'Not only is punishment risky, but it also fails to teach the dog an acceptable alternate behaviour. The dog does not learn what to do the next time he is in the same situation. He only learns to fear the situation.' Emma Parsons, Click to Calm p73

Sunday, 17 April 2011

Reinforcing Cues.

I've been reading Crystal's thought-provoking posts on Kathy Sdao's seminars at ClickerExpo, particularly in regards to why and how a cue may or may not be reinforcing to a dog. I think that there is a scale, of between one and ten, as to how reinforcing a cue is, and the position of each cue on this scale could be affected by a number of factors.

I would suggest two categories of reinforcing cues: naturally reinforcing and artificially reinforcing.
Naturally reinforcing cues (NRCs) would be cues given to behaviours a dog naturally does and enjoys - this could be permission to go and see a strange dog or person, sniffing, playing with a beloved toy, running, etc. Naming these behaviours would be helpful in following the Premack Principle; using a high probability behaviour to reinforce a low probability behaviour. Most NRCs would likely be taught through capturing, rather than luring or shaping, though they aren't exclusive to captured behaviours (the handler could easily shape a 'say hello' cue with a shy dog, for instance).

Artificially reinforcing cues (ARCs) are everything else. In order for a cue to be reinforcing, it must adhere to enough conditions that the dog genuinely enjoys performing it - not simply for the opportunity to get a reinforcer (food, a game, an environmental reward, etc) but for the opportunity to merely perform that behaviour.

I think there are a number of reasons, however, for why a cue may or may not be reinforcing:

- Inappropriate Motivation:
1: Something that might motivate the dog in one situation may not in another (e.g. kibble used in the house versus in a park full of strange dogs).
2: The rate of reinforcement may be too low, whether in regards to what the dog usually expects, or in regards to how 'costly' the behaviour is.
3: The dog may have been put on an intermittent schedule of reinforcement too quickly, or the schedule may be too low for the cost of the behaviour (e.g. the dog only getting a treat once every fifteen or so reps, whilst they still require a reinforcer every five or six times).
4. The handler may be attempting to reinforce the dog with something that they are indifferent to (e.g. a bland food rather than the more stinky, tasty reinforcers they are usually used to), or which is actually aversive for them (e.g. petting the dog on the head - few dgs enjoy that).

- Fear or Dislike of Consequences:
1. The dog may have been taught through positive punishment or negative reinforcement (e.g. pinching a dog's ear to train a forced retrieve will very likely not result in an ARC), which may result in the dog being fearful of the behaviour or of experimenting outside of the behaviours it knows strongly.
2. There may have been an environmental event following the giving of the cue that would result in a lack of enjoyment for it (e.g. the dog being asked to roll over on a slippery surface, and banging down too hard on its side). If the event happened frequently, this could effect the dog's overall enjoyment of the cue.
3. There may have been a one-time environmental consequence to the cue that has resulted in fear in the dog (e.g. a sound-sensitive dog first offered a 'spin' whilst being taught it, and a car backfired at the same time or immediately afterwards).
4. The dog may have been taught a behaviour through physical manipulation of their body, and disliked the coercive aspect of it, which lead to a weakening of the cue as a reinforcer (e.g. modelling a dog's sitting body up into a stand).
5. The dog may have grown bored of repeatedly being asked for the same behaviour, or was too distracted, or had grown bored of the reinforcer being used. If this occured frequently enough in training sessions, the cues that were being asked for at the time may be weakened in turn. The dog may also have been reprimanded for 'blowing off' the trainer (e.g. when showing stress signals when being cued to 'say hello' to a stranger), which would again result in a weakening cue.

- Genetic or Health-related Influences:
1. The dog may have a genetic influence that prevents their body from moving naturally in a certain way. Teaching this dog that behaviour would then likely result in a cue that is not a reinforcer (e.g. asking a greyhound to sit).
2. The dog may have a health influence which, again, prevents their body or mind from certain behaviours. If it was a condition that they gained later in life, being asked to do reps of a cue that was once a reinforcer may reduce its reinforcing status (e.g. asking a dog with hip dysplasia to sit perfectly straight, or to jump up and down).

