrence of a behaviour is dependent on the context in which it is observed, the behaviour is classified as a ‘dependent variable’, i.e., its occurrence depends on the occurrence of another, independent, variable. Putting that science speak in everyday terms, this means that if you can identify aspects of the environment (independent variable) that are responsible for the behaviour that was observed (dependent variable), you will have explained the reason for your observation. In effect, explanations for behaviour boil down to statements describing the relations between dependent and independent variables, what we call functional relations.

Once functional relations have been identified, the mystery surrounding the occurrence of a behaviour disappears. This sounds fairly logical and straightforward, but it is not always as easy as it sounds. First, we have to contend with the conceptual traps caused by mentalism in the search for these functional relations. Second, we might not always know how to collect the appropriate data. It’s all very well looking at simple behaviours in a laboratory, but what about the real world where behaviours are more complicated and contingencies of reinforcement and punishment are also more complicated?

Functional assessment is the general term used for a range of methods that can be used to help address issues on data collection. Functional relations are uncovered by procedures that systematically examine the antecedents and consequences that lead to and maintain a behaviour. This information can then be used to develop individually tailored interventions. In this chapter we examine some of the procedures that have been devised.
There are three main methods in functional assessment:

Direct observation
Indirect or informant based methods
Functional Analysis

Direct Observation

In Movie 11.1, we first see a snapshot of a child kneeling and stretching her arms forward against a white background. If you were asked about what she was doing, you would have to take a wild guess. Maybe she is stroking a dog or fixing a bike.

The next picture adds a background image. You can see that she is actually kneeling on a chair and leaning on a table. If you were asked what she is doing, your guess would be a little more informed. There is no dog and no bike, so the previous guess is not right. You see papers and pens on the table, so she may be drawing a picture or writing something.

The next slide adds some motion and you can see that she is doing her homework and talking to someone. As the scenario unfolds in subsequent slides you observe the normal routine of a child being given homework by her teacher, being told by her mother to do her homework, doing her homework, being praised after finishing it, and then being praised by the teacher after she brings it to school.

The focal segment in this sequence is the child doing her homework. This scene was chosen simply because it provided an opportunity to show visually that the concept of the behavioural stream is not a fanciful abstraction invented by academics in their ivory towers. Rather, it is a simple statement of fact, and one that provides a framework that helps us to understand how components of the behavioural stream are interrelated. The graphics merely help us to visualise the kinds of data that could be obtained from direct observation of relations that can exist between antecedents, behaviour, and consequences.

When it comes to applied research with children who are not doing their homework, it makes sense to examine components of the typical stream shown here with Kalinka to see whether the problem lies with the antecedents or consequences of the behaviour. This is exactly what was done by Millar and Kelley (1994). They used goal setting (antecedent changes) and contingency contracting (changes of consequences) to improve homework performance and on-task behaviour of children. Parents of these children were actively involved and consumer satisfaction for the procedures was high.

Movie 11.1 has an example of another behavioural stream. With ‘Jean’, we see an older woman receiving a visitor and together they reminisce about the ‘good old days’. We have identified examples of the distal and proximal antecedents and distal and proximal consequences. There are more of course, but as in all these examples, we have merely highlighted those that ‘tell the story’ around a focal behaviour, which in this instance is reminiscing, i.e., on-topic self-initiated conversation.

Bourgeois (1993) examined the behavioural stream of older participants who were diagnosed with dementia and lived in day-care and nursing-home settings. She examined ways of helping these participants to engage in on-topic self-initiated conversations. Using a prosthetic memory aid (antecedent) had a positive effect on the frequency of productive utterances, length of conversational turns, appropriate relinquishing of conversational dominance, and increasing acknowledging or affirmative comments (reinforcing consequences).

Finally, in Movie 11.1 we see a mother bathing her baby. Clearly, Martina’s childcare skills are well developed. She comes from a large family, had lots of experience with babies as she grew up herself, and Ellen is her second child. She had plenty of chances to practice bathing a baby and seems quite relaxed. However, childcare skills do not come naturally; young parents usually learn from others. Seeing Martina engaged in these skills offers a starting point for helping a parent who has difficulties caring for their baby. Green at al. (1995) worked with two families in which parents had intellectual disabilities and children were in foster care. Both families were instructed in childcare skills. One mother readily acquired and applied these skills and the child’s phased return home was implemented. However, the other parent did not acquire the skills and the child remained in care.

