Reinforcement Isn’t Real

By Todd A. Ward, PhD, BCBA-D

President, bSci21Media, LLC

Yep, you heard me right.  Reinforcement isn’t real.  Not if you are a functional contextualist.  You see there are at least two dominant philosophical views in behavior analysis: mechanism and functional contextualism.  Skinner was never really clear on what he was, and he seems to have elements of both, but other people have devoted much work to parsing out these important issues.

But why should anyone focus on philosophy, you might ask?  Because to those who study philosophy it isn’t something “out there” — it is something expressed in our behavior every moment as behavior analysts.  You know that functional assessment you do with your kiddos?  How about the way you define verbal aggression topographically in your behavior plans?  You are enacting particular philosophies when you do these things and, if you do both of the things just mentioned, you may be enacting two conflicting philosophies.   

If you haven’t taken the time to figure out your philosophical position, then you might be doing things in a somewhat incoherent fashion.  You might call things “behavior” that really aren’t, you might say “reinforcement” when you really shouldn’t, and you might get lost when attempting interdisciplinary collaborations due to a lack of precise understanding of your own subject matter.

Philosophy defines our scientific subject matter, our analytic goals, and our worldview.  You see, no amount of data can tell you what these things are — they are “pre analytic,” they come before the data.  Most importantly, philosophy can distance you from your own verbal behavior and help you realize the difference between events and constructs — the latter constitute reactions of scientific workers to events around them.

A mechanist might say they are interested in “learning how the world works” and that “if we can just figure out how processes fit together, we will be able to understand our world.”  In other words, mechanists view the world as a machine, and it is our job to put together the parts of the machine…if we can figure them out in the first place.  Sometimes mechanists get lost in the “sandbox” of the basic laboratory, seeking “understanding” at the expense of prediction and influence.  Many times, mechanists talk about “causes” impinging on behavior and mean it literally.  They also tend to talk about reinforcement as something “out there” in the world operating on behavior.  To them, behavior is often localized on or in the organism.  

A functional contextualist, on the other hand, takes the view that behavior is a relation not localized on or in the organism, but as an act-in-context.  To them, this view extends to the behavior of scientists as well.  This means that everything we know about the world is a function of each individual’s unique history of operating in the world.  Looking at a set of data and saying “reinforcement” is a prime example.  The probability that someone says “reinforcement” in reaction to a set of data is probabilistically determined by that individual’s history.  And we never actually “see” reinforcement — we infer it from observations.  A functional contextualist is therefore a-ontological, they say nothing on the existence or non-existence of a “real” world in any absolute sense of the word.  To them, doing so misses the point — the prediction and influence of behavior.  

What do you think about these philosophies?  If you want to learn more, check out some of the recommended readings below.  Also, be sure to subscribe to bSci21 via email to receive new articles directly to your inbox!

Recommended Reading:

Hayes, S.C., Hayes, L.J., & Reese, H.W. (1988). Finding the philosophic core: A review of Stephen C. Pepper’s World Hypotheses: A Study in Evidence. Journal of the Experimental Analysis of Behavior, 50, 97-111.

Hayes, S.C., Barnes-Holmes, D., Wilson, K.G. (2012). Contextual Behavioral Science: Creating a science more adequate to the challenge of the human condition.  Journal of Contextual Behavioral Science, 1, 1-16.

Kantor, J.R. (1958). Interbehavioral psychology. Bloomington, IN: Principia Press.

Pepper, S.C. (1942).  World hypotheses: A study in evidence.  Berkeley: University of California Press.

Todd A. Ward, PhD, BCBA-D is President of bSci21 Media, LLC, which owns and  Todd serves as an Associate Editor of the Journal of Organizational Behavior Management and as an editorial board member for Behavior and Social Issues.  He has worked as a behavior analyst in day centers, residential providers, homes, and schools, and served as the director of Behavior Analysis Online at the University of North Texas.  Todd’s areas of expertise include writing, entrepreneurship, Acceptance & Commitment Therapy, Instructional Design, Organizational Behavior Management, and ABA therapy. Todd can be reached at

3 Comments on "Reinforcement Isn’t Real"

  1. Is “reinforcement” real? Does “reinforcement” exist as a biological property of organisms? To answer this question we need to consider the following (Note, the following is based on the writings of both A.C. Catania, 2012, and a review of A. C. Catania’s 2012 edition of “Learning” by J. Moore).

    The first question might be stated as follows:

    • “Why did responding increase?”

    The second might be stated,

    • “Why do we call the stimulus that is produced by the response a reinforcer?”

    1. To understand “Why did responding increase?” involves answering at least three separate sub-questions:

    The first sub-question might be phrased as,

    • “Why do we say that responding increases?” That is, what is the source of stimulus control for the tact “increases.”

    Presumably, one uses “increases,” “increased,” “improved,” “greater,” etc., etc., in connection with the comparison of numbers on a counter that, itself, derives from socially established and maintained verbal relational frames and networks of frames that connect same/different, comparative (e.g., “more” and “less”), spatial (e.g., “above” and “below”), temporal (e.g., “now” and “then”), deitic (“here” and “there”), and ordinal (e.g., first, then ___) relations that, together, establish a complex private stimulus event that is subsequently and metaphorically tacted as “reinforcement.”

