OC log

[gratefool1]

Hey all,

I was thinking I might start a thread about this new little guy I am manning and trying to train with OC techniques. Prior to this I used the standard manning and training of RTs exclusively, so I am just getting started with clicker training. Please excuse me if my lingo isn't quite correct. I am reading Karen P's book and have been following the threads in here for a while (thanks for setting this up as a separate forum Chris).

Anyway, I thought I would log my journey in case anyone was interested. I always value constructive criticism so please let me know what you think.

So far, we have established food as the positive reinforcer and a clicker as the conditioned reinforcer. He first ate on friday and each tidbit was associated with a click. I have to admit here that I was initially CRing multiple activities that I found desirable, prior to pairing the CR with food. As I understand more, I believe I should have stuck to only clicking with food in the beginning as I am sure he didn't know what the clicks were about.

Saturday, he began jumping to the fist. Monday he creanced indoors very well and has progressed to a variable food reward but still a consistant CR for each jump. Tuesday, I went back to CRing the other positive behaviors such as leaning forward for the hood, stepping to the fist out of the box, etc. Today I plan to creance outdoors.

I know that this is the easy stuff, but hopefully the log will get more interesteing as I condition more complex behaviors... we will see if I can do it =).

[PeteJ]

Just noting a couple of things though before you get too far along. First off...what is 'little guy'? Second, he clearly isn't an eyas so probably having this start under 'imprinting' might be a bit misleading, particularly if you are successful in this attempt. In other words, probably people flying passage birds would be more interested in your progress if they knew it was with a passage bird.
So, if you could clear these omissions up for us, we'll look forward to reading of your developing protocols and responses.
Best of luck!

[goshawkr]

Hey all,

I was thinking I might start a thread about this new little guy I am manning and trying to train with OC techniques. Prior to this I used the standard manning and training of RTs exclusively, so I am just getting started with clicker training. Please excuse me if my lingo isn't quite correct. I am reading Karen P's book and have been following the threads in here for a while (thanks for setting this up as a separate forum Chris).

Anyway, I thought I would log my journey in case anyone was interested. I always value constructive criticism so please let me know what you think.

So far, we have established food as the positive reinforcer and a clicker as the conditioned reinforcer. He first ate on friday and each tidbit was associated with a click. I have to admit here that I was initially CRing multiple activities that I found desirable, prior to pairing the CR with food. As I understand more, I believe I should have stuck to only clicking with food in the beginning as I am sure he didn't know what the clicks were about.

Saturday, he began jumping to the fist. Monday he creanced indoors very well and has progressed to a variable food reward but still a consistant CR for each jump. Tuesday, I went back to CRing the other positive behaviors such as leaning forward for the hood, stepping to the fist out of the box, etc. Today I plan to creance outdoors.

I know that this is the easy stuff, but hopefully the log will get more interesteing as I condition more complex behaviors... we will see if I can do it =).

This promises to be good.....

[tbality]

Hi Ab,
I've played around with clicker training some w/my horses...they love it... even my Italian Greyhound (she's a rescue) that is terrified of my husband, learned to jump in his lap with a couple rounds of clicker training. So my new RT of three weeks, I'm doing what you are and so far so good. She's flying free, at trapped weight, and doing really well, though need to cut her back a bit to hunt I think. I also clicker trained my ferret to ride on my shoulder, it's really cute when you see them "get it"...and repeat the behavior to do it again.

Side note, also working with RT to load in giant hood on her own with clicker...trained the HH to do that a year ago, made it less traumatic for them.

[gratefool1]

Oops!

The "little guy is a 17 weeks old CBMHH. Sorry for that =). I thought I was posting in the OC forum. Did not realize I posted it in Imprinting!! Sorry about that.

Creanced outside for the first time yesterday. The wind was insane but the sun was going down and it was my only chance for outdoor training. He was definitely distracted at first. I would not say that he was as totally focused on me as I am used to seeing in a RT at this point, but he had never been outside without a roof of some sort before.

The good news was he was almost perfect. We started small and went back to consistent positive reinforcement for each flight with both food and click. He did not miss a beat until his tenth flight where we had moved up to sixty feet. He was halfway to me when a burst of wind sent him slightly off course. That distraction caused him to seek a perch in a nearby Crepe Myrtle. He came back down but not to the fist. Called back to fist and started over at a shorter distance. A few more good flights around forty to fifty feet and then he purposefully tried to fly to a different perch and I knew we were done. One more short flight to end on a good note and we were done.

All reinforcement that day was with both food and click and every flight was rewarded. Plan to do the same with variable food rewards today but continue to click. Next day will vary both, but always click with food. I am hoping to fade tidbits out a bit over the following two days and move more and more of his portion to the lure, as long as he is sufficiently reinforced by the clicker. I will continue to reinforce the tangentials such as hopping to the scale and GH, and bending for the hood.

Depending on responses this week, we may be out hunting by the end of the weekend if he is well wed to the lure by then.

I am starting to make some tentative plans for shaping. There are several behaviors I am looking for long term. First, I want to teach the bird to scout a field, circling out in front. I plan to start this behavior by modelling some of Lydia Ash's games progressing from jumpups to remote jumpups and finally to circles.

Secondly, I want to avoid carrying anywhere but back to the fist. I have chased an unruly FHH for a mile through neighborhoods carrying a quail and I do not want a repeat of that ever again. I am amazed by the Troy Moritz article where his MHH returns with kills to the glove. He apparently acheives this by only feeding on the kill from the glove. I am considering some reinforcement with the lure by picking it up with him on it to possibly ready him for the same event in the field. Any input appreciated.

[gratefool1]

Tammy,
Would love to hear more of your training with your FHH =). I think i will start letting Ender front load in to the hood if he can turn around in it. I have been backing him in until now.