The Behavioural Interview

So far, we have looked at the notion of behavioural measurement in situations where the behavioural scientist could directly observe the behaviour of another person. There are, however, many occasions when this is not possible. For example, the behaviour of interest might be private (as in the case of feelings and thoughts), or it might occur in the privacy of someone's home. In such cases we have to rely on the verbal or written reports of others who are in a position to observe the behaviour directly.

All too often, however, the lay person reports their own interpretation of events rather than a precise description of the behaviour in question. This problem with verbal reports was outlined earlier when we discussed summary labels and mentalistic thinking. We also came across a particular example when students were asked to describe Adam's behaviour. Their reports included: "He was rude", "He was cheeky", "He was romantic".

Because these diverse summary labels of the same carefully choreographed scene are open to individual interpretation, they are not appropriate for the precision required by a natural science approach to behaviour. What, then, is the scientist to do? In order to obtain an accurate assessment of events that transpired during the scene, a behaviour analyst would rely on information obtained from a Behavioural Interview. In such an interview the client is asked for a precise description of the behaviour in question, along with the context in which it occurs. So-called "W" questions are the key to this procedure: What exactly did he do? Who did he do it to? How did he do it? Where did he do it?

Let us look at a typical example from childcare practice. Meranda (age 22) is the single mother of two children. Paul is 2 1/2 years old, and Lucy is 9 months old. Meranda states that Paul "is often very cheeky'" and she "has to hit" him at times.

Parents in Meranda's situation are usually told not to worry too much: "Paul is at the 'terrible two' stage and will grow out of it". Meranda may even be labelled an ineffective parent, in danger of physically abusing her child. In order to help Meranda manage Paul's behaviour effectively, we need to find out exactly what she means by "cheeky". His behaviour could be the same as that of Adam, who was also labelled "cheeky", or it could be totally different. In the behavioural interview, the first task is to establish the topography of Paul's behaviour (Movie 11.2).

From this animation we have now established that when Meranda said that Paul was cheeky, she meant that he hit his little sister with his fist on the head. Quite a difference to Adam's behaviour, isn't it! This is possibly not the only behaviour that Meranda labels "cheeky" and we will have to ask her for equally precise descriptions of other behaviours. We must also ask her about the frequency, duration, and latency of these behaviours. If she is unable to give us precise information, the first step of any intervention would be to ask her to record the behaviour accurately.

The next step in the behavioural interview is an analysis of the context in which the behaviour occurs (Movie 11.3). Where does it occur? What happens immediately before it? Who is present at the time? The aim of these questions is to identify possible Antecedents to the behaviour.

We now know in much more detail what Meranda means when she says Paul is 'cheeky'. We know that she means that he hits his little sister with his fist on the head, that this happens in the kitchen after mealtimes when his father is present. The next set of questions in the interview is concerned with the consequences of Paul's behaviour. What happens after he hits (Movie 11.4).

We have noted earlier that consequences influence behaviour. For a full analysis of Paul's behaviour, then, it is crucial that we ascertain the consequences of Paul's hitting behaviour. If particular consequences are responsible for the maintenance of Paul's behaviour, we will be able to identify them with this line of questioning.

We have now conducted an analysis of Paul's behaviour that is much more precise than the summary label 'cheeky' given by Meranda. With this information we can develop a possible explanation for Paul's behaviour that includes reference to the roles of important antecedents and consequences. Once these events are known, it becomes a relatively straightforward task to change Paul's behaviour. The first step would be to specify a desired 'target behaviour'. That is, working with Meranda, the behaviour analyst specifies exactly what it is that she wants Paul to do. Thereafter, the intervention involves changing either the antecedents or consequences that maintain his problem behaviour. At all times during the intervention, records are kept for monitoring the effects on Paul's behaviour. These records help ensure that decisions can be quickly made if changes to the intervention are required.

A Practical for the Teacher

Assessment by behavioural interview
This practical is aimed at developing behavioural interview skills in students. It is structured in a step-by-step fashion thereby ensuring that students will be able to learn how to use their skills in a flexible, focused, and effective manner.