    The second sub-question asks,

    • “In what ways is the increase in responding related to environmental circumstances?”

    Here we conduct an operational analysis (cf. Catania, 2012) to identify whether the probability of responding is “greater” when the responding produces some consequence as compared with when it does not.

    That is, if the charted data establishes the stimulus conditions that adhere to socially established and maintained criteria for tacting “greater,” then both EO and Sd conditions are established for tacting “reinforcement” and/or the complex autoclitic/tact “I see data that indicates the effect of reinforcement.”

    Note that the phrase “the effect of reinforcement” implies a causal relation that is, itself, socially established and maintained, even though it includes a ordinal relation that is not conventional—that is, “cause follows effect” versus “cause precedes effect. It is this lack of a conventional ordinal relation between cause and effect that is most likely behind non-behaviorist criticisms of the term “reinforcement” as being a tautology—i.e., that the term “reinforcement” describes both the cause and effect and is therefore useless.

    The third sub-question asks,

    • “Why does the consequence have the effect it does, namely, of increasing the probability of responding?”

    This question inquires into the phylogenically determined physiological and/or genetic structure of the organism that contributes to the observed behavior. Organisms that are not sensitive to the consequences of their actions are not likely to survive. Susceptibility to influence from environmental stimuli, of which a “reinforcing” consequence is surely one sort, is a characteristic of a behaving organism, and as such is presumably inherited. In any case, the answer to this third question does not necessarily involve a functional relation. In this third sense the question will be answered by physiologists, rather than by psychologists, although psychologists will almost certainly give the physiologists some important clues regarding what they should be looking for.

    2. To ask, “Why do we call the stimulus that is produced by the response a reinforcer?” is to ask why we label a stimulus as a reinforcer. This question inquires about the complex stimulus control that evokes the term “reinforcer.”

    In our language, we use the term reinforcer in connection with certain observed relations between responding and its consequences. Catania neatly specified these relations in Chapter 4 of his book “Learning”: (a) the responding must produce the consequence, (b) the responding must increase in probability, and (c) the increase in probability must occur because the responding produced this consequence. The term reinforcer is not used when these criteria are not satisfied.

    Thus, behavior analysts explain behavior by specifying the contingencies of reinforcement that evoke and maintain the behavior of interest. In part, that does mean specifying the relation between responding and consequences, but it does not mean endowing reinforcement with some sort of mystical potency that connects antecedents, responses, and consequences together. To be sure, specifying that a response does produce some consequence is part of a causal analysis, but to say that reinforcement caused the response is no more meaningful than to say that stimulus control caused the response. Catania’s treatment of this is subtle, but one does not go wrong in following his lead.

  2. Todd, great post! I see the consequence when working with fellow behavior analysts that lack exposure and understanding of functional contextualism in practice regularly. It would be fantastic if this invited curious behavior analysts to contact some of the relevant readings. It really does shed light on many mundane issues.

    Richard, your comment was also appreciated. I am looking to revisiting Catania in this subject.

  3. I think philosophy is hugely important, not just to understand our own views for coherence sake but to also get to better accounts and answering philosophical questions (coherence and experiential anchoring). For example, the question of is “reinforcement” real is a philosophical question if we are asking about ontological status of “reinforcement” and what it means to be “real”. When looking at the philosophical literature, this debate occurs regarding all scientific concepts and laws as well as the differences in how people define “real”.

    First thing we should probably avoid in this discussion are simple (false) dichotomies between mechanistic vs contextualistic thinking as there are many views in between. James Herbert recently wrote a good article in JCBS highlighting this very point regarding the place of ontology in FC.

    For example, there is no contradiction in saying that “reinforcement” is a form of verbal behavior that guides the behavior of scientists and also say that “reinforcement” also is real in the sense that it happens the way it happens independent of how we believe it to be (explains a real relationship in the world).

    Let me unpack that last piece as it’s not easy to convey in a simple sentence. To say that reinforcement is real is to say that the probability of behavior increases following said behavior being reinforced only and if it behavior in fact increase. Reinforcement is not real if we make the claim above but no increase in behavior actually occurs of I thought otherwise, the situation would be different based on me thinking otherwise. In this view, real contrasts with fiction or imaginary. I can make up stories, I can say I can fly but that wouldn’t make it “real” but precisely “fictional” because it depends on how I believe it to be. Reinforcement I don’t think is fictional. It really does happen as does many of the other wonderful things in RFT, ACT, and CBS, and I can’t think otherwise but that won’t change it.

    This is separate from the very specific realist ontological position that this piece hints at by saying is “reinforcement” out there, perhaps in some platonic form or other variations of realism. However, such views conflate concepts like “real” in these sense I described above with “exist” in a concrete sense. The number 2 is real but it does not exist out there in the sense that if we keep exploring the universe, we will find a number 2 running around on a distant planet. However, that is not the only type of real nor even the most popular form of “real” used by scientists or philosophers of science and we should be careful to not make important distinctions when having these discussions.

    It’s especially important when communicating to scientific community at large as well as the public.

Leave a comment

Your email address will not be published.