[gratefool1]

Regrding the Imprinting forum issue, it appears that all posts in OC are copied in Imprinting. Might be that this is a subforum. Should I move this?

[Dirthawking]

I am trying to follow along at home. Dumbbell question I know but... did you just replace a whistle for food/fist with a clicker?

[goshawkr]

Second, he clearly isn't an eyas so probably having this start under 'imprinting' might be a bit misleading, particularly if you are successful in this attempt.

The "little guy is a 17 weeks old CBMHH. Sorry for that =). I thought I was posting in the OC forum. Did not realize I posted it in Imprinting!!

When Chris set up the Operant Conditioning forum, he had a hard time deciding where to tuck it. He ended up putting it as a sub-forum underneath imprint, which is logical since most people using OC to train hawks are training imprints.

However, its not the ONLY time OC is used, and in fact there are several falconers I know that use a lot of OC on every hawk they touch, and a friend of mine teaches his apprentices OC and has them use it on their passage redtails.

[gratefool1]

Mario,

No I still use the Whistle as a call signal. The click and the food are reinforcers. The food is the positive reinforcer and the click is a bridge or conditional reinforcer.

In the beginning, I am using them together so the bird associates his positive reinforcement with the click, or CR. eventually the CR is used without the positive reinforcer every time. It is like a bridge signal telling the bird you like what he is doing and a reward is coming. As I understand it, this will allow for shaping more complex behaviors down the road. Instead of giving a positive reinforcement (food) for each step of 1) lean forward for hood, 2) step calmly on scale, 3) step back on glove, 4) lean forward for hood removal, you give the conditional reinforcer for each step and then the positive reinforcer at the end.

It is supposed to allow you to train complex behaviors faster by reinforcing each step until it is down, then moving on to the next step.

Once the bird associates positive reinforcement with the click, you have just gained a powerful tool to communicate with your bird.

Disclaimer, I am a novice at this and the above is just my simplified understanding of OC. I am probably butchering the semantics, but hopefully the concept is right.

[goshawkr]

In the beginning, I am using them together so the bird associates his positive reinforcement with the click, or CR. eventually the CR is used without the positive reinforcer every time. It is like a bridge signal telling the bird you like what he is doing and a reward is coming. As I understand it, this will allow for shaping more complex behaviors down the road. Instead of giving a positive reinforcement (food) for each step of 1) lean forward for hood, 2) step calmly on scale, 3) step back on glove, 4) lean forward for hood removal, you give the conditional reinforcer for each step and then the positive reinforcer at the end.

Thats close, but not quite right.

Once the animal has the CR down, they really get it, then just CR becomes a reward by itself, but a minor one. You still give a reward after the CR for any real progress that you want to have stick. Of course, the delima always is when are you offering encouragement and when are you spoiling (or bribing) the animal you are training.

Look at a CR that we all know very well - "Good dog". Thats a CR. A dog quickly learns that when you use that phrase your pleased with it - and this sometimes also includes a treat (attention, food, something). In training a dog you can do a lot with just a "good dog", but to really make a lesson stick you better back that up with something that gets the dog excited.

[andy hall]

Thats close, but not quite right.

Once the animal has the CR down, they really get it, then just CR becomes a reward by itself, but a minor one. You still give a reward after the CR for any real progress that you want to have stick. Of course, the delima always is when are you offering encouragement and when are you spoiling (or bribing) the animal you are training.

Look at a CR that we all know very well - "Good dog". Thats a CR. A dog quickly learns that when you use that phrase your pleased with it - and this sometimes also includes a treat (attention, food, something). In training a dog you can do a lot with just a "good dog", but to really make a lesson stick you better back that up with something that gets the dog excited.

This is a good point. Too often trainers have been lead to believe that the 'Click' or conditioned reinforcer is a replacement for a unconditioned or primary reinforcer like food. It is only a place holder and marker to communicate to the animal that the precise behavior occurred and another reinforcer will follow. Check out the last sentence before the Conclusion section in the article "Building Bridges: How to Select, Condition and Use a Bridging Stimulus" in this publication http://www.theabma.org/images/storie...%20v12.1_2.pdf

Steve Martin presented a paper with Dr Susan Friedman on 'Blazing Clickers' a practice of clicking without the followup reinforcer and had many examples of the breakdown of behavior and some aggression and other side effects. I spoke at the same conference so I should have the proceedings and may be able to post it here.

Also keep in mind that everyone uses 'clicks' or unconditioned reinforcers. Every animal will find something that you usually do right before delivering the food and will look to that as a bridge or marker. It can be a reach into the pocket, a nod or many other things. Check out this video I posted and you can find an example of an inadvertent unconditioned reinforcer (I am saying "good" even though there is now audio, but you don't need the audio to find my 'screwup') there is a spot where the bird gets confused and a little agitated that he didn't get his pellet for what he thought I 'clicked'. It happens around the 2:10 mark.

I like the idea of keeping a log...I know I am very bad about keeping records like this so it will be great to see what you accomplish

Andy Hall

[gratefool1]

You put it better, thanks Geoff. In the beginning you like the click to food nil the bird associates the lick with positive reinforcement and it actually becomes a positive reinforcer like Geoff said.

Well yesterday's creance was characterized by me trying to do too much too fast. Started off a bit slow with the bird distracted by all the stimuli. This was our second creance outside and only other stint outside since being pulled was a couple of hours weathering last weekend which seemed to be counterproductive to his manning.

After the first couple of flights where I had to bribe a response by showing the tidbit, he finally "got it" and started showing instant response to whistle. He even started flying back to perch which is always a good sign of eagerness to me. I started varying the reward and he did great.

