Adduction Training

[lizzybeth727]

Now that first bolded part sounds to me more like a couple of behaviors that she is doing one after the other. It does not sound like a unique behavior which arose from two different behaviors, as in the description of adduction. You've cued "go" and "jump." Each behavior has it's own cue.

"Go" means run straight ahead. "Jump" means take the jump I'm indicating; usually for jump I'm pointing at the jump or somehow gesturing toward it. But "go jump" means run straight ahead and take the jump you come to, even if I'm running off in some other direction.

The part that makes this an example of adduction is when the cue is given. I cued her even before she could see the jump in front of her. While she was in the tunnel she heard my cue, and knew that when she came out, she'd have to run straight ahead and then take the jump.... she knew that even before she could see the jump in front of her.

(Of course I don't like to infer about what animals are thinking, because we will never know what they're thinking and believing that we do know what they're thinking generally just causes sloppy training.)

In the 2nd bolded part, what were the other small items on the floor? But how did she know which small item in particular to pick up? Was it the only toy? Or the only thing that smelled like an animal toy?

She picks up anything that she can physically pick up... retrieve is definately well-generalized. But in this example, she picked up the item that was directly in front of her.... but she had to walk a distance to get to it and I didn't prompt her at all besides giving the cue.

 

[release the hounds]

what info do you want? that as you teach dogs behaviors and put them together you get new behaviors? or that as you put things together, dogs tend to more easily put future complex behaviors together quicker as they have success in different previous behaviors?

IMO, if you've been training dogs to do any more than the basics, you've already been doing this. I'm not here to rewrite a book on dog training, i'm still waiting to see what makes this novel or new other than using the word adduction?

 

[lizzybeth727]

what info do you want? that as you teach dogs behaviors and put them together you get new behaviors? or that as you put things together, dogs tend to more easily put future complex behaviors together quicker as they have success in different previous behaviors?

The idea is that you teach the dog the concept - do this behavior, then do this other behavior - and when he understands the concept you can simply cue "behavior THEN other behavior" and he will understand to do one known behavior and then do the other known behavior. Once he understands that concept, you will not have to teach chains or sequences, you simply tell the dog the whole chain from the beginning and then he'll go do it.

Like if I tell you, "Stand up," (you stand up), "walk to the ringing phone" (you walk to the phone), "pick up the phone" (you pick up the phone) "say hello" (you say hello). That's cueing individual behaviors in a sequence. If I were to train that behavior, I'd teach each individual behavior and then teach the sequences (for example: teach you to say hello; teach you to pick up a phone and say hello; teach you to walk to the phone, pick it up, and say hello; and finally from a sitting position, teach you to stand up, walk to the phone, pick it up, and say hello). Then once you understood the sequence I'd put a new cue on it - "answer the phone" - which means to do the whole chain.

With adduction theory, though, I wouldn't have to teach the sequences. I'd teach each individual behavior (stand up, walk to the phone, etc.) individually and separate from each other. Then if I wanted the whole sequence, I'd simply say, "Stand up, THEN walk to the phone, THEN pick it up, THEN say hello, now go" and you'd listen to the whole sequence and do it when I said "go."

Then you can do all kinds of combinations and orders of behaviors, like mix and match. And you wouldn't have to teach the chain at all, because the dog understands the concept. It's not that the chains become easier to train, or the dog learns quicker; it's that you really don't have to train the chains at all, the dog will be able to do new chains on the first try.

 

[marfak9]

Sorry, it's not Lindsay's Volume III, it's Volume I, Chapter 7, "Instrumental Learning". Adduction isn't new.

 

[corgipower]

With adduction theory, though, I wouldn't have to teach the sequences. I'd teach each individual behavior (stand up, walk to the phone, etc.) individually and separate from each other. Then if I wanted the whole sequence, I'd simply say, "Stand up, THEN walk to the phone, THEN pick it up, THEN say hello, now go" and you'd listen to the whole sequence and do it when I said "go."

But as far as I've been able to tell, that's not adduction training.

Adduction training is telling a dog to "down" and to "heel" simultaneously in order to teach him to crawl.

 

[lizzybeth727]

But as far as I've been able to tell, that's not adduction training.

Adduction training is telling a dog to "down" and to "heel" simultaneously in order to teach him to crawl.

The way Ken explained it this weekend, there are two different types... Teaching the animal to do one behavior followed by another, or teaching the animal to do the two different behaviors at the same time. They're clearly different concepts, but he said they were both adduction training.

 

[lizzybeth727]

Adduction isn't new.

Great! Do you have experience with it, then?

 

[Promethean]

This trainer sounds like they are using quackery language to sell something that everyone else calls chaining.

