Aidan

When you go to a schedule of reinforcement and the performance suddenly drops off the first thing to look at is increasing the ratio of reinforcement. This should occur regardless of the pup's age (meaning, adolescence can have an influence but the reliable solution is still to increase the ratio of reinforcement). What you're seeing is known as "ratio strain" and it is a predictable, replicable phenomenon.

So when you faded the food, do you mean you faded it as lure or faded it as a reinforcer? e.g of fading it as a lure is to pretend it is in your hand for a while, but it's actually in your pocket or on a table. e.g of fading it as a reinforcer is that you stop giving it for every response. If you're not confident with your clicker skills, use a verbal bridge instead. Not usually a problem.

The reason he will only do what you ask when you have food in your hand is very simple, that is how he has been trained to respond. It works like this: Antecedent > Behaviour > Consequence Antecedent is anything in the environment that elicits a response, they become paired with the behaviour through repeated association. Behaviour is the thing you want him to do (or the thing that he actually does, so pay attention to that). The consequence is where the food is supposed to be, at the end, not the beginning. There are many reasons for using a clicker. One of which is to "bridge" the gap between Behaviour and Consequence. A reinforcer should be delivered immediately, but unless you have the food in your hand near the dog's mouth this is impossible so we employ a "bridge", commonly a clicker. This means we can mark the response (e.g dog sits) and then feed shortly after that. Using a clicker, we can have the food in a bowl on a nearby table, instead of in our hand or in a bumbag. If you've been teaching him with a food lure, you're going to have to "fade" the lure. Gradually pare it back until the Antecedent is no longer food in your hand, no longer a big hand signal, but a verbal cue or a hand signal if that's what you wish to use. If you see it for what it really is and stop thinking he's trying to overthrow you things will will be fun again and he will learn to respond to Antecedents other than food in your hand.

No doubt you've seen it but I've seen video of a dog cued to go to his bed (rather than rush to the door barking) when the doorbell was used. Reliability occured VERY quickly. Used sensibly, it's very potent. I think there is an article on clickertraining.com that shows how this is trained. My favourite technique with door-bell barkers is to teach them to bark briefly instead of continuously. All you do is click then toss a treat when they have barked for 1 second (or whatever you want, so long as you are consistent). Then click while they are eating, then toss another treat. Knock on the door again to repeat. Next time click for the same amount of barking, but click again for slightly more silence. Simple, effective, works from day one. I may be a bit thick, but I don't get this at all Aidan Are you trying to teach a disobedient partially trained dog not to bark at the door bell, aren't we jumping the gun a bit with door bell behaviour instead of concentrating on general obedience??? Forgive my confusion, but are we talking about the same thing? How could you possibly know that the dog lacked general obedience because he barked continuously at the door bell? My dogs bark at the door bell and both of them have well above average general obedience In any case, what has general obedience got to do with barking at the door bell? My grandmother's Shitzhu has appalling general obedience and she couldn't care less, it doesn't cause her any problems and the dog is easily groomed, seen to by the vet, walked etc If she came to me because her dog barked at the door bell (and she does bark at the door bell but my grandmother likes it, being elderly and living alone) I would just teach Bella not to bark at the door bell and everyone would be happy.

Sounds like a scaled up version of what caused the behaviour in the first place.

No. I would be suggesting a vet check though. I should disclose that I only see reactive and aggressive dogs and that it is very rare to actually need to fast a dog beyond skipping the previous meal. I was just saying that 48 hours is the most I've ever suggested. If you had a very reactive dog who could not tolerate an empty stomach it wouldn't be a major problem though, we'd find ways around that problem (two cases this last summer, health and diet issues).

I have suggested to some clients that they fast their dog for up to 48 hours prior to a session with me. Food gets the fastest results (most reinforcers in a given period of time) and once a dog has learned to eat in a new environment they generally don't need much by way of establishing operations after that point so it's usually a one-off. I do know that the majority of dogs (in my groups) who don't eat have a loss of appetite due to stress so everything I can practically manage to reduce that stress is also done. I've only had one dog who wouldn't take food but the owner did not fast her dog according to my instructions. Funnily enough she still made some (limited) progress, I put it down to habituation and possibly some work at home where the dog would eat.

Possibly, if the correction was completely meaningless but MM has reported an observed decrease in the nipping behaviour so we know that this is not happening. Behaviourists concern themselves with observable, measurable behaviours so you are quite right - it is a possibility - but it would be seen as an increase in nipping. I reinforce calm responses amidst aggressive responses regularly, with no corrections at all, and no dog that I have trained has ever increased their aggressive responses or even stayed at the same level.

