Behaviorism and Philosophical Foundations

Why behavior analysis explains behavior the way it does: what the science tries to achieve, the assumptions it works from and what makes a piece of work count as applied behavior analysis.

Concepts
5
Sections
5
Diagrams
6
In this chapter
  1. 01 What the science is trying to do
  2. 02 The assumptions behind the science
  3. 03 Explaining behavior
  4. 04 Philosophy, research and practice
  5. 05 The dimensions of applied behavior analysis

What the science is trying to do

A science is judged by how far it can go with its subject matter. Behavior analysis names three levels, and each one asks for a stronger kind of evidence than the one before.

Figure 1

Three levels of understanding

  1. Description

    What happens, and how much?

    Evidence
    Systematic observation and measurement
    What it can claim
    Does not relate events to one another
  2. Prediction

    What tends to go with what?

    Evidence
    Repeated covariation between events
    What it can claim
    Does not show that one event produces the other
  3. Control

    What happens when we change it?

    Evidence
    A variable is manipulated and behavior changes reliably
    What it can claim
    Demonstrates a functional relation
Each level adds a kind of evidence to the one below it. Description records; prediction rests on events that reliably vary together; only control, shown by changing a variable and seeing behavior follow, demonstrates a functional relation.
A.1

Goals of behavior analysis as a science

Description is the first level: systematic observation that turns what happens into numbers, such as how many times a student asks for help each morning. Descriptive facts stand alone; they do not yet connect one event to another. Prediction becomes possible once repeated observations reveal that two events tend to vary together, so that one can be anticipated from the other. Covariation, however reliable, does not show that one event produces the other, because a third variable may drive both. Control is the most demanding level. The investigator changes one variable, keeps the rest as steady as possible, and shows that behavior changes reliably whenever that variable changes. A dependency demonstrated in this way is a functional relation, and it tells you what to change.

Example

A teacher logs a student’s off-task behavior for two weeks and finds it in about a third of the observation intervals: description. She notices it is much higher on days with long written assignments: prediction. She then alternates long and short assignments on a planned schedule, and off-task behavior goes up with the long ones and down with the short ones, again and again: control.

Common confusion

Control here is a word about evidence. It means a variable has been shown to be responsible for a change; it does not mean forcing, restricting or suppressing what a person does. And a strong correlation is still only prediction: knowing that problems cluster on Mondays does not tell you which Monday event to change until something has actually been manipulated.

Key points (4)
  • Description is systematic observation and recording of what occurs; it yields quantified facts but does not by itself relate events to one another.
  • Prediction rests on repeated covariation between events, so one can be anticipated from the other; covariation alone does not show that one event produces the other.
  • Control is the experimental demonstration that changing one variable reliably changes behavior (a functional relation); it is a scientific term, not coercion of the person.
  • The three goals are levels of understanding that build on one another: description informs prediction, and both guide which variables to manipulate to show control.
Examples and a practice question: A.1 Goals of behavior analysis as a science Practice this: A.1 Goals of behavior analysis as a science

The assumptions behind the science

Before any data are collected, a science commits to a few working assumptions about its subject and about what counts as knowing it. Each one answers a different question, which is why exam items like to swap them.

Figure 2

Six attitudes, six different questions

  • Determinism

    Is behavior lawful?

    Yes: it occurs in orderly relations with events, so it can be changed.

  • Empiricism

    Where does the evidence come from?

    From objective observation and measurement.

  • Experimentation

    How do we find out?

    By changing a variable on purpose and holding others steady.

  • Replication

    Can we rely on it?

    Only once the effect appears again when the procedure is repeated.

  • Parsimony

    Which explanation first?

    The simplest one built on established principles.

  • Philosophic doubt

    Is it settled?

    Never for good: every finding stays open to revision.

Each attitude answers its own question, which is how to tell them apart. Selectionism and pragmatism, the two remaining assumptions, appear in the next figure and in the text.

Figure 3

Selection by consequences works at three levels

  • Phylogenic

    Time scale
    Generations of a species
    What is selected
    Reflexes and the capacity to be changed by consequences
    Example
    Blinking when something approaches the eye
  • Ontogenic

    Time scale
    One person’s lifetime
    What is selected
    Operant behavior, through its consequences
    Example
    A child learning which requests get results at home
  • Cultural

    Time scale
    Groups over time
    What is selected
    Practices passed from person to person
    Example
    A team keeping a checklist that cut errors
The same logic, variation followed by selection, operates over three time scales. Behavior selected at one level can still change under later conditions; selection describes a history, not a fixed outcome.
A.2

Philosophical assumptions of the science

Determinism assumes that behavior is lawful: it occurs in orderly relations with events in a person’s history and surroundings. Empiricism accepts a claim only when it rests on objective observation and measurement, not on opinion, reputation or a moving story. Experimentation is the preferred way to learn: a variable is changed on purpose while others are held steady, to see whether behavior depends on it. Replication repeats an experiment, or a condition within one, to find out whether an effect can be relied on. Parsimony asks that simple explanations built from established principles be checked and ruled out before more complicated or speculative ones are considered. Philosophic doubt treats every finding, including your own, as open to revision when better evidence appears. Two more assumptions complete the set. Selectionism explains change through variation and selection at three levels: inherited characteristics are shaped across generations by their effects on survival and reproduction, operant behavior is shaped within a lifetime by its consequences, and cultural practices persist or change as they are transmitted and produce outcomes for groups. Pragmatism judges knowledge by whether it leads to effective prediction and control.