The reason you can't find any info on it is because he just made it up and there is no such thing.

Munch, munch, munch..... (takes a swig of a nice chianti)

One thing, it's not on the 3rd volume it's on the 1st and no wonder I didn't remember, the whole section is only 2/3rds of a column.

This is what it reads.

Adduction refers to a training procedure in which a novel response is produced by combining two or more previously learned component repertoires. More specifically, adduction occurs when two previously learned behaviors are evoked by presenting their respective discriminative stimuli together. The resulting adducted response is reinforced and subsequently brought under stimulus control. For example, although training a dog to crawl can be accomplished by gradually shaping the crawling behavior through successive approximations or by utilizing appropriate physical prompts or props (e.g., a low table), the crawling behavior can also be obtained by signaling the evocation of two previously learned behaviors whose combined emission results in crawling. In this case, the dog might first be taught to heel and to lay down on signal. Once both behaviors are under stimulus control, the dog is signaled to lay down while heeling. In response to this arrangement, the dog may start to lay down but continue moving forward at the same time and, perhaps, begin to crawl. As a result, crawling behavior has been adducted from the combined occurrence of laying down and heeling. If the crawling response is reinforced under such circumstances, gradually it can be brought under stimulus control and then trained to occur independently of the antecedent component repertoires

 

[marfak9]

(Yes, Prom, I corrected that in an earlier post. Thanks, though.)

Re, the "quackery" comment. Adduction is a term used in applied behavior analysis and used in advanced "clicker training".

Yes, I’m familiar with it. I have a dog that decided to “saluteâ€; a “sit†combined with a “high fiveâ€. My dog’s high five is standing on her hind legs so she can reach hands, so this was a new behavior combination. I reinforced it and put it on cue, “saluteâ€. But, I’ve never set out to train something using adduction that I can think of.

 

[Dekka]

If you have ever done agility it shouldn't be new. Maybe not named. But teach your dog jump and turn (for example) then you cue your dog to jump and turn at the same time so you can rear cross behind the jump...

Or teach your dog to target, and to assume the 2o/2o position then combine them on the obstacle for a new behaviour.

 

[lizzybeth727]

This is what it reads.

Adduction refers to a training procedure in which a novel response is produced by combining two or more previously learned component repertoires. More specifically, adduction occurs when two previously learned behaviors are evoked by presenting their respective discriminative stimuli together. The resulting adducted response is reinforced and subsequently brought under stimulus control. For example, although training a dog to crawl can be accomplished by gradually shaping the crawling behavior through successive approximations or by utilizing appropriate physical prompts or props (e.g., a low table), the crawling behavior can also be obtained by signaling the evocation of two previously learned behaviors whose combined emission results in crawling. In this case, the dog might first be taught to heel and to lay down on signal. Once both behaviors are under stimulus control, the dog is signaled to lay down while heeling. In response to this arrangement, the dog may start to lay down but continue moving forward at the same time and, perhaps, begin to crawl. As a result, crawling behavior has been adducted from the combined occurrence of laying down and heeling. If the crawling response is reinforced under such circumstances, gradually it can be brought under stimulus control and then trained to occur independently of the antecedent component repertoires

Thanks for the info! Now I guess I don't have to spend $120 on a three-volume set of books.

 

[marfak9]

it's $120 well spent. If you can find the books at a reduced price, even better. If you can't afford them all, just get Volume III.

 

[release the hounds]

120? that's cheap. When I got them they were 70-80 bucks a piece, more like 240

 

[cvarnon]

Alright trainers, I need some info!

I was at a seminar this weekend and heard a very brief description of "adduction theory," combining learned cues to create new behaviors. For example, you can teach your dog "roll over," and teach him "speak," and then use adduction training to teach "roll over while speaking" or "roll over, then speak." The idea is that you say your "roll over" cue, then your "speak" cue, then some other cue that means "put 'em together," then the dog does it.

The idea is really fascinating to me, and so I'm looking for any information I can find about how to actually teach this concept to the dog - or to any animal for that matter - but I'm having trouble finding anything about it online. I did find that Ken Ramirez and Virginia Broitman have given talks about this subject, but I can't find any more information than that. I have Ken's book "Animal Training" at work, so I'm hoping there is some information in there. But I'm wondering if any of you have any information you could share about this subject???

Well I know where you were this weekend. Don't forget to tell everyone else where you were! Its good for (my) business!

Today I was googling adduction to see if I could find anything else on the topic, because it is pretty interesting, and I came across this thread, so I had to register and reply.

Adduction is the combination of two behaviors. When giving two cues leads to a combination of the behaviors, or both behaviors occurring simultaneously, adduction is occurring. If the behaviors happen in sequence, the adduction is not occurring. It does not matter if the the cues are given simultaneously, or in sequence.