Before or simultaneously, it doesn't change what it is. Even if the cue came first or there was never a verbal cue, it would still be -R. The important thing is that a stimulus is removed contingent upon the response which is what happens in the example given. The important part about applying the stimulus before the cue is to associate the cue with the response. You don't want collar pressure or hand on rump to be the cue. But that's a different issue. Generally if you're collar training the sit you apply pressure, cue 'sit', then remove the pressure when the dog sits. It works but it's not a whole lot of fun. It can get quite a quick response though, as a lot of -R procedures do.

Pretty much, as much as possible +R, extinction and some limited -P where it makes sense.

Standard obedience club method in the days I used to go to obedience clubs and still a common method when I wrote that article. Often combined with a firm push on the rump. Thankfully most have moved on these days.

http://ezinearticles.com/?Secrets-of-Dog-T...g&id=201252

A choking sensation occurs when something is tightened around your neck (or stuck in your windpipe), however briefly. The fact that it can happen with virtually any type of collar is irrelevant. Physiologically we find any sort of sensation that threatens biological function uncomfortable and have evolved a nervous system that causes us to immediately react to and/or avoid them. Does anyone think that I am arguing that they should be called "choke chains"? I would be disappointed if that were the case.

The only issue with the name was that people had suggested to the OP that perhaps if she didn't refer to them as a "choker" she would feel better about it. I suppose that is probably true but there are genuine and credible reasons to use aversion that don't rely on having to deny the fact that the collar at some points must produce an uncomfortable choking sensation.

I'm confused, who said there was anything wrong with calling them a correction or check chain? The only thing I am opposed to is denying that they cause a choking sensation. Talk about your dancing bears and being PC!

No doubt you've seen it but I've seen video of a dog cued to go to his bed (rather than rush to the door barking) when the doorbell was used. Reliability occured VERY quickly. Used sensibly, it's very potent. I think there is an article on clickertraining.com that shows how this is trained. My favourite technique with door-bell barkers is to teach them to bark briefly instead of continuously. All you do is click then toss a treat when they have barked for 1 second (or whatever you want, so long as you are consistent). Then click while they are eating, then toss another treat. Knock on the door again to repeat. Next time click for the same amount of barking, but click again for slightly more silence. Simple, effective, works from day one.

isnt that the long way round the short track? Why not simply apply an aversive so the dog not only gets no reward but sees there is a consequence to the action which further decreases any rewarding factor in the behavior, then show the alternative wanted behavior and highly reward that. My statement was in response to a specific example and was evidence that dogs undergoing extinction procedures do expect the reinforcer (which is why extinction only works for previously reinforced responses, and why it works better after continuous reinforcement). But yes, you can use punishers. Whether it would be any better or not would depend on the circumstances. The very broad statement regarding punishers, first noticed by Skinner and repeated over and over again, is that they are "unreliable". Reinforcing an alternative response improves the outcome but then you are no longer taking the long way around the short track regardless of whether you use a punisher or not. Differential Reinforcement of Alternative responses (DRA) gets very quick results and there is no risk of fall-out. Not really. Once you put it under extinction the classical conditioning is dissociated very quickly. Good question though. Yes, DRA is usually preferable. But please don't make the mistake of thinking that the dog needs to be a great problem solver for this to be effective, none of the data suggests that this is required and cognition isn't even required for a lot of this learning. It can even happen subconsciously with no learner awareness whatsoever - and does, to HUMANS, every day! Horses are relatively poor problem solvers and they have no trouble learning through these mechanisms. I would consider something like that to have an element of self reward in some dogs. A bit of food or a toy is not always enough to overthrow what the dog sees as rewarding. So if rewards are difficult do we then start deprivation in order to get our goal? We need to establish operations one way or another. Deprivation is sometimes used. Habituation is another possible mechanism. Extinction might be employed. There are many ways to skin that cat. If I can make one thing clear the reinforcer used need not be "greater" than the reinforcer freely available in the environment. There is boatloads of data to support my assertions on this.

This is actually something I've been thinking about a lot recently. I'm wondering if all these things we call rewards - prey, food, attention, praise, sex - are way less important to dogs than the one, overarching "meta reward" of being able to control the environment (and therefore predict and control access to the good stuff, and predict and prevent the bad stuff). That's an interesting thought and I think you are on to something. On the topic of what you have called "meta rewards" you might also be interested in Panksepp's concept of "SEEKer circuits" and the possibly related concept of "contra-freeloading".