Example

A website claims a new weighted bracelet ends tantrums and quotes delighted parents. Empiricism asks for measured data instead of quotes. Parsimony asks first whether tantrums are simply being followed by attention or escape. Experimentation would compare tantrums with and without the bracelet while other conditions stay the same, and replication would check that the result holds more than once. Even a positive result is then held provisionally, as philosophic doubt requires.

Common confusion

Determinism is not fatalism. Saying that behavior has causes is exactly what makes it changeable, because the causes can be changed. Philosophic doubt is not a refusal to accept findings or to act until everything is certain; it means acting on the best evidence while staying ready to revise. Keep the others apart by their question: empiricism asks where the evidence comes from, parsimony asks which explanation to test first, and pragmatism asks whether the knowledge works.

Key points (5)
  • Determinism: behavior is lawful and depends on identifiable events, which is precisely why changing those events can change behavior; it is not fatalism.
  • Empiricism: claims are accepted on the basis of objective observation and measurement, not opinion, authority or testimonials.
  • Selectionism: consequences select behavior at three levels, the species (phylogeny), the individual's lifetime (ontogeny) and cultural practices.
  • Parsimony: simpler explanations that require fewer assumptions are evaluated and ruled out before more complex or abstract ones are entertained.
  • Pragmatism: knowledge is judged by whether it works, that is, whether it leads to effective prediction and control of behavior.

Explaining behavior

An explanation says where to look for the causes of behavior. Radical behaviorism looks at the environment, past and present, and treats thinking and feeling as behavior to be explained rather than as the explanation.

Figure 4

A circular explanation, and a way out

Explanatory fiction

  1. Leo leaves worksheets unfinished
  2. “So he is unmotivated”
  3. “He leaves them unfinished because he is unmotivated”

The only evidence for the cause is the behavior it explains.

Behavior-analytic account

Current environment
Worksheet is long; a tablet game is within reach
Behavior
Stops after the first few problems
Consequence
Gets to play; the worksheet is taken away
History
Stopping has often ended hard work before

These events suggest a testable hypothesis. Observe the current contingencies and assess the relevant history; the past itself cannot be changed. Experimental evidence is needed to establish a functional relation.

On the left, a label inferred from the behavior is offered as its cause, so the explanation returns to its starting point. On the right, the account points to events that can be observed and changed.
A.3

Explaining behavior from radical behaviorism

Radical behaviorism is the philosophy behind behavior analysis. Radical here means thoroughgoing: the same analysis covers every kind of behavior, private events included: thinking, remembering and feeling, which only the person engaging in them can observe directly. Methodological behaviorism accepts that private events exist but keeps them out of scientific analysis because nobody else can observe them. A radical behaviorist explanation looks for the causes of behavior in the history of the species, in the individual’s learning history and in the current environment. Mentalism looks instead to inner agents such as a mind, a will or a trait. When that inner agent is inferred from the very behavior it is then used to explain, the account runs in a circle and is called an explanatory fiction.

Example

“Why does Leo leave his worksheets unfinished?” “He is unmotivated.” “How do you know?” “He leaves his worksheets unfinished.” The label returns the question as its own answer. A behavior-analytic account asks instead what happens right after Leo stops working, what else is available at that moment and what usually follows when he does finish.

Common confusion

Radical behaviorism does not deny that people think or feel. It counts those events as behavior governed by the same principles as public behavior. A private event can be a link in a sequence, such as telling yourself the bus leaves in two minutes before you start running, but it is not a first cause outside the environment; it has its own controlling variables. Methodological behaviorism and mentalism are different positions: the first sets private events aside, the second treats inner constructs as causes.

Key points (4)
  • Radical behaviorism includes private events (thinking, feeling) as behavior governed by the same principles; methodological behaviorism leaves them out of science because they are not publicly observable.
  • Mentalism explains behavior by inner constructs (mind, will, traits); a label inferred from the behavior and then used as its cause is a circular explanatory fiction.
  • A radical behaviorist explanation looks for causes in the person's phylogeny, learning history and current environment.
  • A private event can take part in a sequence of behavior, but it is itself behavior to be explained, not an initiating cause.

Philosophy, research and practice

The same field works in four domains. They are told apart by what each one is for, not by who takes part or where the work happens.