I was talking to Ken some more about adduction at dinner one night. I was asking him what his procedure for creating adduction was, and if he had a specific cue for adduction.

Ken told me that he has a cue that basically means, pay attention while I tell you what to do. The dogs will not perform the behavior until he gives a separate release cue.
He will give the dog the "pay attention and wait" cue, then two cues for different behaviors, then release the dog to preform the behavior. He will also give supplemental cues while the dog is performing the behaviors. For example, if he gave the cues for bark and rollover, and then the dog begins to roll when released, he will give the bark cue while it is rolling over. Pretty soon the dog will perform both behaviors at once. After training this with four or five sets of behaviors, the dog seems to learn to combine any behaviors it is given in between a "pay attention and wait cue" and the release cue. No specific training is needed, it just sort of happens. He helps it develop and captures it when it does occur.

Basically, to take out all jargon adduction means do this AND this, as opposed to a chain which is do this THEN this.

I don't really know when it would be useful. Ken will be the first to say that some of those things are fun, but not really useful. He also has developed a technique for teaching an animal a mimic cue. One animal does something, and he says "Copy" and the second animal will do what the first one did. He says it really isn't useful, but its fun. Same with adduction and even the "insight learning" that Dr. Epstein talked about. Its probably just something more useful for fun.

It may be useful to consider when your animal (or you!) does something strange. Perhaps it has been given two cues somehow, even environmental cues, and is adducting?

As far as the term itself... It is pretty uncommon, even in the scientific behavior analysis literature.
You could call it by what Dr. Epstein calls it... but I don't really seen any reason to ever use his terminology.

Did you notice that when answering the question about adduction in the panel discussion Ken adducted the words satisfy and gratify to form grastify? How appropriate!

 

[corgipower]

He also has developed a technique for teaching an animal a mimic cue. One animal does something, and he says "Copy" and the second animal will do what the first one did. He says it really isn't useful, but its fun.

Oh, that could be very useful. If I could train all my dogs just by having them watch one that's already trained.

I do have one dog that's a natural mimic, and I have often tapped into it while training him. I haven't been able to cue it specifically though, nor have I been able to get any other dogs to even attempt to mimic.

And, welcome to chaz!

 

[Doberluv]

Well I know where you were this weekend. Don't forget to tell everyone else where you were! Its good for (my) business!

Today I was googling adduction to see if I could find anything else on the topic, because it is pretty interesting, and I came across this thread, so I had to register and reply.

Adduction is the combination of two behaviors. When giving two cues leads to a combination of the behaviors, or both behaviors occurring simultaneously, adduction is occurring. If the behaviors happen in sequence, the adduction is not occurring. It does not matter if the the cues are given simultaneously, or in sequence.

I was talking to Ken some more about adduction at dinner one night. I was asking him what his procedure for creating adduction was, and if he had a specific cue for adduction.

Ken told me that he has a cue that basically means, pay attention while I tell you what to do. The dogs will not perform the behavior until he gives a separate release cue.
He will give the dog the "pay attention and wait" cue, then two cues for different behaviors, then release the dog to preform the behavior. He will also give supplemental cues while the dog is performing the behaviors. For example, if he gave the cues for bark and rollover, and then the dog begins to roll when released, he will give the bark cue while it is rolling over. Pretty soon the dog will perform both behaviors at once. After training this with four or five sets of behaviors, the dog seems to learn to combine any behaviors it is given in between a "pay attention and wait cue" and the release cue. No specific training is needed, it just sort of happens. He helps it develop and captures it when it does occur.

Basically, to take out all jargon adduction means do this AND this, as opposed to a chain which is do this THEN this.

I don't really know when it would be useful. Ken will be the first to say that some of those things are fun, but not really useful. He also has developed a technique for teaching an animal a mimic cue. One animal does something, and he says "Copy" and the second animal will do what the first one did. He says it really isn't useful, but its fun. Same with adduction and even the "insight learning" that Dr. Epstein talked about. Its probably just something more useful for fun.

It may be useful to consider when your animal (or you!) does something strange. Perhaps it has been given two cues somehow, even environmental cues, and is adducting?

As far as the term itself... It is pretty uncommon, even in the scientific behavior analysis literature.
You could call it by what Dr. Epstein calls it... but I don't really seen any reason to ever use his terminology.

Did you notice that when answering the question about adduction in the panel discussion Ken adducted the words satisfy and gratify to form grastify? How appropriate!

Welcome! Very interesting description. Thanks!

Well, I certainly have utilized adduction training and have found that it IS useful for certain things. I know I've used it when teaching some tricks...crawl, for instance. It was begotten by blending down and come and soon got it's own cue.

