1. AGPC ™ Home
  2. Parrot Articles
  3. African Grey Forum
  4. Parrots For Sale
  5. Parrot Rescue Centre
  6. AGPC ™ Blog
  7. Parrot Shop
  8. Contact Us

Training With Positive Reinforcement Explained (Sort Of)

10% Discount in The Online Shop
  • December 17, 2017, 09:51:44 AM
  • Welcome, Guest
Please login or register.

Login with username, password and session length
Advanced search  

News:

Pages: [1]   Go Down

Author Topic: Training With Positive Reinforcement Explained (Sort Of)  (Read 2984 times)

0 Members and 1 Guest are viewing this topic.

lovebirds

  • U18
  • * Head Pecker *
  • *
  • Reputation: 3
  • Offline Offline
  • Posts: 15624
    • View Profile
Training With Positive Reinforcement Explained (Sort Of)
« on: May 16, 2011, 11:43:32 AM »

A few months ago I've done a parrot behaviour course and have learnt a great deal from it so would recommend it to anyone. It is free, but a donation of $50 to a chosen by the organisers charity is desirable. Here you can read more about it http://www.behaviorworks.org/

I thought it might be useful if i put a short summary of what I have learnt (obviously I can't just copy it but we were encouraged to "share our knowledge") and perhaps Karmen could add some more terms and explanations, or more information to it, to give us all a better understanding of our parrots behaviour and how to manage it humanly and effectively.

I will try to put it briefly and reasonably simply but it does take time to understand, but if you have time it might be worth trying and understand the gist of it and if you have any questions please ask  :biggrin:

First of all, we were told to throw "labels" out of the window when we encounter a problem. For example, we commonly say the bird is aggressive because it is "hormonal". "Hormonal" is a label which we often hang on an animal, and that prevents us from looking deep and objectively at what the problem really is and how we can rectify it.

Every behaviour serves a purpose and to be able to modify it we need to find out what the purpose is before trying to modify it. In other words every behaviour has a function

Additionally, every behaviour has an antecedent and consequence

Antecedents -  are events or conditions that precede the behaviour and predict its occurence. Antecedents can be distant and immediate.

Distant antecedents are the ones which don't happen immediately before the behaviour and are more generalised ones, like gender, species, eating routing, daily schedule, etc.

Immediate antecedents are those which happen right before the exhibited behaviour. So if this antecedent happens then we know it will be followed by a certain behaviour straight away. It's the antecedent behaviour which signals to the animal the behaviour-reinforcement contingency. For example - you tap a perch on the playstand, the bird flies over to the perch, you give her a treat. The immediate antecedent here is you tapping on the perch and it lets the bird know that the treat will follow if he flies over to that perch now.

Another important term is the consequence . Consequences are those events or conditions which occur straight after the behaviour has been exhibited. The consequences let the parrot know how to behave in the future. The consequence can intensify, weaken or maintain a behaviour.

This unit of behaviour is called ABC (antecedent, behaviour, consequence). Almost every behaviour can be analysed using ABC.
To modify the behaviour we can either change antecedent or consequence or both.
Using this formula we will be changing the environment and not the animal.

Here is an example of how it works:

Antecedent: Nicholas comes to sit on the sofa
Behaviour: Digby bites Nicholas
Consequence: Nicholas gets up and moves to another sofa

There are a few different ways of how we can change Digby's behaviour - we can either change the antecedent to prevent the behaviour from happening, or the consequence to alter the behaviour.

So moving on, we come to shaping up the behaviours. The best, most humane and effective way to shape a behaviour is by using Positive Reinforcement.

Positive reinforcement doesn't mean necessarily training by using rewards. Both "positive" and "negative" reinforcement will maintain or increase the desired behaviour. But the difference is that with positive reinforcement the consequence that follows will add something to the environment. So positive reinforcer is something the learner wants and tries to get  . Negative reinforcement is something the learner wants to escape and avoid.

