Who Redefined Operant Conditioning into the Quadrants of Dog Training?

This article is really only going to be of interest to you if you are a behavioural science nerd, I am not going to offer any apologies for this being a nerdy article. It is what it is.

This is less article, more explanatory timeline.

What is the premise that I am writing this article with? There are two;

1) The quadrants of dog training are an erroneous interpretation of operant conditioning. I’ve written about this in previous articles, but I have not explicitly traced the timeline from Skinner to today. I will lay this out in this article.

2) There is a theory floating around that the quadrants of dog training was something dog trainers and vet behaviourists created 30 years ago by making a mess of B.F. Skinner’s work. This is categorically untrue and I will demonstrate that.

Prof Skinner published “The Behaviour of Organisms” in 1938. In that book he described Operant Conditioning in a mechanistic or technical sense, writing about how the changes of behaviour were impacted by changes in the environment; he was studying the effect of reinforcements and increase or decrease in behaviour due to the consequences of actions.

The necessary context of the purpose of operant conditioning did not appear until his later works, the most concise and clear version of which came in 1974 in About Behaviourism, this is where Skinner connected his lab based experiments to the real world. The real word context that Skinner introduced after his lab work cannot be understated in its importance to being able to understand behaviour, and as I’ve previously written, this is the biggest problem with what passes as operant conditioning in today’s common understanding and use.

Skinner moved from working with operant conditioning as a rigid process to bring in the philosophical and functional understanding that the process exists as an evolutionary mechanism geared wholly to ensuring survival of the individual and species. This was a little known groundbreaking addition to the theory, but got overtaken by events before Skinner even published it, and as I’ll get to later, this is the first step in the divergence of Skinner’s operant conditioning to the common version in use today.

Skinner’s later addition to the theory explains why the animal's emotional state matters with respect to responses to stimuli, and therefore why competing motivators exist. To give a really narrow specific example, Skinner explains why a treat offered to a dog when there is another dog 5 metres away is not a positive reinforcement to the dog at all - the evolutionary survival and mating brain functions are in full swing trying to assess the potential threat right ahead of the dog, so of course the treat is just an annoyance, it’s been out competed in the hierarchy of evolutionary needs i.e. something else in the vicinity is a stronger reinforcement and therefore gets the attention.

Skinner’s original concept of operant conditioning as a mechanistic procedure, published in 1938 survived less than 12 years in the initial form. So who changed it?

Psychologists. Professors and Dr’s at prestigious universities in the USA in the late 1940’s.

Let me get specific; the two key players in the 1940’s were Prof’s Keller and Shoenfeld. They were followed in the 60’s, 70’s and 80’s by Prof Jack Michael.

Keller and Schoenfeld took Skinner’s 1938 works, agreed with it, but decided that the terminology was not precise enough and introduced additional clarity in the procedure; The field of behaviourism realised that Skinner was using the words positive and negative to describe both the stimulus properties and the administrative procedure, and this did not stand up to academic rigour. The blunt truth of that academic rigour was that freshman students couldn't get their heads round the imprecise language, and this annoyed Profs Keller & Schoenfeld. Skinner didn't seem to care.

Keller and Schoenfeld delineated between what the experimenter does and what happens to the organism’s behaviour, in short, they forced the process to make it clear whether the experimenter was augmenting or diminishing the environment. To get even more precise, they turned operant conditioning into a mathematical formula;

They came up with a shorthand notation; ($ + $) and ($ - $) to represent the nature of the stimuli (appetitive vs aversive) which later got mapped onto the procedure of operant conditioning (Prof Jack Michael was involved in this) which became the concept of adding or removing a stimulus.

The peculiarity of Keller and Schoenfeld’s work is that they were convinced that operant conditioning was a system of logic, which has some merit, it's where Skinner started, but when the philosophical and evolutionary mechanisms of Skinner in the 70’s is placed into the conversation, we get a logic that does not play in nice straight lines, it gets messy in places, because emotions and instincts get involved.