INTRODUCTION
Introduce the session and explain to the students that while direct observation is the most reliable assessment method for behavioural intervention, this is not always possible. In many cases we have to rely on a verbal description of the problem and its possible causes. This can be obtained by careful questioning. The behavioural problem should -if at all possible- be described in such a way that the client, worker, and anyone else involved, would agree on whether the behaviour had occurred or not; in other words, what is sought is a specific and reliable description. This type of description is important for intervention planning as well as ethical and non-judgmental recording. For example, in regard to Social Enquiry Reports (these are reports regarding the social background of offenders that are required by courts and furnished by Probation Officers) recent criminal justice legislation in the UK demands consideration of concrete issues rather than nebulous and subjective material which has often been the hallmark of Social Enquiry Reports in the past.

EXERCISE
The following exercise is designed to guide students through a behavioural interview one step at a time. It is important that they have mastered each step before proceeding to the next one. Ask students in pairs to role-play an interview situation, one acting as the client, the other one as the behavioural worker. The case details are as follows:
Petra is ten years old. Her mother, Mrs. Smith, has come to you for advice because she is worried about Petra's future in secondary school. She thinks Petra will not be able to cope because she is very self-conscious and has low self-esteem. Can you help?
First step
The first objective is to describe the problem behaviour in detail. In other words, your task is to get the students to produce an acceptable definition of the Behaviour. Ask students to use 'W' questions such as 'What' or 'How' to obtain a verbal description of the behaviour. What does Mrs. Smith mean by 'self-conscious' and by 'low self-esteem'. What would we have to see Petra do for Mrs. Smith to consider her as 'self-conscious' and with 'low self-esteem'? Interview questions should centre around topography, frequency and duration of behaviour. This role-play is to last for approximately 5 minutes.

Examples
After the role-play, ask the students to give you some examples of behaviours which they have defined and write these up on the board. Examples could be "She does not talk", "She does not keep eye contact", "She blushes", etc. Ask students for more precise details if their definitions are not precise enough.

Second step
In the second step of the exercise the students should establish the context in which the behaviour occurs. In other words, they are to identify the antecedents of the behaviour. Ask the students to use one particular behaviour that is written on the board for their role-play. They should use 'W' questions such as "Where does the behaviour occur?", "When does it occur?", "Who is present?", "Who is present when it does not occur?", etc. Give the students approximately 5 minutes for the role-play.

Examples
After the role-play, ask the students to give you some examples of the antecedents they have identified and write these up on the board. Examples could be, "In new situations", "When she meets “new people", "In the classroom", etc. Ask the students for more details if their definitions are not precise enough.

Third step
In the next step, students should concentrate on what happens after the behaviour has occurred. In other words, they should identify the consequences of the behaviour. Use one of the behaviours mentioned in the previous step and again use 'W' questions such as "What happened afterwards?", "What did you do?", "What did Petra do?". Give the students approximately 5 minutes for the role-play.

Examples
After the role-play, ask the students to give you some examples of the consequences they have identified and write these up on the board. Examples could be "Teacher looks at Petra", "Mother speaks to her", "No-one speaks with Petra", etc. Ask the students for more precise detail if their definitions are not precise enough.

Conclusion
After completing the three steps, you can continue this exercise by role-playing how one might define and establish target behaviours; continue along the same lines described in each step. Parental participation in the establishment of aims and objectives in an intervention is particularly important.

Functional Analysis

The examples of the functional assessment that we used here were fairly simple and obvious (see also Movie 11.5), but of course there are many times when behaviour is much more complex and it is difficult to be clear about which antecedents and consequences bracket a focal behaviour.

In more scientific terms, we say that it may be difficult to identify which antecedents and/or consequences play a key role in maintaining the behaviour; i.e., the variables of which the behaviour is a function. We say this because we know that some antecedents have acquired discriminative control over the behaviour (e.g., the antecedent is a Sᴰ) and some consequences have reinforcing or punishing effects on the behaviour. But we may not know which. Consider, for example, a very complex behaviour, such as self-injurious behaviour (SIB). The effect of this behaviour is damage to a person’s own body. Typical examples include hitting, scratching, or biting oneself or banging one’s head against a wall or other hard object which often leads to bruises or cuts. Extreme cases may also include the behaviour of gouging out one’s own eyes. Self-injurious behaviour is a serious issue not only for the individuals who engage in it but also for everyone around them. SIB reduces the ability to focus on education, and depending on severity, physical injury may require medical attention. For family and staff, it is very upsetting to watch a loved one hurting him/herself and in trying to help you can end up in the line of fire yourself. Traditionally, interventions have relied mainly on medication or physical restraint, such as arm braces or head gear

If we are not clear about which antecedents and consequences play key roles in maintaining SIB, we will not be in a position to help the person change this behaviour. Simply trying a variety of treatment approaches might eventually achieve the desired treatment outcome. However, this would neither be very efficient nor very ethical. Furthermore, the possible success of this kind of approach would be by default rather than by design. Even if you are successful with one person, you cannot guarantee that repeating a successful intervention with other clients would be equally successful.