Figure 5

Four domains, told apart by their purpose

  1. Behaviorism

    Purpose
    Clarify what counts as an explanation of behavior
    Product
    Conceptual analysis
  2. Experimental analysis of behavior

    Purpose
    Discover basic principles and processes
    Product
    Basic research findings
  3. Applied behavior analysis

    Purpose
    Identify what improves socially significant behavior
    Product
    Applied research findings
  4. Professional practice

    Purpose
    Benefit a particular client
    Product
    Services
The purpose decides the domain, not the species, setting or participants. The arrows show the usual direction of influence; in practice each domain also informs the others.
A.4

Behaviorism, basic research, applied research and practice

Behaviorism is the philosophical side of the field. It works on conceptual questions, such as what counts as an explanation, and its product is analysis rather than data or services. The experimental analysis of behavior is basic research: its aim is to discover principles and processes, whether its participants are pigeons in a chamber or adults at a computer. Applied behavior analysis is also research, but its aim is to identify the variables responsible for improving behavior that matters socially, and to show that the improvement is due to them. Professional practice guided by behavior analysis delivers services to clients, using procedures that came out of that research, for the benefit of those clients rather than to produce new findings.

Example

A laboratory study varies how long a reinforcer is delayed to see how pigeons divide their pecks between two keys: basic research. A study that uses a reversal design to show that a self-monitoring card raises a teenager’s homework completion: applied research. An Instructor who introduces a self-monitoring card with a new student and adjusts it from his weekly data: practice.

Common confusion

The participants, the setting and the possible future uses do not decide the domain. Basic research can be done with people in everyday settings, and applied research can take place in a clinic. Ask what the work is for: clarifying a principle, demonstrating why a socially significant behavior improved, or serving a client.

Key points (4)
  • Behaviorism is the philosophy of the science of behavior; it produces conceptual analysis, not data or services.
  • The experimental analysis of behavior is basic research aimed at discovering principles; the purpose, not the species or setting, defines it.
  • Applied behavior analysis is research that identifies the variables responsible for improving socially significant behavior.
  • Professional practice guided by the science delivers services to clients, using procedures derived from the research, to benefit those clients rather than to produce findings.

The dimensions of applied behavior analysis

Seven dimensions are used to judge whether a piece of work is applied behavior analysis. They are independent: a study can meet several and still miss one.

Figure 6

Seven dimensions, seven checks

  • Applied

    Does the target matter in the person’s life?

  • Behavioral

    Is the behavior itself measured directly?

  • Analytic

    Is it shown that the intervention produced the change?

  • Technological

    Could someone repeat the procedure from the description?

  • Conceptually systematic

    Are the procedures tied to basic principles?

  • Effective

    Is the change big enough to make a practical difference?

  • Generality

    Does the change last, or appear in other settings or behaviors?

Each dimension is a separate question about the same piece of work. Answering yes to one says nothing about the others.
A.5

Dimensions of applied behavior analysis

Applied: the behavior chosen matters in the life of the person and of the people around them. Behavioral: the behavior itself is measured directly and precisely, instead of reports about it or states inferred from it. Analytic: the work shows that the intervention, and not something else, produced the change. Technological: every procedure is described completely enough that a trained reader could carry it out the same way. Conceptually systematic: the procedures are tied to the basic principles they come from. Effective: the change is large enough to make a practical difference for the person. Generality: the change lasts after the intervention ends, appears in other settings or spreads to related behavior.

Example

A program teaches a young adult to order at a café. Ordering matters to him (applied). Observers count his complete orders (behavioral). Training begins at different times in three cafés, and orders rise only once training begins in each one (analytic). The report spells out every prompt and correction (technological). It explains how differential reinforcement strengthens independent ordering and how fading prompts transfers stimulus control to the café’s natural cues (conceptually systematic). He goes from no independent orders to ordering on every visit (effective) and is still doing it months later at a café never used in training (generality).

Common confusion

Technological has nothing to do with devices or apps; it means written out well enough to replicate. Analytic and effective are separate: a study can demonstrate a clear functional relation for a change too small to matter, or report a large change without a design that rules out other explanations. Generality is about how long the change lasts and how far it spreads, not about how big it is.

Key points (4)
  • Applied means the target is socially significant for the person; behavioral means the behavior itself is measured directly, not reports or inferred states.
  • Analytic means a functional relation between intervention and behavior is demonstrated; effective means the change is large enough to matter in practice. A study can meet one without the other.
  • Technological means procedures are described in enough detail to be replicated; conceptually systematic means they are tied to basic principles. Technological has nothing to do with devices.
  • Generality means the change lasts over time, appears in other settings or spreads to other behaviors.

Keep going

Practice the whole chapter in the course

Original questions on every task of area A, with explanations and a record of which tasks you have mastered.

Open the course

Try a sample free, without an account