Science into Practice: Labels Don't Train Dogs. You Do.

Somebody asks me a version of this question during almost every training: In which quadrant does this fall?

It’s a reasonable question. R+, R–, P+, P–. The four corners of the operant quadrangle. Every trainer learns them, most of us can label what we’re doing in them, and a good chunk of internet debates live and die inside them.

Here’s the thing. The quadrants were never meant to tell you what to DO. They were built to DESCRIBE what had already happened. Skinner was doing a functional analysis of consequences, sorting them by whether a stimulus was added or removed and whether the behavior got stronger or weaker (Skinner, 1938). That’s a classification system, not a training plan.

When “which quadrant” is your first question, you’ve already narrowed the whole problem to consequences. That’s a lot of training territory you just gave up.

What the quadrants don’t see

Motivating operations. Satiation and deprivation. Arousal. Whatever the dog brought to the session that shifted the reinforcer's value BEFORE the behavior occurred. A cookie that was worth a sit yesterday might not be worth a sit today, and no quadrant change fixes that (Laraway, Snycerski, Michael, & Poling, 2003).

Antecedents. The cue, the setup, the environment, the handler position, and the time since the last repetition. Most “the dog won’t do it” problems turn out to be antecedent-arrangement problems. Moving where the handler stands alters things that no amount of reinforcement-schedule tinkering will.

The respondent layer. Classical conditioning runs underneath every operant procedure you use (Rescorla, 1988; also see the last blog). The environment, the cue, your body language — all of it takes on reinforcing or aversive properties. You can be delivering R+ with perfect timing on top of a conditioned stimulus that’s making the dog uneasy. The quadrant is clean. The picture is not.

Behavior chains and differential reinforcement. A behavior can be reinforced by the chance to do the next behavior. You can reinforce one response by weakening another (DRI, DRA, DRO, DRL). Those are compound procedures, and they don’t live neatly inside a single quadrant (Cooper, Heron, & Heward, 2020).

Take Raven.

We were working on climbing and environmental work — getting her to commit to novel, unstable surfaces and STAY there. I was doing what has worked with the majority of my dogs and looks like textbook R+: close to the object, mark the contact of all four feet when on the object, deliver food. By any quadrant audit I was adding something pleasant.

Except that the behavior was going in the wrong direction. She hesitated longer. Offered a shorter duration of time on the platform. When she did get on, she jumped off as if it were electric. If I’d stayed inside the quadrant frame, I’d have said I was reinforcing and tried to clean up my timing or raise the rate. Which I had already done both in previous sessions.

What I actually needed was to back out of her space and switch to a toy. Same problems.

No hesitation.

The procedure I’d been running wasn’t R+. The behavior I wanted didn’t increase. My body pressure was a punisher, and the food was riding on top of it. The quadrant label was never going to tell me that. The behavior did.

Better questions to ask instead of “which quadrant.”

· What’s the motivating operation right now?

·  What’s the antecedent picture? Is the cue clean? Is the context clean? Is my position helpful?

·   What’s the dog’s arousal and motivational state, and is anything here a cue for something I didn’t intend?

·   What else is being reinforced right now that I’m not thinking about?

·   Am I trying to teach a new behavior, maintain one, strengthen it against a competing response, or put it on a cue? Because the procedure for each of those is different.

The reframe

The quadrants aren’t wrong. They answer a narrow question — what kind of consequence did that behavior get — and they don’t scale up to “how do I train this dog.” Treating them as the whole map is what limits the training.

Stop asking which quadrant you’re in. Start asking what the dog is actually responding to and if your behavior is increasing or decreasing. Those are not the same question.

Be curious.

Robin Greubel

K9Sensus Foundation

References

Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied Behavior Analysis (3rd ed.). Pearson.

Laraway, S., Snycerski, S., Michael, J., & Poling, A. (2003). Motivating operations and terms to describe them: Some further refinements. Journal of Applied Behavior Analysis, 36(3), 407–414.

Rescorla, R. A. (1988). Pavlovian conditioning: It’s not what you think it is. American Psychologist, 43(3), 151–160.

Skinner, B. F. (1938). The Behavior of Organisms: An Experimental Analysis. Appleton-Century-Crofts.