Before I get into Prof Jack Michael, I need a quick mention of C.B. Fester, one of Skinner’s closest collaborators (co-author with Skinner of Schedules of Reinforcement 1957) and M.C. Perrott, who took Fester’s work and published “Behaviour Principles” in 1968 which contained the following phrases:

Reinforcement: Always increases behaviour

Punishment: Always decreases behaviour

Positive: Always adds a stimulus

Negative: Always subtracts a stimulus

Can anyone see the quadrants in those statements?

So the controversy is this: Without Skinner’s co-author of Schedules of Reinforcement, Prof Fester, author M.C. Perrott would not have come up with and printed the 2×2 grid which became the foundational material for the use of operant conditioning in Psychology in the 1970’s and onwards thanks to the influence of Prof Jack Michael.

None of these people are dog trainers. This is important.

So Prof Jack Michael. What is his role in all this?

Prof Michael was the person who wrote the behaviourist courses for many university students, he, through his own works and published papers, became something akin to “the guy” in terms of behaviourists, which is why he had such influence on the field and was the architect of how behaviourism was taught post the 1960’s. He specifically set out to communicate behavioural experiments in unambiguous language and terminology. He was instrumental in preventing stimuli being explained in terms of how they feel, instead he forced students and therefore the next crop of behaviourists to describe stimuli in terms of what we do with them. This was a departure from Skinner, who was beginning to wrestle with needing to include the “feeling” or what was going on in the brain’s limbic system to understand behaviour.

Prof Jack Michael, perhaps counterintuitively, reinforced the need to understand an animal’s motivation to understand the behaviour and therefore influence it. He posited that if you do not understand the motivation of the animal, you won’t know if a positive or negative reinforcement would be the correct response to influence the behaviour.

His 1982 paper “Distinguishing between discriminative and motivational functions” is considered his greatest piece of work (not my words), from there I have this example (paraphrasing); “a no entry sign on a door only works as a reinforcing contingency to stay away if you wanted to go inside, otherwise it has zero bearing on your behaviour at all.”

So I deduce that Prof Michael meant that the innate desires of the animal are vital in bringing understanding of behaviours. I absolutely agree, and so did Skinner. How can we see those innate desires, they are in the brain? Good question. I’m not answering that here though.

Prof Jack Michael was not above controversy himself, in one paper; “Positive and Negative Reinforcement, A Distinction That is No Longer Necessary” he proposed exactly that; there is no longer value in delineating between positive and negative reinforcement, instead we only need reinforcement. He goes on to cite Skinner who said that punishment is defined as the withdrawal of a positive reinforcement or presenting of a negative reinforcement. Prof Michael’s proposal to use reinforcement and punishment without positive or negative prefixes was rejected by the community. I find this an odd proposal from someone who was borderline obsessed with clarifying language and definitions.

Incidentally, Prof Michael agreed with Skinner here that the removal of a negative reinforcement is felt by the animal as a positive reinforcement, and vice versa for the removal of a positive reinforcement. But isn’t that punishment? Welcome to the imprecise language game.

Punishment is only the definition if behaviour is reduced. If it is increased, its reinforcement, but it is felt as punishment emotionally. You’ll go round in circles if you’re not careful. Remember Skinner became emphatic on the need to involve emotions in operant behaviour, the remainder of the field wanted to get as far away from that as possible, as evidenced by the beginning of the 2×2 matrix and mathematical application (see a few para’s above). Though Prof Michael’s assertion that understanding the animal’s motivation for behaviour is key doesn't sit neatly in that paradigm. It seems that part has gotten lost in the mists of time.

Here is Skinners 1974 definition of Operant Conditioning: “A very different process, through which a person comes to deal effectively with a new environment, is operant conditioning. Many things in the environment, such as food, water, sexual contact and escape from harm, are crucial for the survival of the individual and the species, and any behaviour which produces them therefore has survival value. Through the process of operant conditioning, behaviour having this kind of consequence is more likely to occur. The behaviour is said to be strengthened by its consequences, and for that reason the consequences themselves are called reinforcers.”

This is the macro level definition, not the micro, but understanding this context first is truly vital. Everything operant conditioning has to be viewed through the lens of the animal undergoing the process and what it is trying to achieve in the moment as well as in the bigger picture of its life, ignoring parts of that puzzle will lead to very peculiar outcomes.