A treat, for example,  is going to be a positive reinforcement as we know. But if you look at my example with Digby earlier, Nicholas going away from the sofa after Digby has bitten him is a positive reinforcement too - Digby got what he wanted, and that positive reinforcer will maintain or increase Digby's biting behaviour in this situation.

Now, reinforcers can be primary or secondary.

Primary reinforcers are associated with physiological necessities, like food, water, etc.

Secondary reinforcer (or conditioned reinforcer) - is a previous neutral action/thing which has acquired its reinforcing value by being paired up with primary reinforcer or with another existing secondary reinforcer. The best example of it is a cliker - the sound of a clicker becomes a secondary reinforcer by being paired up with the primary reinforcer (a treat/food).

When conditioning new reinforcers or training using the reinforcers there are a few things which will make it effective:

- the reinforcers must be delivered immediately after the desired behaviour. Even a second of delay will reduce the effectiveness by loads and loads.
- the reinforcers must be delivered consistently after the desired behaviour
- the intensity of the reinforcers is of the utmost importance (for example using the most favourite food for training will be a lot more effective than using a second favourite food)
- the size of reinforcer can help to establish the behaviour quicker if using sparingly and appropriately
- availability of the reinforcer must be limited outside of the training situation to make it most effective during training. (for example, if using sunflower seeds as a reinforcer don't give any during the rest of the day).

So if for example you want to teach your bird to wave, pick the reinforcer your bird will react strongly too, let's say sunflower seed. Make it unavailable anywhere anytime else. Break a few apart but keep a small amount of seeds whole. This will be your "bonus" reinforcers for doing the job super well done, rather than just well done. Now, here we will have to use so-called "Shaping" - that is shaping the final behaviour (a wave) by reinforcing small approximations of that behaviour first. For example "a wave" can be broken down into the following approximations:

- a foot is slightly raised off the perch
- a foot is raised off the perch by 1cm
- a foot is raised off the perch by 2 cm
- a foot is raised off the perch by 3 cm
- a foot is raised off the perch and slightly turned up
- a foot is raised off the perch, and turned up in "a wave"

When training you will start reinforcement from the first approximations progressing further as the next behaviour step occurs.
It is important to reinforce the next step as soon as it occurs and drop the reinforcement of the previous step to make sure the behaviour progresses and doesn't just stop on the first approximation. You can use your "bonus" reinforcers (in this case full size sunflower seeds) to reinforce the first occurrence of the next step in the approximations for example to make it more effective.

this way you can shape up almost any behaviour/trick you want. But as always a lot of patience, consistency and a good eye required.


Modifying unwanted behaviours

Coming back to labels and analysing behaviours - it's important to understand the purpose of the behaviour exhibited instead of just labelling it with something generic. It's important because once you know why the behaviour happens you can offer an alternative without negative effect to your bird. As we know each behaviour serves a function, if we know what the function is we can offer the bird an alternative. This alternative behaviour should give the bird the same or larger amount of reinforcement for him to choose to exhibit the behaviour we want, rather than that we don't want.
For example, screaming is very natural for parrots but not very nice for human ears. By offering and reinforcing alternative way of vocalising, for example whistling or talking we will solve the problem of neighbours complaining about the noise and the bird will still be happy. Let's say your parrot starts screaming when he has to sit in the cage while you leave the room. While vocalising he might wedge in a word or a whistle in his vocalisation. That is when you need to reinforce the desired sound by either quickly returning to the room (you have to be swift to return exactly when the desired sound was heard and not during an undesirable one) or replying in the same manner (by whistling, talking etc.).
At the same time you can condition the bird to perform an incompatible behaviour (that incompatible with the undesirable behaviour). For example, the bird can't play with the toy and scream at the same time (well, it can if it really tries, but not as much). So whiling teaching an alternative behaviour you can also reinforce each instance of your bird playing with his toys while he is in the cage by praising, providing treats and using any other positive reinforcers.
This sort of strategy should work every time if performed correctly because an animal will always choose that behaviour which is reinforced more. (if screaming is ignored by whistling is reinforced then the parrot will prefer to whistle - unless you ignoring him in that instance is a reinforcer in itself  :haha: ).

