OC log

[andy hall]

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.

I have used a laser pointer as a target with several types of animals including owls, but not for falconry.

Now for the second part about using a click after removing an aversive. It may help to more deeply understand the types of consequences that can occur after a behavior. There is a functional (how the animal responds to it in the future) and a operational (how the trainer or environment acts to create the consequence). So the functional aspects are a consequence can either increase the likelihood an animal does the behavior which is reinforcement, or it can decrease the likelihood which is punishment. The consequence is also either positive or negative, which is a algebraic concept. If something is introduced as the reinforcer or punisher, it is positive, like introducing food as a reinforcer. If something is removed, then it is a negative reinforcer or punisher. So negative is not synonymous with aversive or bad in behavior or training. I presented a paper a number of years ago to really define these terms called "When Negative is Good and Positive is Bad"

So Steve is using a clicker to pair with the removal of an aversive stimulus to create a negative reinforcer and CR. The click still becomes a Conditioned Reinforcer and can signal the comfort that follows whatever is being removed. So when done properly the bird will actually have some sort of chemical release or decrease of something to make it feel better.

I would guard against using this technique unless you have no other positive reinforcement route to try. There are several downsides to using a negative reinforcer as a CR. The first, which has been talked about a lot in this thread, is that a weak CR like a clicker must have continuous pairing with the another reinforcer to maintain its effectiveness. Which means the aversive stimulus must be introduced before the behavior occurs, which increases stress with all the bad side effects of using negative reinforcement.

The only time I rely primarily on negative reinforcement is when I have first trapped a bird and they are in the time period where they are even too afraid to eat (a day or two). I then enter their comfort space when they start looking stressed (a slight head turn usually), then I wait a couple of seconds for them to return to a comfortable body posture and I reinforce that posture by leaving. I don't use and purposeful CR, but I don't use one with raptors anymore at all. As soon as they start eating I rely more and more on food as a positive reinforcer.

I hope this helps a bit. There are some other great resources out there about behavior, just not too many using raptors as models, but the principles are all the same. Check out anything by Paul Chance about Applied Behavior, Good Bird, Inc., the Avian Ambassador's blog, and even some paper at NaturalEncounters.com I even recommend taking the Living & Learning with Parrots online course from behaviorworks.org I used a falcon as my study bird for this course.

Keep reading and learning. It took me several years to really grasp many of these concepts and I still find myself calling up mentors and researchers several times a year to clarify nuances of strategy and terminology, even after 17 years of training hardcore. But spending this time learning can really improve your ability to look at behavior and train much more effectively.

And definitely spend as much time hunting as you can, you can work on all this stuff while out there too and when the season ends as well.

Andy Hall

[kitana]

Andy, that is exactly how I used negative reinforcement in my hawk: when he was afraid of me at the beginning and wouldn't eat yet in my presence, I waited for him to become calm with a relaxed body posture then I left, thus the R-. It has been used by some animal trainer friends of mine in zoos with very fearful animals that would jump on the fence when approached at less than 200 foot, they would approach just close enough to make the animal wary, then leave as soon as the animal relaxed.

One has to be very proficient in reading the body language of the animal to use R- in that manner, because if you put a little bit too much pressure you'll cause a real fright and teh animal only learns to fear you... But when correctly done, it's an awesome tool. I use it a lot as a veterinarian too, when I see a fearful animal unable to eat because fo stress, I'll always use R- with this animal before resorting to the muzzle and contention. It works in about half the animals I see.

[kimmerar]

I even recommend taking the Living & Learning with Parrots online course from behaviorworks.org I used a falcon as my study bird for this course.

Andy Hall

I used an american kestrel when I took the course. The ABC's do help with learning to study the student.

KISS is a motto I've heard many times. Think plan do is another - along with knowing when the plan needs to be modified.

[gratefool1]

You guys are awesome! I appreciate the patience as I continue to try to improve my grasp on this. I swear I haven't wrestled with anything this ephemeral in awhile. I feel like I understand completely as I read, and then i stop to review and what I just learned fades like smoke.

This is the first time I have heard of the online course. I will look ino it for sure. Andy, Thanks for making me feel better by insinuating it took you awhile as well . Kit, I am very interested to read more about your kestrel.