I noticed him drop a piece of his tidbit and show no interst in going after it which bothered me a bit. He clearly is uncomfortable on the ground with me near. I began calling to the fist and tossing tidbits but he balked. I would have to lower him to the ground, wait until he was comfortable to hop off, then he would eyeball me and sun deny snatch the tidbit, turning away. I reinforced each tidbit this way with click, but it was slow going and he was clearly losing interest.

Okay so my next mistake. I thought that maybe a bigger reward would solidify the response so I moved to the lure with a big piece of rabbit. Problem one. He had only seen the lure twice previously and it was inside. Problem two, I introduced swinging it around as a call which he has never seen before. I also started a new whistle trill as the lure call. I totally forgot that while theses things were routine to me, he hadn't seen them yet.

To his credit he showed immediate interest and launched but then the newness of whistle, swing, and ground proved too much for him and he flared off. I had to coax him back on and reinforced with clicks once he started pulling bits. I did decide to pick him up on lure to finish the meal, which he did fairly well.

Lessons learned. I need to train myself a little better before training him. I need to break the lessons into separate stages for each goal and work on them separately. We will see what tomorrow brings.

[James Idi]

My training experience revolves around people and dogs, but not raptors. however, the same concepts apply to pretty much all animals, as training is nothing more than learning, and we are all wired to learn in many of the same ways for survival.

In the beginning you like the click to food nil the bird associates the lick with positive reinforcement and it actually becomes a positive reinforcer like Geoff said.

Keep in mind that a conditioned reinforcer is just that, conditioned, and conditioning NEVER stops. When you create a conditioned reinforcer, you do so under a specific set of conditions which begins to have a specific meaning (click = food). Change the conditions, and you will start to change the meaning (click = nothing? Food?), and will lesson the strength of your conditioned reinforcer.

While some may not agree (imagine that...), I prefer to condition and use 3 markers for shaping behaviors.

1. The conditioned reinforcer.
- That's the biggy and lets the animal know that the behavior will be immediately rewarded.

2. A keep going signal.
- A mark that means "Keep doing what you are doing because it will lead to a reward."

3. A no reward signal.
- A mark that means that is not the behavior that will lead to a reward.

Animals are often times smarter than we give them credit, and, at least with dogs, they are fully capable of learning and understanding a wide variety of signals and communications, especially if you are consistent. IMO, once the markers are conditioned and understood, the communication process is so much easier, and this leads to the ability to more efficiently shape behaviors because you can communicate what will, and what won't, lead to the reward.

[gratefool1]

Danny,
I like the idea of a no reward signal. Can that be something as simple as a "no" spoken? I may already be doing that but not aware...
I think the click is supposed to transition to a keep going signal, but I am not sure.
In your model, how do you incorporate variable rewards?
Thanks for your input

[gratefool1]

Today was a great day. I broke training up into three short segments, each with a clear goal and plan.

The first segment was straight creance training. Food was on the glove and rewards were variable. 15-20 flights with spot on whistle response.

Second segment was learning jump ups and weaning the fist / food association. I was amazed at how fast he took to this and how quickly he would come right back to the glove from the ground. He was back up almost before I blew the whistle. I don't think he particularly likes being on the ground. We learned jump ups today while working on this behavior.

The third stage was tying these behaviors together. I would call to the fist from the perch and give a click. I would then toss a tidbit 10 ft away and he would launch. As he landed and grabbed the food I clicked again. I would the call back to the fist with a whistle and give another click. It was awesome!

The only two hiccups were when he jumped the gun again, once on the first and once on the last session. On the first he flared back to the crepe myrtle and was reluctant to come down. On the last, he started to flare but he saw the glove coming up, corrected, and landed on the glove. That was a good sign to me. I was so tempted to reward that but I elected not to, so as not to reinforce the false start.

He was hooded for at least an hour between each session. Best day yet by far

[James Idi]

Can that be something as simple as a "no" spoken?

Absolutely.

As long as the "mark / signal" is consistent and understood, it can, and WILL, be whatever you condition it to be.

"No" can certainly be used and conditioned as a no reward signal. Others use the ol' "lower the fist and turn away" approach, which is, in itself, another type of signal, a physical / visual signal. I prefer the audible, as it is simply faster, and accomplishes the same thing.

I use "yes" as my reward mark / C-R, "good" as my keep going signal, and "wrong" as a no reward signal. The first thing that gets conditioned is the C/R, then I condition the keep going signal, and lastly the no reward signal.

I think the click is supposed to transition to a keep going signal, but I am not sure.

The "click", if used as your conditioned reinforcer for food, should ALWAYS mean that food is on the way. When you want to extend a behavior, or transition to a different reinforcement schedule, you simply don't "click" until you get the behavior you want, or until the behavior has been repeated as much as you want.

In your model, how do you incorporate variable rewards?

Once I feel the animal understands what I am asking it to do, then I'll start to transition to a variable reward schedule while keeping an eye on the motivation level and attitude, but initially, I will use a continuous schedule to shape the behavior how I want it. I'm also a big fan of the "jackpot" reward where the reward size is periodically adjusted on the large size.

As I stated before, my experience stems from people and dogs, and as I continue to research raptors, I can see why Falconers use the approach they do. Because it works. And it has worked for millenia. "Modern" O/C simply helps refine the communication process IMO.

Best to you and yours.

[gratefool1]

Thanks a ton for your responses Danny. I like "wrong" as opposed to "no" as there are fewer words with that sound so there should be less confusion. I think I will work those in.

