Saturday, December 26, 2009

The Principles of Operant Conditioning

I have briefly discussed Operant Conditioning in a previous post after looking at it I thought a better explanation is required. Operant conditioning is a basic tenant of animal behaviour and learning. It says that all living things repeat a behaviour that is rewarding to them and deminish behaviours that make bad things happen. The four principles of operant conditioning are:

Positive reinforcement:(written in behavioural shorthand as "R+") The dogs behaviour makes a good thing happen, so the behaviour increases. He sits so you give him a treat. He likes getting treats, so he sits more.
Positive punishment:(P+) the dogs behaviour makes a bad thing happen, so the behaviour decreases. He jumps up on you so you knee him in the chest (not recommended , I will blog about this another time). He doesn't like the knee in the chest, so he jumps up less.
Negative punishment: (P-) The dogs behaviour makes a good thing go away, so the behaviour decreases. When he juimps up to grab the ball from your hand, you hide the ball behind your back. He doesn't want the ball to go away so he jumps up less.
Negative reinforcement: (R-) The dogs behaviour makes a bad thing go away.A puppy struggles when restrained, so he is held until he becomes calm and them you let go of him. Teaching the puppy that calm behaviour makes the restraint go away. He does not want to be restrained, so he learns that to be calm in order to make the restraint go away.

Because training methods that involve intimidation, coercion, and physical force can cause undesirable side effects, including fear and aggression, positive trainers use primarily positive reinforcement and especially avoid positive punishment. When using negative punishment it works best with it works best when using positive reinforcement for the behaviour you want instead. So using our earlier example of hiding the ball, when he sits you throw the ball (R+) so he sits more and jumps up less.

Tuesday, December 8, 2009

so how does operand conditioning positive reinforcement training work

All animals learn the same. This is a fact whether dog, chimpanzee, cockroach, rat or human!

In the past and most evident in dog training a method of punitive discipline was  (and in some places still is) the preferred teaching methodology
If we look in the dictionary we find the following definitions 
punitive is relating to, involving, or with the intention of inflicting punishment.   
discipline Control obtained by enforcing compliance or order.

I think the best examples of seeing this in action is watching a dog wearing a choke/check chain being reefed around untill the owner can place it in the preferred position. The general mesasge being sent to the student or dog in this case is "do this or else". This is not to say that at times that these repercussions are necessarily inappropriate but, rather, that this method of education seems to frequently mean living up to the expectations set by the trainer/educator rather than working with that subject.

Burrhus Frederic Skinner (March 20, 1904  August 18, 1990) was an American psychologist, author, inventor, advocate for social reform. He is one of the great reformers of pedagogical theory. 
His early paper on Radical behaviourism led to his founding theories on reinforcement processes. His theory hypothesizes that positive reinforcement is the strengthening of behavior by the application of some event (e.g., praise after some behavior is performed), negative reinforcement is the strengthening of behavior by the removal or avoidance of some event (e.g., opening and raising an umbrella over your head on a rainy day is reinforced by the cessation of rain falling on you). Both types of reinforcement strengthen behavior, or increase the probability of a behavior reoccurring, the difference is in whether the reinforcing event is something applied (positive reinforcement) or something removed or avoided (negative reinforcement). Punishment and extinction have the effect of weakening behavior, or decreasing the probability of a behavior reoccurring, by the application of an aversive event (punishment) or the removal of a rewarding event (extinction).

So in carrying this theory further he also found these roadblocks to animal education.

  • a fear of failure.
  • The task is not broken down into small enough steps.
  • There is a lack of directions.
  • There is also a lack of clarity in the directions.
  • Positive reinforcement is lacking.

Skinner suggests that any age-appropriate skill can be taught using five principles to remedy the above problems

  • Give the learner immediate feedback.
  • Break down the task into small steps.
  • Repeat the directions as many times as possible.
  • Work from the most simple to the most complex tasks.
  • Give positive reinforcement. 

Part of Skinner's analysis of behavior involved not only the power of a single instance of reinforcement, but the effects of particular schedules of reinforcement over time.

Skinner's types of schedules of reinforcement involved:
 interval (fixed or variable) and ratio (fixed or variable).Continuous reinforcement - constant delivery of reinforcement for an action; every time a specific action was performed the subject instantly and always received a reinforcement. This method is prone to extinction and is very hard to enforce.

Interval (fixed/variable) reinforcement (Fixed) reinforcement is set for certain times. (Variable) - times between reinforcement are not set, and often differ.
Ratio (fixed or variable) reinforcement (Fixed) - deals with a set amount of work needed to be completed before there is reinforcement. (Variable)amount of work needed for the reinforcement differs from the last.

Pigeons are superstitious
Skinner placed a series of hungry pigeons in a cage attached to an automatic mechanism that delivered food to the pigeon "at regular intervals with no reference whatsoever to the bird's behavior." He discovered that the pigeons associated the delivery of the food with whatever chance actions they had been performing as it was delivered, and that they subsequently continued to perform these same actions.