Happy New Year all!

[Steve Roberts]

What a fascinating thread.

I don’t want to take it off at a tangent, but can someone direct me to any information sources that deal with best thinking / practice regarding OC use training falcons for game-hawking.

Some years back Karen Pryor wrote a couple of Hawk Chalk articles and I recall an article by Jimmy Fustos chaining behaviours for sage grouse hawking. There seemed to be very little written beyond this at the time.

Personally, I have used clicker training to a limited extent for some basic behaviours with falcons (basic handling, hooding etc) and training all basic commands with my pointer, but most of the difficult/distance stuff remains traditional in approach.

Probably, much of traditional falcon training practice fits OC principles so maybe there isn’t a large gap with how OC practitioners would go about training longwings. Nonetheless, I think it’s at least interesting to consider training techniques from an OC point of view.

To take simple examples, swinging a lure would seem (I think Karen Pryor made this comment) to be a bit of a compromise (as a bribe)compared to a proper return/recall signal but perhaps no more than a common method of teaching the sit where the dog is shown the treat above its head. I wonder also whether kite training would be viewed with suspicion by OC trainers, and whether this is a serious objection or not.

The most vital technique training gamehawks is of course rewarding the act of climbing. However, there is no CR with the reward (a serve) being instant, and many have difficulty timing the reward i.e. how long do you wait each time. On this theme, how could ‘keep going’ be built into the training scheme: again Karen makes reference to such matters but has the thinking moved on?

If a falcon fails to climb and/or sets its wings then many falconers stand still (not ideal and probably a ‘no’/’that won’t work’ signal is better as some do e.g. stick arms out) and if that doesn’t work bring the falcon down to the lure with limited reward or immediately go to the kite. These are all valid techniques but it would be interesting at least to see them analysed and indeed critiqued in terms of the rigorous framework that OR provides.

Hope this doesn’t derail the thread but then again maybe it’s all relevant to training Redtails or Harris Hawks to wait on.

So does anyone know of articles , web-sites etc that address this topic?

Best of luck Ab - I hope you get the rewards you deserve for your considered approach with this hawk.

Steve

[gratefool1]

Hey Steve! Thanks for jumping in. You are hitting on one of the many frustrating points I am trying to wrap my head around in regards to OC training and these birds.

The first is the temporal component of when to CR in relation to the reward, ie., how far apart can these two be without loosening the association. I hope that trail and error will show that that time interval can be slowly extended once the association is solid. I think this will allow more extensive use of the CR as a communication tool in shaping more complex behaviors.

The second issue is spacial. How do you CR a behavior that is occuring 1500 feet above you? Or for that matter, 100 yards away? I suppose the flash CR that Karen describes coupld be utilized here. I don't know that much about kiting, but I suppose you have the reward ready to be served once the behavior is acheived. I believe it was in Karen's book where she spoke of using a bright flashlight as a CR. You mention handler gesticulation that might be just as effective. Perhaps a beeper attached to the kite with a remote trigger? I see weight issues with that. Not sure a whistle will travel that far either.

I am wondering how to do the same shaping of a behavior at a distance when you do not have the reward near to where the behavior is occuring. For instance. I have several slips where, ideally, I would like to use the bird to circle the lake, pinning the ducks to the water and then take a perch in a snag on the far end. I would take postion with the dog opposite him and move in towards the flush, basically flushing the ducks right towards him. I realize that this is a tremendously complex behavior that will take forever to shape, but hey... I dream big.

My concern is that in shaping this behavior, I will need to call the bird back to me for the reward after the CR is given (I plan to shift to the flashlight or laser pointer cr at this point). The added action of flying back to me changes the behavior, as I want the bird to stay in position until the flush, which may be immediate, or may be delayed up to fifteen - twenty minutes.

I am rambling a bit, but I don't see a way around the spacial issue without reliance on the time issue. I plan on trying to really get my bird well wed to the CR and be able to stretch out the time interval even though some have cautioned me not to. I think it is the only way I will be able to shape the complex behaviors.

Just my $.02, which when weighted for inflation is actually only $.0034 in 1970 value...