You mention that the click should always mean that food is on the way. Is there a time limit in this? What I mean is am I confusing the bird by giving a click with every good creance flight but only giving the food reward as a variable? Should the two be synchronized always? I thought that they needed to be synchronous initially while you condition the click as a reinforcer but that eventually, the click became its own reinforcer or at least a bridge to a more desirable/powerful reinforcer. I think I was hoping my click was allowing me time to get a couple more flights in before the reward, giving some positive reinforcement while allowing me to vary the more powerful reinforcer, food.

[gratefool1]

Today was basically a repeat of yesterday. His first response in the morning was sluggish. He sat the perch taking it all in. Longer than yesterday actually. It was much windier today and there was a lot going on between ducks and dove flying over and an occasional crow circling for a looksee.

I refused to show the food and waited him out for what felt like five minutes but was probably more like two. Once he made the initial 30 ft flight he was dialed in. Immediate whistle response every time. Same pattern as yesterday but I adjusted the location of the perch and made him fly to me from different angles.

Second session was thrown tidbits and jump ups. Much better.
Third session was again a combination of creance flights rewarded with thrown tidbits. Again, one attempt to fly to a tree perch. He had just enough creance to reach out and pull a small twig under him. Instead of waiting for him to come down, I put gentle pressure on the creance unil he let go and came down. Probably not the best choice, but after climbing 40' up in a tree on Thanksgiving after his sister who had a floppy Jess tangled, I thought it best.

I would click for the creance to glove flight, toss the tidbit and click as he flew and grabbed and ate it. Another whistle call to the fist got a click as well. Then I would start walking him back to the perch to repeat. If he jumped to the perch of his own volition, he got another click.

At the end of the last session, I called and threw the lure out in front of me four or five feet, stepping on the tether. He left the perch immediately and clearly knew what the lure was but did not hit it initially. Instead he pulled up and hovered a sec right in front of me until I reflexively put my glove up and he landed on this. From the fist, he studied the lure for a few seconds before hopping down a foot in front of it. He walked in and snatched at the lure and tried to carry. When he discovered that he could not, he started to mantle. I made in and stroked feathers and feet and when he bent to eat, he got a click.

His third session was also his first with his transmitter on. He did not seem to notice this unil later in the evening while manning and he began to pick at it.

Tomorrow will be more of the same with the introduction of new anklets and a bell. I will also move him out to the mews tomorrow between the second and third sessions to see how much that impacts performance.

[James Idi]

You mention that the click should always mean that food is on the way. Is there a time limit in this?

The faster the better, however, after the C/R has been conditioned, the animal knows, and more importantly, EXPECTS the reward to follow the C/R, so there is no issue if the animal has to come to you from a distance, or you have to go to it, for the reward.

One of the reasons the use of a C/R became popular in training is that it allows you to capture and mark the exact behavior you want, exactly when, and where, it happens, even from a distance. There will be times when the behavior you are trying to shape or reinforce will necessarily be performed too far away from you to instantly reward with food, however, since you have conditioned the C/R, the animal knows that when it gets the C/R, it has performed the behavior you are looking for, and will get its reward.

What I mean is am I confusing the bird by giving a click with every good creance flight but only giving the food reward as a variable?

Yes. More than likely the bird is wondering why it hasn't been rewarded after it hears the click.

Conditioning is a never ending process, so if you conditioned the clicker to mean "reward", but you start NOT offering a reward, you will change the conditioning and the meaning of the "click". The C/R should ALWAYS be followed by the reward.

Should the two be synchronized always? I thought that they needed to be synchronous initially while you condition the click as a reinforcer but that eventually, the click became its own reinforcer or at least a bridge to a more desirable/powerful reinforcer.

Initially, you are trying to create an association between your C/R, and what you intend to use as a reward, food. Once you have created that association, then the reward should follow the C/R.

I think I was hoping my click was allowing me time to get a couple more flights in before the reward, giving some positive reinforcement while allowing me to vary the more powerful reinforcer, food.

Without the reward though, it's as if you offered food, and took it away. That's not good for anyone. Instead, simply ask for the behavior, and monitor the response and attitude. In time, you'll be able to get a better read of your bird and learn where the limits are.

Remember that the reward you use must be valuable to the critter. While a bird may feel that a tid bit is worthy of a simple hop to the fist, that same tid bit may not be considered worth the effort of a long creance flight, or more than a couple jump ups. That's why the "jackpot" reward is valuable, as it sets the precedence for the bird to not know what it's going to get, and it will be more motivated to keep going in search of a "jackpot" reward. Make the reward worth the effort, and you are likely to get the behavior you desire.

Keep at it. The concepts of modern O/C are not difficult. The real challenge is learning how to interpret the signals your critter sends to you, and developing a progressive training approach to achieve your goals.

[andy hall]

You mention that the click should always mean that food is on the way. Is there a time limit in this? What I mean is am I confusing the bird by giving a click with every good creance flight but only giving the food reward as a variable? Should the two be synchronized always? I thought that they needed to be synchronous initially while you condition the click as a reinforcer but that eventually, the click became its own reinforcer or at least a bridge to a more desirable/powerful reinforcer. I think I was hoping my click was allowing me time to get a couple more flights in before the reward, giving some positive reinforcement while allowing me to vary the more powerful reinforcer, food.

The clicker does give you time to deliver the reinforcer, it usually doesn't give you extra repetitions (animals' response to different CR is different). It marks the correct behavior, like taking off from the perch and flying, so that it knows another reinforcer will be given once it lands on the glove. The more you click for repetitions of the behavior without following up with food, the weaker the click and behavior will become. Pavlov's dogs didn't always salivate when they heard the metronome, he also showed that they also quickly stopped salivating when the food was withheld. Also the body's response to a conditioned reinforcer is never as strong as the response the initial eliciting stimulus.

