Generalisation Versus Specialisation


The above picture will look completely abstract to the vast majority of readers,,, and that is exactly how I want it.

You see, until 2014 I was involved with American Football at the semi-professional level in the European leagues. During my years as a coach I had a few wonderful opportunities to push myself and lead others in their pursuit for athletic success.

Then came my opportunity to start to strategise and take control of the game in a totally different manner. You see American Football is much more fragmented than the Rugby codes, or Aussie Rules. Each play is a microcosm of the game itself. Moves and counter moves, layers of fundamental skills heaped upon athletic prowess and quick wits. Each play has several nuances and specialties that you cannot prepare for in other codes of contact ballsport.

Let us look at the above diagram of a specific play I coached. On the left side you will see two circles labelled X and L and on the right there are two circles labelled R and Z. Earlier on I would hold this card up and the players - represented by the letters - would run the play as the red arrows directed them. Once we had that down, I would send A - the circle in the middle at the bottom of the picture - to the left side. Remember that we had the play down pretty good as it is shown in the picture. What do you think happened?

Sit on the beach

Let us perhaps look at something a little more canine in its application. Something like a simple pair of obedience exercises. Something like Sit and Down are fairly simple movements any dog owner knows and has had trouble with in the past.

You work on Sit until you have your dog able to Sit and wait for a period of time at home, no worries. It did not take too long and your dog was able to focus on you perfectly. Then you started cooking dinner and your dog could not Sit still. No matter how often you said Sit or Stay, your dog simply could not comply with your command.


Down at home

How about another well known command, the Down? This is an even better example. Your dog Downs perfectly when your watching a movie or you are eating dinner. Not a problem for when guests arrive at your house either. Your dog even Downed well at that BBQ you had.

But then you went and walked your dog to the beer garden and would your dog stay in the Down? Not a fat chance in heaven.

You see, dogs reflect us very well, we simply do not generalise well either. This is why fast food chains, supermarkets and other chain stores tend to organise themselves into a similar fashion. Familiarity breeds return. Novel environments get in the way as do novel situations.

As soon as we inject a situation with a variable we have to think, those items that where crutches before are now exposed as being crutches. If a Give Way sign disappeared from an intersection, how many people would stop compared to give way? The crutch is gone. To put things into a behavioural perspective: If we set up a conditioned stimulus (Sit) to a conditioned response (Butt hits the surface and stays there) and then attain fluency; we are gold!

If we then change something about that, say for example you are in a shallow pool, your dog will enter the Sit but will hold it so that their butt is on the water, not the pool. Try it out. If we set up a down at home, then move out to a curb on a busy road we have enough variance in our stimulus that our dog will vary their response to such an extent that they may not comply with a Down at all! They may Sit or stand there looking at you.

This is due to the fact that a neutral (hitherto unknown) stimulus has affected the communication chain and is now clouding the student dogs processing capacity. However, since the fundamental is in itself still conditioned - validated by compliance in a known environment - this inhibition will not last long. You see, when I flipped the roles around for X, L, R and Z they ran around doing pretty well...anything, believing they where right. But after a few collisions and discussions I realised that they where indeed all right in being all wrong.

They where used to being taught in a very specific way and I simply took it for granted that if I said flip the play, that they could all jump on board, all 11 moving parts. Had I printed the mirror image of the play, there would not have been any problems for them. It is exactly the same for our dogs, if we take the time to show them the variability that we require. The common vernacular would be generalisation, since the skill set is able to be produced in a variety of situation it can be referred to as a generally applied skill. I prefer to look at this as specialisation. Most of the people I talk to find this much easier to understand. We teach our dogs to apply known skills to novel situations, we do this by enabling them to overcome their own temporary inhibitions which are created by the novelty of the new situation. In the vast majority of cases we can use differential reinforcement strategies to improve the level of specialisation. In these cases we can use extinction by giving our dogs the "Wrong" signal and the re-issuing the desired command. In most cases this will be over and done with pretty quickly, with correct reinforcement, through the correct application of the "Yes" signal, we can quickly disinhibit our dogs and shape the correct action in a very similar way to how you will have taught the progression at home in an otherwise controlled environment.

Extinction works fine here, since we are not looking to eliminate a behaviour, we are looking to call upon a behaviour. When we add enough novelty into a scenario to create inhibitions, that is non-compliance, we are seeking to establish a go-to behaviour for our dogs to perform. Their previously inhibited behaviour is not reinforced and as such our reinforcing a specific behaviour biases the student dog heavily towards desired compliance efficiently. Unless the distractions are of such a magnitude to create intense conflict within your dog, the disinhibiting process - Specialisation - will not take more than a few attempts. Certainly not the 25-50 attempts as previously required to condition an unknown behaviour. Keep this in mind the next time you are in a novel environment for your dog, such as a group class, the dog park, the beach or other such place and your dog is "acting up". Revisit the training sessions, did you provide enough progression to be certain of a fundamental level of fluency in the behaviour previous to your dog "acting up"?

#extinction #generalisation #specialisation #inhibition #disinhibition #behaviour #sit #down

Featured Posts