[gratefool1]

Well the CR stuff has taken a back seat to hunting of late. I have taken the bird out to hunt every opportunity I have had and only turned back to training when the weather or work would not cooperate.

On a good note, we killed our first rabbit last week. It was the first time that he saw rabbits and i was impressed. Moreso because he was hunting over dogs for the first time. A friend brought his beagle pack out and was kind enough to let us fly over them. Flick has been around our pates at home with no problem but never inthe field with a dog. He was awesome and understood exactly what they were doing. In fact, as we barnched ahead of the dogs, he would frequently dart back to check on them and then back up front into position. He had four gorgeous wingovers that resulted in misses due to thick cover (I had never seen that type of flight from him before) and then connected on the last flight of the day just prior to calling it quits. Beautiful head hold. Beautiful flight. It is amazing to me what is pure instinct to them as he had never seen a rabbit before, much less seen one killed.

On a few days, I went back to the OC training to correct a few minor behaviors I was growing frustrated with. The main one was allowing me to leash him in the mews. I freeloft this bird and when at weight, he would dart to the glove but then flare off as I moved to leash him. If time were pressing, I would simply grab his jess and leash him while he was bating which certainly made the problem ten times worse (dumb).

To fix this, I started with food in the glove hidden and would call. If he would not come immediately, I would show the food the cover. This would get him on the glove. Once on the glove, I would click and tidbit. On successive flights to the glove, I would touch the jess then CR and reward building it to where i could play with the jess, clip him in, and then CR/reward. I added weighing as the last step. He did this great for two days while it was stormy.

The bad weather persisted, so I decided to go back to the hooding process with CR as well. This did not go as smoothly and in fact, undid some of the shaping I had just acheived. I would call to the fist, clip in, and bring out the hood. This negative started the bating all over again, but he would gradually allow me to get the hood on with a struggle. Once it was on, I would click and tidbit. He started looking for the tidbit in the glove when the hood was on, so I moved to tossing the tidbit after the behavior was acheived. Unfortunately, the adversive of the hooding was stronger than his desire for food, so he would not come back to the fist.

I am going to have to start over with tossed tidbits and go extremely slowly here. Once I have comfortable glove response in the mews without a wingflare, I think I will go somewhere else to work on the hood, so as to separate these behaviors.

[Steve Roberts]

Ab, Everything I've read suggests only a few seconds between click and treat - never seen any mention of being able to extend the time interval and be sure the reward is still associated in animal's mind with the click. However, several people stress that the animal works for the click so reward every time is not essential and variable reinforcement schedules important too.

My pointer cannot take a treat when she has had lots of scent e.g. a point or even after the flush but I still click on occasions being confident that's a more timely/direct form of 'good girl' - something I did read somewhere in book/article on clicker training gundogs.

I think the distance click is maybe not such an issue. You just have to ensure something which can be seen or heard at a distance, whether light or whistle. The early 90s Hawk Chalk Pryor/Layman articles talk about this. So rewarding being at 1500 feet is probably not difficult (if indeed that's the right thing to do) but it's the act of climbing that needs rewarding. Traditionally falconers have simply served and waited longer each time. OC would suggest that clicking first would be more effective. What I don't fully get are the references made to 'keep going' cues i.e. is that a matter of clicking once during a prolonged activity, clicking many times, or having a cue which continues throughout. This is possibly relevant to some of the things you want to shape.

Re your circling over pond requirement, in those old articles, maybe Layman or Steve Martin, talks about shaping a hawk to not land when cast off the fist and also (I think I recall but maybe I'm dreaming) shaping a redtail to climb/wait on. This seemed initially to require just the immediate click and return for food that you mention but in the confidence that the hawk 'gets it' quickly and you can move on and extend the time the activity runs. By their approach you'd start like this and build up to longer flights. My fear would be that the hawk would stay flying longer each time but stay very close anticipating the click/reward but maybe that would just be a short-term problem and later you could start to click only when the bird goes out a bit. Presumably attaching a cue to all this would be important.

When you say 'stretch the time interval' if you mean extend the time interval of the activity then surely that's fine but if you mean extend the time interval between click and reward then surely you will weaken the association between the click and reward - accepting that variable reinforcement may be important anyway.

Regards,
Steve