I try to be a close to 1:1 on my CR to other reinforcement as possible. (I actually don't use any intentional CRs with most raptors, I use clickers with really fast learners like corvids, chickens and vultures) I also shy away from varied schedules of reinforcement until the behavior has been fluent for quite some time and generalized in many ways. Then I work on getting more intensity, quicker response or longer duration by not reinforcing for slow, latent or short response. But they almost always get reinforcement for any behavior that isn't completely fluent.

For verticals, I reinforce for both coming to the fist and then again to the ground at first. Then I fade out the reinforcement at the fist. This creates a chained behavior of flying to the glove and then to the ground chasing the food. I don't think I want them leaving the glove for the ground without something to chase, so I stop there. I can then work on height and response time from there. I am currently working on getting my new HH to go to tree branches instead of my glove so I can get more height workout in each repetition (on his fat days after a bunny baggy, trying to get him to chase mammals with the same enthusiasm as he chases birds lol).

From your descriptions, it sounds like your bird is not anywhere near a fluent behavior stage yet, so I would continue with 1:1 (continuous) reinforcement. One thing that has been shown over and over in the 'lab' and in practice, is that more reinforced repetitions are the most powerful tool to getting to a strong behavior.

Andy Hall

[gratefool1]

Andy,

Awesome input. Thanks much for it. I pretty much knew going into this that I would end up doing a few things wrong, but was hopeful that I would receive the good input you guys are giving me. I am going back to 1:1 reinforcement today with small tidbits for most and the occasional jackpot.

Will a varied schedule of 1:1 click/food association one day and varied association another day be of benefit? Also, I am worried that I will not get the repetitions I want if I give food every flight. It doesn't take much to get this guy's weight up.

Another question for all that has been weighing on me. Does this engender more food / glove association that most trainers of HHs try to avoid? I believe that most people recommend minimizing food on the glove to avoid this association which has been blamed for excessive vocalization. The FHH i trained this fall did very well with the tossed tidbit routine and was minimally vocal until exposed to an extremely vocal sister...

To all, thanks for reading and critiquing. My hopes for this thread is to highlight my conceptions (misconceptions) of OC that I think are fairly general to all novices. As I go through the process of fixing these, I will try to log them so others trying this out can have a smoother road. Please keep following and offering your insights!

[James Idi]

Will a varied schedule of 1:1 click/food association one day and varied association another day be of benefit?

IMO, the animal has a lot to do with dictating the reinforcement schedule based on their attitude and motivation levels. 1:1 reinforcement is best for teaching and shaping new behaviors, but after they know what you are asking, and YOU know that THEY know what you are asking, then it's time to transition to a variable schedule.

An approach I learned to help the transition from fixed to variable is to chain the new behavior with a simpler old behavior/s, and reward after the "series" has been completed. So if you are training a creance flight and you believe you can start changing the schedule, once the bird completes the behavior (landing on the fist), ask it to do something else before rewarding, something simple that doesn't take a lot of effort, like going to a nearby bow perch, and when it does, THEN C/R and reward. Then the next time you do the flight, have it drop down to the perch, then call it back to the fist, THEN C/R and reward. By incorporating additional "low cost" behaviors before "payment", you can gradually reduce the expectation of immediate payment after any specific behavior, however, you maintain the fact that payment WILL come. You just have to keep an eye on the attitude and motivation of your bird, and adjust accordingly.

Also, I am worried that I will not get the repetitions I want if I give food every flight. It doesn't take much to get this guy's weight up.

I had a discussion with another forum member about this very topic. Food as motivation is certainly a limiting factor due to weight concerns and/or satiation. A variable reinforcement schedule will stretch out your training time, however, you've got to take the time to get the behavior down before switching up, which means more food. Incorporating a percentage of washed meat for rewards can help keep the weight down, however, developing a good training plan is probably the best approach. In the end, you can only feed so much, so look at the behaviors you plan on training, and figure the food "cost" per behavior, even with a variable schedule, you are still going to have to make "payment". The best you can do is try to balance the amount of behaviors with the amount of reward available, but some days the animal will be more motivated, other days....not so much.

I believe that most people recommend minimizing food on the glove to avoid this association which has been blamed for excessive vocalization.

I would think that a simple way to avoid this potential issue would be to simply mix up the way you "pay" the bird. Sometimes on the fist, other times tossing the reward, etc...

As I go through the process of fixing these, I will try to log them so others trying this out can have a smoother road. Please keep following and offering your insights!

Don't think of it as "fixing" as much as simply refining, stay positive!

There's a dog trainer named Michael Ellis that has some vids on You Tube discussing modern OC / marker training. The dude is very articulate and does an excellent job of conveying the "how and why" of the techniques involved in an easy to understand manner. His approach isn't much different than any other trainers (OC is OC is OC....), however, it is his ability to articulate and demonstrate it that I truly appreciate, and since he is a protection dog specialist, he has a better understanding and appreciation of aggression, and how to channelize and focus it, more so than the "pet / OB" trainers.

Keep up the good work!

[andy hall]

Will a varied schedule of 1:1 click/food association one day and varied association another day be of benefit? Also, I am worried that I will not get the repetitions I want if I give food every flight. It doesn't take much to get this guy's weight up.

I would not see any benefit to introducing a variable schedule of reinforcing and you will probably find it counterproductive. You can vary the reinforcer though, give different size pieces of different types. I always do this once I start hiding the food in the glove, so they don't make behavior decisiOns on the magnitude of the reinforcer. I have always found they I start off using larger reinforcers with less reps in the beggining of training an animal than I do once a trusting relationship has been formed. I go from feeding whole rodent carcass in the first day or two to 1 cm pieces by the end of a week with a new bird. But by clicking and not offering food, there is good chance you may be prolonging this trust building period.