So if I apply all said above to the situation with Digby my plan will be as follows:

I will have to offer here an alternative and incompatible with the naughtiness :biggrin: behaviour and provide more reinforcement for it than he gets from biting nicholas. The incompatible behaviour in my case will be him staying on his playstand while Nicholas is sitting on the sofa. I will start with one minute and increase it gradually of him staying on the playstand, while reinforcing it every let's say 10 seconds to start with by providing a treat. After one minute Nicholas will get off the sofa before Digby even has a chance to get on the sofa. Digby's staying on the playstand will be incompatible with him being at the same time on the sofa, obviously and it will be well reinforced. After some time, when we have clocked up enough minutes and he is comfortable with Nicholas's presence on the sofa I will stop treating frequently but will just give a reinforcer intermittently to maintain the behaviour.



















Logged
Make sure you click me every day and tell the world how we are the best parrot site on the internet!!

bkhowe

  • Guest
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #1 on: May 16, 2011, 12:15:11 PM »

Cheers Irina, that all makes perfect sense  :thumbsup:
How long did the course last?, I'm doing it next year, although not much point if you are posting it all on here  :biggrin:
Logged

lovebirds

  • U18
  • * Head Pecker *
  • *
  • Reputation: 3
  • Offline Offline
  • Posts: 15624
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #2 on: May 16, 2011, 12:29:43 PM »

oh no, you will get a lot more information, with lots of examples and will work with the teachers on your own, so it is definitely worth taking, Bob, those are just a few basic principles  :thumbsup:
The course is made up of 7 lectures and 6 homeworks, (1 lecture per week) plus another week for a final exam
Logged
Make sure you click me every day and tell the world how we are the best parrot site on the internet!!

Pat

  • I have a nut allergy, I can't tolerate idiots.
  • Regular Pro
  • * Head Pecker *
  • *****
  • Reputation: 90
  • Offline Offline
  • Gender: Female
  • Posts: 14131
  • Hatch Date: 11th June 2009, 12th Dec 2004
  • Parrots Owned: Charlie and Penny
  • Yoohoo!! Tis Me!!
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #3 on: May 16, 2011, 12:31:12 PM »

That is really interesting, thanks for sharing.  I can see I'm on the right track with Charlie lol.  I'd like to have a look at that course though, do you have a link for it please?

Karmen

  • *Queen Of Africa*
  • Jabbajaws
  • *****
  • Reputation: 7
  • Offline Offline
  • Gender: Female
  • Posts: 1569
  • Parrots Owned: Buku (CAG), Bobby and Snow Flake
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #4 on: May 16, 2011, 12:34:25 PM »

I know Susan Friedman - her approach is great  :thumbsup:
Logged
Karmen


Certified Animal Behaviour Consultant (Avian, Canine)

lovebirds

  • U18
  • * Head Pecker *
  • *
  • Reputation: 3
  • Offline Offline
  • Posts: 15624
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #5 on: May 16, 2011, 12:37:02 PM »

you certainly are on the right track, Pat  :thumbsup:
Pat, you can see the description of the course here http://www.behaviorworks.org/htm/comp_professional_overview.html

but to see the actual lectures you have to join a closed yahoo group and the lectures are posted there one by one. The group is only open while the course lasts, so i am afraid those not on the list for taking the course can't join it, but there is a group you can join and is open for everyone also run by the same people and it does contain some information in the files http://groups.yahoo.com/group/ParrotBAS/join
Logged
Make sure you click me every day and tell the world how we are the best parrot site on the internet!!