So now we arrive into the 1990’s. The name that we need next is Dr Mary Burch. Dr Burch studied at Auburn university. She is a human Psychologist. Curiously she branched out and worked closely with American College of Veterinary Behaviourists and American Veterinary Society for Animal Behaviourists. She was not a founder or graduate of either.

Dr Burch was taught by material developed by Prof Jack Michael. So Dr Burch is a fully paid up member of the quadrants of operant conditioning. Why Dr Burch is important in this story is for two reasons:

1) She brought human psychology into the dog training world (Karen Prior and Dr Ian Dunbar also feature here) via her psychology degree and doctorate. This was allegedly to counteract the “dominance theory” style training that was prevalent at the time;

2) Dr Burch linked Col. Konrad Most’s “Dog Training; A Manual” from 1910 to the 2×2 grid version of operant conditioning, but referencing Skinner (an error as the 2×2 grid does not appear in any of Skinner's published works) in an article called “A Toast to Konrad Most” in 1990. What is particularly ironic is that Konrad Most is the author of the dominance methods Dr Burch was so against. If you have read K. Most, you’ll be familiar with his instructions for punishing and / or beating dogs into submission; if you can’t get a dog to behave properly through primary and secondary inducements and compulsions (yes that’s all the rewards, treats, prong collars and the beatings too) then dispose of the dog, its broken.

Yes, Dr Burch mapped Most’s work to the development of operant conditioning as coined by her psychologist predecessors. Specifically, K. Most’s definition of primary and secondary inducements and compulsions. A quick translation for you: inducement = positive reinforcement, compulsion = negative reinforcement or punishment. She almost went as far as crediting K. Most with discovering operant conditioning. This article is long enough already, I’m not tearing that assertion down here, but for the record, I disagree 100%, perhaps a glance at Skinners full operant conditioning definition above will assist you in becoming confused as to how the quadrants relate to it.

I know K. Most is from a different time, and we have to take his book in the context it was intended; training GSD’s for the first world war, nothing else. It was never intended to be a pet dog book. K. Most was friends with a certain Herm Sprenger by the way.

At some point since 2000, the dog training community split into its balanced training and force free training paradigm, and both camps claim they are right ones and they are using behavioural psychology correctly. All dog trainers actually did was follow one of two camps: K. Most (balanced or compulsion training) or Dr Mary Burch / Karen Prior / Dr Dunbar in what became Force Free.

Well, if you know me by now, you’ll know what I’m about to say. They are both wrong.

I am an unashamed Skinnerian behaviourist in my leanings, and in my personal experiences, based on all the material I’ve read from the above names, the one that makes sense and works in practice is Skinner’s “About Behaviourism” version of operant conditioning, which I quoted a few paragraphs back. The quadrants make perfect sense in a laboratory setting, no quibble there from me, but out in the real world, they are woefully inadequate to explain how animals learn to navigate the world around them. Skinner’s inclusion of evolutionary psychology and internal motivations is the crucial distinction that makes it work over and above the quadrants.

I am not the only one to think this either. If you want credentialled persons who share my views, Dr’s Perone and Stadden are a good start. I’m sure there are others.

So in summary, against my initial premises:

Operant conditioning has been redefined. Skinner refined his own works, but it seems the world took little notice, and built the quadrants without understanding macro level context and motivations, I think that is a fatal error.

Dog trainers and dog training had nothing to do with redefining operant conditioning. That was all human psychologists. Psychologists created the quadrants, but K. Most discovered his own version of them by trial and error. The two didn’t meet until 1990.

It’s one of those weird coincidences that what K. Most wrote about maps so neatly onto the quadrants of operant conditioning. A more appropriate name for K. Most’s discoveries and the quadrants would be “a system of autonomous behavioural influence via applied environmental manipulation”.

Skinner’s later addition of evolutionary psychology to his operant conditioning discoveries have in my opinion and experience brought about the best way of understanding how to overcome behavioural problems with dogs, but by that time, his own colleagues and students had run away with his original idea and turned it into something else that is not a good fit for anything beyond the strict controlled environment of a laboratory.

If you wish to understand why I think this way, my other two blogs What is Operant Conditioning? and How Does Operant Conditioning Work? will be of interest.