A Functional Analysis is a systematic approach that has been developed to help identify, on an individual basis, the role of antecedents and consequences in maintaining behaviour.

This image shows Danny engaged in self-injurious behaviour. He tends to pull his own hair until it comes out in big tuffs and he punches his own eyes until they are bruised. His parents say that “He does it all the time!”. He usually wears a helmet and sometimes his arms are restrained. He also takes medication. This kind of behaviour obviously prevents him from leading a fulfilled and happy life.

Imagine that you were charged with helping him to stop engaging in such destructive behaviours and trying to take him off physical as well as chemical restraints. Where would you begin? Well, one of the most important things you have learned to date from the study of schedules of reinforcement is that a discrete behaviour observed at any point in time may be part of a more general pattern of behaviour. The explanation for the observed behaviour, then, is to be found by making repeated observations until the pattern is uncovered. Data collection would involve details of his interactions with the environment, including details of various antecedents and consequences to his behaviour.

Sometimes, though, the antecedents and consequence of a behaviour are not readily identified. If that is the case, then we need to devise a specific strategy to determine which antecedents (distal and proximal) occasion his behaviour and the role that consequences play in maintaining it. We call this approach an ‘experimental’ or functional analysis.

In a functional analysis you would design a variety of test conditions in which you could observe Danny’s behaviour. The purpose of these tests would be to systematically assess different conditions until you identify the functional relations between his SIB and specific antecedents and consequences that cause and maintain it. In other words, by using the scientific method you would be able to determine which conditions reliably produce his SIB. You would then use the results of your investigation to design a treatment package that is specifically tailored for him.

Let’s now have a look at a few of things we could do. Here are four different test conditions that are typically used in the search for functional relations:

Condition 1: Attention

We could design a test to establish the effect of the most common consequence provided for someone engaging in SIB, social attention in the form of a reprimand and disapproval.

Danny and you would be in a room. Toys would be available. Danny would be told to play with the toys, while you do some work.

You might observe that Danny engages in self-injurious behaviour. If he does, you would show disapproval by saying “Don’t do that!”. If he does not engage in self-injurious behaviour, you do not give him any attention.

Analysis of Condition 1: Attention
If Danny engaged in SIB during this condition and then continued with it after social disapproval, we would conclude that social disapproval or a reprimand do not function as ‘punishers’, i.e., they do not reduce the probability of the behaviour. In fact, they may even be functioning as reinforcers given that the behaviour persists.

Condition 2: Demand

We could design a test to establish the effect of another common consequence provided for someone engaging in SIB, allowing the person to discontinue a demanding task.
You would ask Danny to engage in a relatively difficult academic task, completing a puzzle, or threading beads, etc.. You might observe that he engages in self-injurious behaviour at some point. If he does, the tasks would be terminated and you would simply turn away from him.

Analysis of Condition 2: Demand
If Danny engaged in SIB when presented with a task, or sometime during the task, but stops when the task is removed, we would conclude that termination of a task functions as a negative reinforcer. In other words, SIB is maintained by an escape contingency.

If Danny engaged in SIB during this condition and then continued with it after social disapproval, we would conclude that social disapproval or a reprimand do not function as ‘punishers’, i.e., they do not reduce the probability of the behaviour. In fact, they may even be functioning as reinforcers given that the behaviour persists.

Conidtion 3: Play/enriched environment

This control condition simulates an ‘enriched’ environment with the experimenter and lots of Danny’s favourite toys are made available to him. This condition is designed to control for the effect of the mere presence of the experimenter and toys, and it examines the consequences of delivering praise for NOT engaging in SIB. A variety of toys would be made available to Danny in a room. You would stay in close proximity to him. You might observe that he does not engage in SIB. You would praise Danny for not engaging in SIB and you would ignore any occurrence of SIB.

Analysis of Condition 3: Play
If SIB does not occur, we would conclude that the mere presence of experimenter or toys does not function as an antecedent to SIB. In fact, praise may function as a positive reinforcer for the absence of SIB and ignoring it may function as ‘punisher’ for SIB when it occurs.