Karmen

  • *Queen Of Africa*
  • Jabbajaws
  • *****
  • Reputation: 7
  • Offline Offline
  • Gender: Female
  • Posts: 1569
  • Parrots Owned: Buku (CAG), Bobby and Snow Flake
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #6 on: May 16, 2011, 12:43:29 PM »

Pat, Susan Friedman also offers a professional course, which can be done by pet owners as well, you don't have to be a vet to do it. You can get more info about it here: http://www.behaviorworks.org/htm/lla_professional_overview.html

Perhaps you should drop her an email and she will explain to you what your options are, costs etc. sg.friedman [at] usu.edu
Logged
Karmen


Certified Animal Behaviour Consultant (Avian, Canine)

lovebirds

  • U18
  • * Head Pecker *
  • *
  • Reputation: 3
  • Offline Offline
  • Posts: 15624
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #7 on: May 16, 2011, 12:58:30 PM »

the professional course involves communicating via a video camera I think so I didnt want to do that  :rofl: plus it costs $700 for 8 weeks  :biggrin:
Logged
Make sure you click me every day and tell the world how we are the best parrot site on the internet!!

Karmen

  • *Queen Of Africa*
  • Jabbajaws
  • *****
  • Reputation: 7
  • Offline Offline
  • Gender: Female
  • Posts: 1569
  • Parrots Owned: Buku (CAG), Bobby and Snow Flake
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #8 on: May 16, 2011, 01:02:47 PM »

Oooo... no.... video camera? Would also not feeling comfy with that  :rofl:
Logged
Karmen


Certified Animal Behaviour Consultant (Avian, Canine)

Mrs Piker

  • * Crazy Horse *
  • Jabbajaws
  • *****
  • Reputation: 90
  • Offline Offline
  • Gender: Female
  • Posts: 9306
  • Hatch Date: 07/07/2009 & 18/09/2010 & 27/01/2010 & 15/06/2004
  • Parrots Owned: Mr Frodo Baggins, Mr Bilbo Baggins, Miss Rosie Cotton & Miss Charlee Chica Baggins (The Hobbits)
  • 333 cus I'm only half evil
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #9 on: May 16, 2011, 01:31:01 PM »

That's some brilliant information Irina  :thumbsup:
At this rate I'll never be able to come off this site, so much reading but very valuable info on here.  :thumbsup:

Julie
Logged
Julie & The Hobbits, Frodo, Bilbo, Rosie and Charlee Chica

Ann Summers is to women, what B&Q is to men... D.I.Y.

lizduncan

  • Guest
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #10 on: May 16, 2011, 10:22:56 PM »

Great info Irina. It all makes sense. Glad to hear the course is good.

Good luck Bob for next year
Logged

Lesley

  • *Brave Heart*
  • Jabbajaws
  • *****
  • Reputation: 14
  • Offline Offline
  • Gender: Female
  • Posts: 4416
  • Hatch Date: 1996
  • Parrots Owned: Flint: begged, borrowed or stolen for the moment....
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #11 on: May 17, 2011, 07:01:21 AM »

Thanks for posting that Irina.

Its really helpful. Generally fits with what I try to do but it good for understanding why I'm doing it  :biggrin:

You sofa biting example is particularly relevant for me at the moment!
Bring on the play stand!
Logged
Lesley and Flint x

Pat

  • I have a nut allergy, I can't tolerate idiots.
  • Regular Pro
  • * Head Pecker *
  • *****
  • Reputation: 90
  • Offline Offline
  • Gender: Female
  • Posts: 14131
  • Hatch Date: 11th June 2009, 12th Dec 2004
  • Parrots Owned: Charlie and Penny
  • Yoohoo!! Tis Me!!
    • View Profile
Re: Training With Positive Reinforcement Explained (Sort Of)
« Reply #12 on: May 17, 2011, 12:24:54 PM »

Thanks very much Irina and Karmen for the info, will take a look x
Pages: [1]   Go Up