Dog training: a science or an art? Surely it’s a mixture of both. We may be a scientist or at least be scientific in our approach, but if we have little rapport or empathy with dogs (or animals generally), lack the personality or temperament of a trainer, then it’s a non-starter. On the other hand, I wonder how many shepherds have read ‘How Dogs Learn’? You don’t need to be a mechanic to drive a car. Just do it! As with any profession, there is an element of jargon and sometimes a tendency to make things more complicated than they actually are – ‘reinventing the wheel’ – even over analyzing; but isn’t that what scientists do?
In Part 2 of this series I spoke about the International Dog Trainers Winter Summit 2020. Yesterday Bob Bailey gave an engaging presentation. He spoke about “science, or the scientific method as being a systematic way of asking questions and making it difficult to lie about the answers……………..science is a process of studying the complex world around us……………….scientists usually break apart, or simplify, complex phenomena into its component parts. Science APPLIED to solving specific problems is one definition of TECHNOLOGY. The technology of animal training involves the sub-sets: psychology (and ethics), biology, biochemistry and mechanics.”
This will be the last blog in the current series. What follows is a random selection of theories that have, hitherto, been a source of bemusement to me. A THEORY is just that; it is said to be scientific until it has been falsified or becomes an enshrined FACT.
THE QUADRANT OF PUNISHMENT
This takes the science of dog training to a new level! I mentioned the danger of OVER analyzing and here I think we may have ‘lost the plot’ (see next item). Trainers will understand the concept of ‘The Quadrant of Training’. However, how many understand, or have even heard of, ‘The Quadrant of Punishment’? Cesar Millan, in his book ‘Cesar’s Rules’, discusses this, though I should point out he does not take credit for it! The four quadrants of punishment are:- 1. ‘aversive punishment’, 2. ‘non-aversive punishment’, 3. ‘aversive non-punishment’, 4. ‘non-aversive, non-punishment’. 1 and 2 are self explanatory. However, non-punishment implies INACTION or the opposite – REINFORCEMENT!
It is possible to over analyze an action rather than let it occur naturally. The over analysis of an action or string of actions, for example in sports, could lead to paralysation and non-action or incorrect action. A goal keeper about to defend a penalty shoot will have to decide whether to dive to his left or to his right. As the ball leaves the foot of the striker he must make an analysis resulting in a nanosecond of inaction or freezing (then he dives the wrong way!)
DIFFERETIAL CLASSICAL CONDITIONING (DCC) Vs. DIFFERENTIAL REINFORCEMENT OF AN ALTERNATIVE BEHAVIOUR (DRA)
Differential = 1. the relationship between two points on a sliding scale, 2. graduated.
Also in Part 2 I mentioned Ian Dunbar‘s presentation at the Summit and gave examples of Ian’s theories. However, the thought processes firing in a dog’s mind are, of course, complex and we can only guess as to what a dog is thinking (an educated guess nonetheless). Going back some years to his ‘Growl Class’, Ian spoke about ‘Differential Classical Conditioning’ (DCC). This was in relation to a dog-reactive dog lunging and the following string of events to avoid a potentially dangerous situation arising. (However, there is little reference to DCC in dog training books or the internet other than the defensive withdrawal reflex of a sea hare (slug) from a noxious event – an electric shock!) Ian went on and concluded the class by stressing the importance of giving the dog feedback for ALL behaviours – i.e. lunging, then withdrawing, followed by the WANTED behaviour – not simply ignoring the unwanted behaviour! Think of our sliding scale analogy.
Grisha Stewart discusses in her book, ‘Behavior Adjustment Training 2.0’ (BAT), a procedure referred to as ‘Differential Reinforcement of an Alternative Behaviour’ (DRA). She describes this as extinguishing an unwanted behaviour by withholding reinforcement for it whilst reinforcing an alternative behaviour. I quote, “the handler is NOT applying ‘differential reinforcement’, rather ‘differential reinforcement of an alternative behaviour’. With BAT, we don’t apply operant extinction, because it’s not ethical………………to force a dog to stay in a stressful situation just because he is not offering the ‘right’ behaviour to get out of it. If the dog [lunges], for example, we move him away from the trigger…………….it may however reinforce the [lunging]. That’s risky but not as bad as it sounds. We can just arrange our set-ups so that [lunging] or other such behaviours are unlikely. That way the dog gets lots of reinforcement for behaviour we want, and not much reinforcement for the behaviour we are trying to eliminate. This has a similar effect to differential reinforcement, without the added stress and disempowerment of extinction.” (Stewart, 2016).
Of course, as responsible handlers, we will be looking ahead for potential areas of conflict with a view to avoidance as Grisha espouses. In the Summit, Jamie Pound later spoke about the importance of ‘scent work’ and the positive affect this has on a reactive dog. For example the laying of a food reward as a means of desensitisation (in this case, to another dog) and counter conditioning – here the two are NOT mutually exclusive. Dogs were born to sniff (and run)! Sniffing is self rewarding and signals sent to the brain induce a self calming effect. We could argue that the reactive dog has simply been distracted, but nevertheless, he is aware of the trigger in the distance. The dog, in effect, is handed control of the situation – a fundamental philosophy of BAT!
An easier concept to understand may the the scenario of our pet dog sitting under the meal table whining. We can do one of two things. 1. Apply ‘differential reinforcement’ by offering ‘not much‘ reinforcement or, indeed ignoring the behaviour, praising for NOT whining and relying on extinction – though Grisha describes this as disempowering (for the dog)! 2. Apply ‘differential reinforcement of an alternative behaviour’ by luring the dog away with, for example, a toy (preferably NOT food) to his anchor mat. The danger here is that the dog may perceive this as reward for whining! Only WE can judge.
THE PREMACK PRINCIPLE – GRANDMA’S RULE
Sometimes referred to as ‘Premack’s Principle’ named after the researcher David Premack in 1959. The idea here is that a high-probability behaviour reinforces a low-probability behaviour. For example, Grandma explains to her Grandchild that he can have his sweet providing he eats his greens! So, if he eats his greens (low probability) he gets to eat his sweet (high probability). The child has been bribed! This principle is more commonly applied in a classroom setting with children (or adults) and would not necessarily apply to dogs because it requires self management and understanding of what is in one’s best interest.
However, in dog training it has the potential to increase impulse control and/or recall probability. The disadvantage here is that three handlers are required – one to hold the dog, one to call the dog and one to apply the high-probability behaviour. In the scenario our dog is aware that handler 3 has a high value reward (a piece of smelly frankfuter) – but handler 2 (preferably his guardian) is calling him. This gives him a predicament. Does he control his impulse to run to handler 3 or run to his guardian (handler 2) in order to reach his ultimate goal – the frankfurter? To help the dog, handler 1 may ‘point’ the dog in the right direction! Unfortunately we will be applying negative punishment (P-) by not allowing the dog access to the reward for the ‘wrong’ behaviour, thus adding the element of frustration and the dog potentially walking away in disinterest.
The following related blogs may be of interest:-