As for the vocalization of hawks fed on the glove, I have not had this problem since changing to reinforcing so often. I don't only feed on the glove though. I throw food on the ground, feed while weathering, in the mews etc. I worked with some educational falcons (Merlin's, prairies, and peregrines) that screamed alot and they were never fed on the fist and only trained with negative reinforcement (before I learned better) so food association isn't the only thing that leads to screaming.

I went through some of the same thought processes when I began getting interested in training too. "Dont Shoot The Dog" has something to do with this. I have been hearing for years that Pryor is rewriting the sections on variable schedule reinforement and negative reinforcement to correct things, but have not seen any changes yet.

Keep up the training and learning !

Andy Hall

[gratefool1]

Danny and Andy, can't thank you enough for following along and offering input. I can't tell you how nice it is to get rather immediate feedback when I should be doing something different rather than figuring it out after the error results in behaviors that are difficult to "undo".

A question on the lure. Is it appropriate to click multiple times while the bird is feeding? I ask this because Flick initially seemed very reluctant to feed from the lure. The response was odd. He was introduced to the lure indoors and was from the start reluctant to leave the fist for it. The reward is always at least 5x greater than a typical tidbit reward. Initially I attributed this to him feeling vulnerable eating on the ground with me towering over him. He will now fly to the ground for tidbits and is quick to get back on the fist. Yesterday, a bit off kilter from new jewelry, he immediately launched for the lure but flared off at the last second. Took him a full minute to jump down beside the lure and walk up to it. He snatched and tried to carry but could not because leash was restrained. When he settled in and bent to eat, I clicked initially or each bite. As I made in and started slowly stroking his back and legs, he started to mantle. I smoothed his wings back and when he took a bite without mantling, I would click. Is this appropriate reinforcement or should the click be simply when he is served the food? I am a bit concerned that I am click happy.

[James Idi]

Is this appropriate reinforcement or should the click be simply when he is served the food? I am a bit concerned that I am click happy.

The "click" is merely the mark/signal that food/reward is on the way. Initially, giving the C-R/click multiple times while food is being eaten is used to create an association between the C-R and the food/reward. However, once this association is created and understood, there is no longer a need to continuously give the C-R, and by doing so, it could lead to eventual confusion.

Edit: In addition, by varying the size of the reward with the same C-R, you are creating anticipation in the fact that ONE of those clicks will equal a jackpot reward. This same concept and approach is used in slot machines, and keeps folks cranking on handles for hours, and hours, and hours, all in hopes of landing that jackpot reward.

[gratefool1]

Okay. I seem to have a better grasp on the basics now and I believe my hawk does as well. We are creancing extremely well and responding to rewards both from the fist and tossed. Rewards are hidden and varied in size. I started incorporating jackpot rewards and I have added steps to the typical creance. Step one was flight to the fist = reward. Step two was thrown tidbits and return. Step three was combining these into a four step chain where he flies to fist, flies to thrown tidbit, flies back to fist, and then flies back to perch. The second two steps are not rewarded, but they are prerequisites to the "important" steps (to him) of fly to fist and fly to tossed tidbit.

I have two more behaviors I hope to start shaping now. Number one and easiest is hooding (I hope). I plan to reward first for sitting the fist and not shying away as the hood nears position. Once he is calm for this, I plan to reward not moving when i put the beak barely into the beak opening and progress until i have the hood on un-braced. Once he is doing this steadily, I will start to tighten and reward acceptance. I imagine he will regress during this stage and an accelerated repeat of the earlier steps will be needed.

Does this sound like a reasonable plan?

My next behavior is much more complex and needs much refining. I am looking for the perfect tradeoff and a stepwise plan to acheive this. No big.

I actually started this last night by allowing Flick to kill a quail in the dark in an enclosed environment, untethered. I used a flashlight to highlight the quail and once he launched and struck I made in. He did try to move away and mantle initially, but could not move far. I pushed his wings back and spoke while soothing him a bit and secured the quail with the glove. I then tossed the lure with two bloody legs (a huge portion for him) beside him and moved the flashlight to this. Took him a minute to let go and hit the lure. I clicked and let him eat. Pocketed the quail and made in low with soothing talk and gentle moving of the wings to correct the mantelling response. I once again clicked when he relaxed enough to bend down and eat, but I will stop doing that in the future.

When he was finished, he hopped right to the glove held low in front of him and I put him up in mews (he is free lofted). Biggest crop with me so far and he had the satisfaction of a kill (I know, baggie but still...), so I hope his reinforcement in this was strong. I plan to do this several more times to reinforce the tradeoff concept before shaping the behavior to tradeoffs during the day and in a hunting environment. Not sure if I will always use the lure either, but for now it is a huge motivator and I want to make sure he does not feel robbed early on. I don't plan on shaping this until after probably three more sessions like the above or until he jumps immediately and willingly to the lure over the kill.

[James Idi]

I have two more behaviors I hope to start shaping now. Number one and easiest is hooding (I hope). I plan to reward first for sitting the fist and not shying away as the hood nears position. Once he is calm for this, I plan to reward not moving when i put the beak barely into the beak opening and progress until i have the hood on un-braced. Once he is doing this steadily, I will start to tighten and reward acceptance. I imagine he will regress during this stage and an accelerated repeat of the earlier steps will be needed.

Does this sound like a reasonable plan?

Sounds like a very reasonable plan, and that same approach can be used to shape any number of desired behaviors.

Part of the enjoyment I get out of training is developing the process and watching the progression. The great thing about the "innernets" is that if you ever get "stuck", you can usually find someone who has trained your task before and can give you a hand.

Keep up the good work.