- Personal Preferences:
1. The dog may, quite simply, have a personal dislike for a behaviour, either generally or situationally (e.g. a dog that is naturally very excited may dislike being cued to 'stay'). This could also change over time, as the dog ages and matures - for instance a dog may have enjoyed something greatly as a puppy, but may not when they are elderly.

- and Other:
1. The dog may have been asked to give the behaviour so much that it has become background noise; the dog appears to be listening and 'obeying', but is simply going through the motions (e.g. if the dog is asked to sit every time--and only every time--the handler stops mid-heel, the reinforcing nature of 'sit' may deteriorate as the dog simply pre-empts the trainer and performs the behaviour itself).
2. Alternatively to above, the dog may have been asked to give the behaviour so sparsely that it is no longer reinforcing. If the dog was taught a behaviour over a couple of days (even if there was a hugely high rate of reinforcement for it) and then not asked again for it for several years, the cue may have been all but forgotten - it is therefore no longer a reinforcer.


Crystal Thompson said...

I love your categories of NRCs and ARCs. I was trying to get to that, but the clever little acronyms are so awesome! Really helps drive the point home. :)

I also love the way you've structured a lot of the ideas around why cues may not be reinforcing into neat little categories. I especially love the one on personal preferences- I got entirely too sterile and scientific about my thoughts, and failed to consider the dog beyond its role as an input/output machine in regards to conditioning. Your thoughts under "Other" are great too. I've definitely seen both things happen, especially that second one.

andrea said...

I can honestly say I had never thought of the cue itself being a reinforcer ... the behaviour is a reinforcer sometimes (for example Sally will weave all by herself with no one anywhere near her) and lots of things are non traditional reinforcers - smiling, clapping are two I use but the cue itself .. I will have to think about this lots I think

Ninso said...

Clicked over from Crystals facebook page. It seems to me that these are all reasons that might affect the enthusiasm level of a dog's response to a cue. So it seems to me that rather than saying that a cue can be reinforcing, we could say more accurately that a cue that the dog LIKES performing can be reinforcing. Seems much simpler that way. I know which cues my dogs enjoy, so this way, if I needed to reinforce and didn't have a treat/toy, I can give my dog a favorite cue. I do think strength of the reinforcement value needs to be considered as well. If my dog comes when called off of a very interesting scent, I doubt the dog will find "roll over" to be much of a reward, no matter how much he likes rolling over in other situations.

Cool post and cute dogs!

Sophie said...

Crystal, thanks for popping over! :) I don't think I've covered everything, but it's all I can think of at the moment. I do think that the dog's personal likes and dislikes is a major component; even if you do everything else "right", I think there's still a chance that the dog might not find the cue reinforcing if it is something they don't personally enjoy.

Andrea, thank you for commenting. I agree that the behaviour can be reinforcing, but I think the cue can definitely be reinforcing too; it's the opportunity to do something that the dog really enjoys. I also use non-traditional reinforcers - my puppy genuinely seems to enjoy it when I start doubling up with laughter at her.

Ninso, I'm glad you stopped by! I agree with your simpler way of saying it; I think these are just reasons why a dog may like or may not like a cue, which in itself would lead to the cue becoming a reinforcer (in my opinion). The motivation is definitely situational as well as general - like with your example, my JRT puppy loves to jump in my arms in the house, but outside scents are still quite a lot more interesting. And thank you!

Anonymous said...

Coming over here from Crystal's blog.

Absolutely love the concept of natural reinforcers! Yes - that is what I have felt for a long time without being able to express. My hound is almost solely impressionable by natural reinforcers whereas my other two dogs (a Pom and a Portuguese Water Dog) will work happily and repetitively for artificial reinforcers.


Sophie said...

Beachcomber, I'm glad my awkwardly-explained concepts have helped :) My older dog, Jess (a Beagle crossed with a Staffordshire Bull Terrier, but all scent-hound when we're outside) will only work outside for things she naturally finds reinforcing. If I'd thought when she was younger to put cues to those behaviours (following a scent, greeting a person, etc) I might have been able to make her listen more. As it is, she's pretty hard to motivate outside!