It is nice to be aware of the technical labeling of concepts with animal behavior. This is just one more thing to make training more fun.

 

[lizzybeth727]

Today I was googling adduction to see if I could find anything else on the topic, because it is pretty interesting, and I came across this thread, so I had to register and reply.

OMG somebody besides me has been googling adduction.

Ken told me that he has a cue that basically means, pay attention while I tell you what to do. The dogs will not perform the behavior until he gives a separate release cue.
He will give the dog the "pay attention and wait" cue, then two cues for different behaviors, then release the dog to preform the behavior. He will also give supplemental cues while the dog is performing the behaviors. For example, if he gave the cues for bark and rollover, and then the dog begins to roll when released, he will give the bark cue while it is rolling over. Pretty soon the dog will perform both behaviors at once. After training this with four or five sets of behaviors, the dog seems to learn to combine any behaviors it is given in between a "pay attention and wait cue" and the release cue. No specific training is needed, it just sort of happens. He helps it develop and captures it when it does occur.

Awesome! Thanks for that info!!

Basically, to take out all jargon adduction means do this AND this, as opposed to a chain which is do this THEN this.

Didn't you hear, though, during Ken's answer during that panel discussion, he said that adduction could mean "do this AND that" (rollover and bark at the same time) OR "do this THEN that" (roll over, then bark)? Maybe the "THEN" one is called something else? And it IS different than a chain!

I don't really know when it would be useful. Ken will be the first to say that some of those things are fun, but not really useful. He also has developed a technique for teaching an animal a mimic cue. One animal does something, and he says "Copy" and the second animal will do what the first one did. He says it really isn't useful, but its fun. Same with adduction and even the "insight learning" that Dr. Epstein talked about. Its probably just something more useful for fun.

Oh, I can think of many applications. If nothing else, it gives us some insight on how animals can learn concepts, not just behaviors. Just like Ken's modifier cues.

As far as the term itself... It is pretty uncommon, even in the scientific behavior analysis literature.
You could call it by what Dr. Epstein calls it... but I don't really seen any reason to ever use his terminology.

Dr. Epstein calls it "insight learning"? Or is that a different concept?

Did you notice that when answering the question about adduction in the panel discussion Ken adducted the words satisfy and gratify to form grastify? How appropriate!

Gotta love Ken.



WELCOME to Chaz!

 

[cvarnon]

It is nice to be aware of the technical labeling of concepts with animal behavior. This is just one more thing to make training more fun.

I agree completly.

Oh, that could be very useful. If I could train all my dogs just by having them watch one that's already trained.
I do have one dog that's a natural mimic, and I have often tapped into it while training him. I haven't been able to cue it specifically though, nor have I been able to get any other dogs to even attempt to mimic.

My mom had one that was a good natural mimic. It can be pretty useful. Ken has talked about his procedure a little, but I'm not sure how many of the details he is ready to reveal, so I won't go into detail. The point he makes is that it is very labor intensive to train a mimic on cue, and he could train the new behaviors faster without the mimic cue.

Didn't you hear, though, during Ken's answer during that panel discussion, he said that adduction could mean "do this AND that" (rollover and bark at the same time) OR "do this THEN that" (roll over, then bark)? Maybe the "THEN" one is called something else? And it IS different than a chain!

Well... I had been there all day, and I was pretty tired because I was running back and forth between my house feeding baby pigeons... so I may have zoned out a bit. I'm just going by what we talked about at dinner that night.

Maybe that kind of chain is a new behavior to the dog? I don't know.

Dr. Epstein calls it "insight learning"? Or is that a different concept?

Different concept.
He showed this video and a few others.
YouTube - A Pigeon Solves the Classic Box-and-Banana Problem

The argument most people make is that the pigeon "thought" about the program and had some "insight" to the solution.
What actually happens is that you have to train each of the component behaviors, and then the animal puts them together. I could be a little more elaborate, but I'm not sure I could do that and be accurate at the same time. I really want to try some of that though, it looks like a lot of fun.

 

[lizzybeth727]

Different concept.

The argument most people make is that the pigeon "thought" about the program and had some "insight" to the solution.
What actually happens is that you have to train each of the component behaviors, and then the animal puts them together. I could be a little more elaborate, but I'm not sure I could do that and be accurate at the same time. I really want to try some of that though, it looks like a lot of fun.

Yes, but that's not the same as adduction theory.

 

[cvarnon]

Yes, but that's not the same as adduction theory.

I agree. I didn't mean to say it was... Lets see...
Oh I mentioned it initially as another example of something that looks like a lot of fun, but not necessarily the most efficient way to train a behavior.