Failure to do something isn't a response, but whatever the dog was doing at the time is a response and that is what you are likely to see extinguish (probably very quickly in the example given). It's not a good example because it would be very difficult to discontinue all reinforcement for this response. If you reinforce an alternative response you can get that alternative response very quickly. The interesting thing about that example is that if you discontinue reinforcement for the alternative response, the previous behaviour often comes back very quickly. The same is true when a punisher is used.

LOL, yup!

So, I have the treat in my hand, dog expects treat. I lure the dog into a sit but it doesn't. So no treat. Are you saying the dog doesn't interpret that as something being taken away? No, he didn't have the treat so it wasn't taken away. Don't forget, behaviourists make observations and do not make assumptions about things like what a dog might "interpret" (not that it would make any difference in this case). In an extinction procedure the reinforcer is likely always expected (as it is in your example). Dog expects treat, doesn't get it, response goes into extinction, which typically follows a curve and is subject to well-studied phenomena associated with extinction. One of the fastest ways to extinguish a response is to reinforce it continuously first so that the dog expects a reinforcer and figures out quickly that something has changed when there isn't one. If the addition of the halti decreases the behaviour, I'm a bit stumped as to how it can't be +P. Consistent aversive effect aside,surely the contingent +P is that the dog feels pressure when it pulls. Pulling = pressure on the face AND potentially failure to progress. Dog pulls, leash tightens (because I allow it to) and dog's progress is thwarted. I'm doing this, not the dog. I didn't say there wasn't ever going to be +P involved. There is a certain amount of escape and a certain amount of avoidance, or a certain amount of extinction that goes into the learning process. Then the dog is conditioned. It's sometimes analogous to a prong collar. Ideally with a prong collar (or head halter) the dog learns quickly - pulling results in discomfort. So he no longer pulls. He is in control of that consequence. It is not a constant aversive. This is true for either tool. The problem comes when distractions are not introduced correctly, or if the owner accidentally tightens the leash or attempts to use poorly timed corrections. If the dog doesn't feel that he can control the consequence through his behaviour then the result is stress, you get more fall-out, you get slower learning, you get displacement behaviours etc Most people are oblivious to it. Very easily managed with a double-ended leash though. The dogs who find the head halter uncomfortable at all times can usually be conditioned to accept that something is on their face, although some do appear to be resistant to this. I think the vast majority who do not are those who have not been given reason to believe that they can control the consequences through their behaviour. The example above is just one way a head halter might work. I have seen dogs who keep attempting to pull, they clearly don't find pulling into the head halter aversive. They just don't get where they want to go so the response extinguishes, often fairly quickly. It's the same as pulling into a flat collar, only with added leverage. We should not assume that the head halter is always aversive.

Don't bother, skateboards operate outside time, space and conditioning. Arrr a sense of humor! Very important in dog training! You don't know how serious I am about this.

Settle down, there is no place for this sort of reasoned, logical thinking in a thread about NDTF vs DELTA!!!! Now if you'll excuse me, I'm off to watch "Aikido vs Ju Jitsu" and "Judo vs Kenpo" videos on YouTube. I reckon the Aikido master will win because he's non-violent and just uses the Ju Jitsu guy's violence against him...

Don't bother, skateboards operate outside time, space and conditioning.

Withholding a reinforcer is "extinction" and does not equate to punishment, either by definition or by function. -P requires that something (not necessarily a reinforcer) is taken away ('-' = take away or remove, which means that the organism had to have had the thing in the first place). I think it is incomplete to say that you "only use +R", or possibly naive, but nearly every purely positive trainer will openly state that they use negative punishment and extinction, and most will admit to using -R these days. Karen Pryor has always admitted to using +P, I think people on both sides of the camp have a funny idea about what anyone else does. Possibly, but preventing self-reinforcement does not equate to +P, nor does removing an aversive contingent upon a response. Adding something to the dog is not the same as adding a consequence for a behaviour. I think it pays to be accurate, +P should be absolutely minimal if using a Halti correctly. I could be wrong, I know that Delta asks their graduates not to use certain tools, but I'm not sure where they say that they are "purely positive"? They have recommendations on tools and methods and have a policy of minimising harm. A head halter used appropriately should minimise harm, "appropriate use" may be up for debate though. I don't think Delta are making unsubstantiated claims regarding their published policies, but I'm not expert on Delta so feel free to correct me (just not with a prong collar!)