[Harris C]

Hey all,

I was thinking I might start a thread about this new little guy I am manning and trying to train with OC techniques. Prior to this I used the standard manning and training of RTs exclusively, so I am just getting started with clicker training. Please excuse me if my lingo isn't quite correct. I am reading Karen P's book and have been following the threads in here for a while (thanks for setting this up as a separate forum Chris).

Anyway, I thought I would log my journey in case anyone was interested. I always value constructive criticism so please let me know what you think.

So far, we have established food as the positive reinforcer and a clicker as the conditioned reinforcer. He first ate on friday and each tidbit was associated with a click. I have to admit here that I was initially CRing multiple activities that I found desirable, prior to pairing the CR with food. As I understand more, I believe I should have stuck to only clicking with food in the beginning as I am sure he didn't know what the clicks were about.

Saturday, he began jumping to the fist. Monday he creanced indoors very well and has progressed to a variable food reward but still a consistant CR for each jump. Tuesday, I went back to CRing the other positive behaviors such as leaning forward for the hood, stepping to the fist out of the box, etc. Today I plan to creance outdoors.

I know that this is the easy stuff, but hopefully the log will get more interesteing as I condition more complex behaviors... we will see if I can do it =).

In the 3rd paragraph of the first post there is a HUGE problem that is going to cripple your attempts right off the bat... but with a little insight from a professional trainer (me) hopefully I can help you understand what it is exactly that you need to know. A clicker is a BRIDGE (bridging stimulus) - NOT a conditioned reinforcer. A conditioned reinforcer is a learned reinforcer, or what we professional trainers call a "secondary reinforcer". A good example of secondary reinforcers for different animals are head scratches, praise etc. Humans are also motivated by Secondary Reinforcers such as money. We did not evolve needing money to survive, the dog did not evolve to need head scratches, therefore these are just conditioned stimuli that the amimal has come to view as pleasant or desireable. With a BOP you should just stick to Primary Reinforcers (actually, just 1 Primary Reinforcer - food) because these "play/social interests" that comprise every secondary reinforcer under the sun are not going to motivate a raptor the same way it would motivate a parrot for example. I suggest you try joining the IAATE (International Association of Avian Trainers & Educators) and furthering your training knowledge. I'm a professional-level-member now and it has been very informative and it is definitely worth the $40 or so dollars per year membership. Good luck with your training, I only read the one post so far and thought this was important to clarify for you so I will comment if I see any other issues when I read the rest.

[Harris C]

I am trying to follow along at home. Dumbbell question I know but... did you just replace a whistle for food/fist with a clicker?

No, the clicker is a bridging stimulus (communication tool) and the whistle is a cue for the animal to perform a trained behavior. In a nut shell, you blow the wistle and put a tidbit on the glove (this is a cue to fly to you) and when he complies by flying to your fist you bridge (click) which tells the animal he did something right and food is on the way. The only way the clicker is effective is if the bird gets a food reward EVERY TIME he hears a click. The clicker bridges the gap between the human and animal mind thereby allowing us to communicate using a language we both understand. A great way to learn how to train with clickers effectively is to play a little game with your friends: Try "training" them to do a simple task such as opening the door or pouring a glass of milk or making a sandwich etc. As you do this you realize how many steps are involved in a seemingly simple task such as making a PB&J sandwich! You can't use any words, only clicks to let them know they did something right and moved toward the end goal you are working towards. It is like the "Warmer/Colder" game except there is no "colder", we ignore unwanted behaviors and reward even the smallest approximations or steps towards desired behaviors.

[gratefool1]

Thanks for your input Christopher. I have read quite a bit about all of this and honestly this has been a confusing thing for me to understand. My main resource has been karen Pryor through her book and articles in Hawk Chalk as well as forums.
I do not think I am alone in voicing the confusion about what exactly the clicker is supposed to be, and I appreciate you chiming in to help clear it up. Is it possible that role of the clicker is supposed to change during the training? I ask because so many articles imply that the click is a bridge as you suggest and others describe it as becoming a secondary reinforcer after its use has been ingrained. This led to much of my confusion as to "when to click" etc.
Karen recites times where the click was used alone as a secondary reinforcer, I specifically recall a story about a horse who needed to have a hoof soaked and was resisting until the clicker was used. The horse was clicker trained and took that click as a communication tool indicating that leaving the hoof in the soak was the desired behavior and he complied. In that example, it would seem that the click became the secondary reinforcer.
What I am starting to feel out is that maybe the role of the click can change over time. Initially a bridge signal indicating that a reward is inbound, with enough repetition, do you believe that the click can start to become a reward in and of itself in a way?
Again, I am a newb navigating through a lot of info that at times is a bit confusing. I am trying to highlight my hangups because I assume that other people new to this may run into the same stumbling blocks. Thanks for being a part of this thread.
Ab

[gratefool1]

Okay, so I haven't been updating my progress as much lately because we were in a time of repetition mostly. Up until saturday, Flick was perfecting his creance work and working on hooding, which is progressing, but not as fast as I would like. The main reason for the hooding issues is the falconer not sticking to the plan and I will explain below.

Permit me a moment to brag. Flick had his first free flight on Sunday. He was more than ready for this (delayed a few days due to timing issues). Everything went perfectly with absolutely perfect following and whistle responses. His perches were nice and high and it was an absolute joy to watch him stretch his wings and learn what he can do in the air. Towards the end of our time, we came across a borrow pit pond that I checked every trip out last year for duscks and never saw any. Well they were there on Sunday =). Eight mallards who saw us and made their way to the middle of the pond. They actually let me get fairly close with Flick on the wrist before they flushed. He was after them like a flash and chased down a hen from behind and struck her but did not bind. She hit the bank and hopped back into the water, leaving a few feathers behind. He stayed on the bank staring at her trying to process what had just happened! I tried to flush her again, but she swam to the other side of the lake away from Flick and then flew off. Pretty awesome first flight, even if we didnt end up with game in the bag. I was incredibly impressed with his speed to be able to tail chase a mallard from the fist with a twenty yard plus head start.