Condition 4: Low stimulation

Because of growing evidence that SIB may arise from sensory stimulation, we could examine this possibility by designing a test to simulate an ‘impoverished’ environment. Essentially, this means is that there would be no opportunities for social or other antecedents or consequences to occur. Clearly, this condition would not be conducted with anyone whose SIB is very severe when left alone. In order to

safeguard Danny, before setting up this condition we would need to assess the levels of Danny’s SIB very carefully and of course, the condition would be terminated immediately if SIB became severe. We could arrange such an environment by leaving Danny in a room by himself without toys. We might observe him engaging in SIB.

Analysis of Condition 4: Low stimulation
If Danny continued to engage in SIB, we would conclude that the sensory stimulation produced by this behaviour functions as an ‘automatic’ reinforcer.

Conclusion
We have seen that when we want to understand the reasons why a behaviour occurs and why it is maintained we may need to use carefully crafted test conditions. In Danny’s case, this approach helps us to identify precisely those circumstances in which his SIB is likely to occur, and they can identify those events that function as reinforcers for either the presence or absence of SIB.

Functional analysis is discussed in detail in a special edition of JABA (1994) devoted to behaviour assessment. In this issue, Brian Iwata and colleagues report how they carried out a range of functional-analytic tests with individuals with developmental disabilities who displayed self-injurious behaviour. They describe how the detailed functional analysis of self-injurious behaviour led to focused and effective interventions.

The conclusion from all test conditions is that Danny’s SIB is a function of a number of different variables. Decisions regarding appropriate intervention are based on the identification of these variables. In Danny’s case, here are some suggestions for interventions based on the findings in each condition.

Condition 1: Attention
When Danny was reprimanded for SIB and continued to engage in it, we concluded that social disapproval and reprimand did not function as ‘punishers’. In fact, we also drew attention to the possibility that this kind of social attention could function as a reinforcer. If Condition 1 had been the only condition in which Danny engaged in SIB, it would follow that the intervention should involve the replacement of social disapproval and reprimand with non-contingent attention or access to material of food. An intervention should also involve differential reinforcement of other or alternative behaviours (DRO and DRA respectively) in order to replace SIB with less aberrant behaviours that serve the same function, i.e., gaining social attention.

Condition 2: Demand
If Danny stopped engaging in SIB after the termination of a demanding task, we would conclude that escape from the task functioned as a negative reinforcer for SIB. Had this been the only condition in which Danny engaged in SIB there are a number of ways in which in which he could be helped:
1 - The difficulty of the task demand could be reduced.
2 - He could be taught the necessary skills to accomplish the task.
3 - He could be taught to ask for a break appropriately to allow for time off-task or for access to leisure activities.
As part of the general effort to individualise an intervention and generate motivation for task completion, positive reinforcement for appropriate behaviour should be provided frequently.

Condition 3: Play
When Danny was praised for the absence of SIB while at the same time SIB was ignored, he did not engage in SIB. We concluded that praise functioned as a positive reinforcer for the absence of SIB. We also concluded that ignoring SIB may function as a ‘punisher’ for it. To increase the likelihood of him not engaging in SIB, we should use praise for appropriate behaviour more often and continue to ignore SIB.

Condition 4: Low stimulation
When Danny was left alone and engaged in SIB, we concluded that the sensory stimulation produced by this behaviour functioned as an ‘automatic’ reinforcer. If this had been the only test condition in which Danny engaged in SIB, it would follow that an intervention should involve substitute sensory stimulation.

It can be difficult to identify substitutes, though, because some topographies of SIB produce multiple sources of sensory stimulation. Alternative sources of stimulation may include leisure materials. Availability of these should be combined with keeping Danny occupied with preferred tasks.

Summary

To sum up, functional analysis offers an experimental tool to help assess the function of a behaviour in a way that goes beyond simple direct observation in natural environments. Functional analysis was first published in a review of 152 cases when Iwata et al. (1994; reprinted from 1982) noted the following;

“All of these interventions when applied to the function of SIB for which they were designed, reduced behavior to below 10% of baseline in over 80% of cases. More importantly, uniformly positive results were obtained in spite of the fact that reinforcement-based interventions were used two to three times more often than punishment” (p. 233).

Since then, functional analysis has become the key methodology for the assessment of severe problem behaviours, including SIB and many others. It has been used in experimental settings, but also in schools and in the home (Movie 11.5).