This comes into play when thinking about the timing and placement of a reward. For example if your dog goes over a jump and then you go to reward the dog for going over the jump it comes to you and sits patiently waiting for you to get the reward, the dog will think that it was rewarded for sitting not going over the jump. I will go further into this in my next entry.

Sunday, December 6, 2009

What does dog training mean to me.

I am so lucky. I bought my first dog as an adult to have as a running friend. I was living on my own and running about 80 km per week and thought that having a dog along would be good fun and some company. Its what some guys did and I thought I could be a guy like that. But getting a dog left me with a lot of questions , how do I train it for starters? As a boy we had Doberman Pinschers and cattle dogs. These were trained with a choke/check chain and basically did what ever they wanted such as attacking local kids on bikes and eating cats. Here I was at 30 something years of age and getting in touch with feelings (freaky stuff). So I had no idea how to raise my new 8 week old puppy. Though I did have a stranger help me. I was out walking my puppy one night in Balmain and some young macramé wearing hippy feral guy walks up to me and ask about my puppy. He says that if you treat a puppy with love and gentleness you will have a friend for ever.

So when I was a kid I watched our dog trained with aversion techniques and attack people and bite them. My father was somehow proud of this vicarious machismo display, most of the time I was just embarrassed.

So I took this strangers words in to my heart and treated this dog with gentleness and love. I never hit, yelled or scolded.  I must confess now that this dog Scout was a very special dog. She was a cattle cross bull terrier. I bought from a pet shop in Newtown for $60. She was the most perfect dog I have ever owned in my life. She would do any thing for a ball and pat. I ended up with a dog of friendship and loyalty she was unlike any dog I had ever been associated with before in my life. She would never bite anyone. She would not bark at anything she was so sweet and friendly. My perception of dog ownership was changing.

Go foward a couple of years and I take my super fit dog to flyball to give her an outlet for all her energy. I turn up with all these crazy people and their dogs, they all seem quite nice (people and dogs). They talk about some weird stuff like staking my dog when I am setting up equipment. Mind you I had not taught  my dog a sit or a stay she just seemed to do what I wanted her to do.

So this flyball things seems like a bit of a fun, I meet some nice people and go off to a seminar. Run by a guy named Steve Pitt, he introduces me to a concept called operand conditioning positive reinforcement training an extension of classical conditioning. With these new found concepts I am dog to complete tasks/tricks like shake hand and roll over. I am excited by the possibilities. A light is switched on in my head and the fun begins...

Saturday, December 5, 2009

Flyball, what to do next?

I have been playing flyball for about 9 years. In that time I have trained 3 of my own dogs and helps countless other train their dogs as well. I have been lucky enough to run with some fantastic team mates and on the other side of the coin I have run with some that are less than fantastic. At times I think hardest part of playing flyball is being in a team with other people. Most other dog sports do not involve working together with other people it is generally a person and their dogs relationship.

I have my own personal theories about people that do dog sports and their anthropomorphic relationships with their dogs but lets not do to far down that track.
For the most part it is great to see people getting out and having fun with their dogs.

So back to me and my situation I have competed in flyball and agility with great success. In flyball, my main sport, I set out on a goal to capture the national championships (St George Phoenix). Whilst we were not awarded the actual title (because we did not run the fastest time) the official time sheets show that we won the most amount of heats against the eventual winners. This particular situation gave rise to a lot of debate within flyball in Australia leading to my putting forward a new standardised mothod for calculating competition winners in the event of a tie. This proposal was strongly supported (and accepted) by the members.

The confusion and controversy around those nationals was to me a real bummer for me. A situation like that takes a toll on the team, we knew we were not the fastest but we did know that we were a competitive team in division one. To loose on speed countbacks on several occasions was also becoming disheartening to me. My fun in flyball was waning at our team working so hard and loosing every time.

On a personal note my own life had taken a lot of hits I separated from my 14 year relationship (she is now my best friend). I was unemployed for 5 months due to Global Financial Crisis. I lost my house, I was diagnosed with autism (this was actually a relief). I was loosing my passion for flyball in the big picture

Then the World Dog Games were announced, this was an opportunity to show case the sport to a potential audience of millions. I asked a few people if they wanted to put a team together and compete. I was excited once again and built new enthusiasm. As time moved on though and as events unfolded it became obvious that a change was needed, thankfully I was not the only one thinking this way and it ended up I did not need to do anything.

I decided to take a respite from flyball for a while. I had started to try my hand at yard dog trials and was thinking of getting my own sheep dog, perhaps a another koolie from borahview. An announcement was made on an email list about a prospective dog border/staffy mix. I had been thinking of going to the USA and buying my own mix but a lack of funds had prevented this.

This mix (border/staffy) are proving to be a great build for flyball the current world record 14.963 secs is held by a team of four (view video here).

So a few emails were exchanged. I asked that few test be placed on the puppy (10 weeks) the results came back and suited what I was looking for. I now have a new dog. Ricco (named after Lavazza Gusto Ricco coffee It's a strong, intense coffee with a liquourish sweet taste).

I will try and journal my progress (or lack of) as I train him up for flyball, lets see where this journey takes us.