Okay, back on topic. I am continuing to click with every reward and Flicks responses are excellent. I have varied the size of the rewards and incorporated jackpot rewards. He is very motivated and doing well when training. There is still a ramp up to training that is getting easier. He is still a bit slow to come to the glove in the mews to get started, but no longer hesitates once we are outside. I attribute this more to the fact that the bond is still tenuous rather than weight. Also, I try not to feed in the mews, and he knows that the glove is likely empty at this point.

Anyway, the only behavior I am really shaping right now is hooding. I had a good plan in place to get him to hold still for the hood, but found it was hard to give a tidbit for each click while holding the hood and clicker in one hand and the bird in the other. As such, i tried to just go ahead and put the hood on loosely, place a tidbit on the glove, click, remove and he found his reward. I did this for two days and he still would fight the hood by twisting his head away, but happily look for the tidbit upon removal. I was hoping he would get the fact that you have to be first hooded to get unhooded, but...

Anyway, I realized i was simply rewarding being unhooded, not being hooded so i am going back to the drawing board. i believe that I will try Harry McElroy's method of having a whole breast in the glove and simply allow access ater a click. This will necessitate some remedial training to get used to a new delivery method but no big. I tried this briefly but knowing there was food in the glove, Flick just beaked the glove repetitively and ignored other stimuli. I assume if i am patient enough, he will give up and focus on me again. I will PM Harry and ask him to chime in here.

So that is where we are. Happy overall and hoping my mistakes aren't overcomable (is that a word?). Thanks for following!

[tbality]

Sorry to delay in responding! holiday season and all. I've ready Karen's book, some chapters several times and have taken a BOP OC class thru Andy Hall in Colorado; and then just done a lot of experimenting on my own. My horses love it and get very animated about it so does our Italian Greyhound rescued dog that has some "issues" from prevoius owner. My ferret loved it as did a kestrel I fostered, my FHH and my newly trapped RT respond well too.

When you click, you MUST treat...always. Just remember that. Karen's book didn't get that home to meet the first time I read it....and I tried to replace treat with the click or just clicked for what I liked and didn't always treat, so that doesn't work. You can up the anty and increase the behavior by withholding the click, and you will see them try different behaviors searching for that click. The click MARKS the thing you want exactly when they do it. Then you can get the treat out and they will wait for it, but know what the treat is for. For instance, when I was working on lead changes on my horse, I would click at the moment he changed leads, then he would stop and get his treat. He knew he was being rewarded for the action in which I clicked, but not for stopping to get the treat. That was key for me...I click when my RT or HH take off quickly at a distance to come to me...that in my thinking tells them taking off NOW is what I want. Or I click just when she hits the lure...then she gets the meat off of it. Don't do too many things at once, get a behavior down, then move on, they get bored doing the same thing over and over. And you can increase the effort when they "have it down" by withholding the click, to increase speed or distance.

It's really fun to see the "get it"...when they understand what that click is, they really show it and it brings out a cute, playful side of the personality of every animal I've tried it on at home.

[James Idi]

A clicker is a BRIDGE (bridging stimulus) - NOT a conditioned reinforcer. A conditioned reinforcer is a learned reinforcer, or what we professional trainers call a "secondary reinforcer"

In clicker training the terms are often used synonymously.

From Karen Pryor's web page:

Conditioned reinforcer

A neutral stimulus paired with a primary reinforcer until the neutral stimulus takes on the reinforcing properties of the primary. A clicker, after being repeatedly associated with a food treat or other reinforcer, becomes a conditioned reinforcer.

Bridging stimulus

An event marker that identifies the desired response and “bridges” the time between the response and the delivery of the primary reinforcer. The clicker is a bridging stimulus.

The clicker is first paired with a primary reinforcer, and becomes a conditioned/secondary reinforcer by association in the classical sense. This conditioned / secondary reinforcer is then used as a bridge / event marker to shape behaviors.

[gratefool1]

Late update. We have not been focusing as much on the OC stuff as we have been trying to get as much time in the field as possible. I have to admit, the clicker training has gotten lax with the excitement of hunting. I am focusing on correcting some of the more typical early hunting issues by simply getting the bird out and hunting. He seems almost a bit overloaded with just this so clicker training has taken a back seat.

After his near miss on the mallard on the first day out, he has chased feather fairly well, missing on several duck slips by being out of position. Having entered several birds on ducks at this particular place, it definitely takes time for them to learn how to make the slip work. These are wood ducks in a small lake off the river with numerous cypress snags out in the middle to perch on. To be successful, I need the hawk to take a desirable perch and wait for me to flush. The young hawks seem to get overly excited and try to strike on the water which never works. Hawk gets wet, ducks fly away. Longterm plan is to get the hawk to sit the fist until i see exactly where the ducks are, take a perch of my choosing, and wait for the flush. May consider laser pointer for this. Any successful hawkers doing something similar?

Regarding the clicker. The quagmire of my understanding just keeps getting deeper. I have read recently some of Steve Layman's OC work involving removal of a negative. I am not clear but I think he is clicking for the desired behavior and then removing the negative as a reward. To me this is a shift in the meaning of the clicker once again and maybe supports the idea of the clicker becoming it own independant reinforcer indicating something positive happening in the near future but that positive is not necessarily food.

Whew...

Just when you thought it was safe to go into the forrest...