;)Missed that post along the way. BTW, I’ll second all the things you’ve said here. Fun stuff! We taught one of our dogs to open (unchained) gates. Seemed like a cool thing to do, until she got so good at it that we now have to be meticulous about locking the gates or else we have horses in our front yard.
I lived on a rural ranch for some years, where the owner had horses. One of them was an escape artist. Nobody had to teach her to open (unchained) gates – she figured that out all by herself.
She also figured out how to open chained gates by the simple artifice of ripping the whole gate off its hinges!
Fortunately, however, that particular horse was also a highly people-friendly sweetie pie, who would follow people around like a puppy dog. When she had a foal just a couple weeks old, I was out there petting the foal and draping pieces of rope over its body and face, even while it was laying down, to get it accustomed to people and ropes and halters, etc. Mama Horse didn’t interfere, except to make it clear that she wanted some of the attention too.
BTW, I just noticed that the OP is none other than dolphinboy himself!
Didn’t we see, from some earlier thread, that dolphinboy has also some history in working with dolphins? Or am I imagining that?
ETA: Yes, I knew it!
Agree with most of the preceding – dogs work for your approval (it’s one of their most admirable traits) and small treats of things they especially like are ideal for training, much as we ourselves nibble on luscious hors d’oeuvres for a treat. Immediate reinforcement is critical for training.
But don’t overlook the big treats too, on special occasions or just because he’s been a good boy. On his birthday or if he had to endure a trip to the vet, my dog would have a big grilled rare steak on a bed of kibble for dinner, and ice cream (Haagen Dazs vanilla) for desert. And he always had real food on his kibble anyway. That has nothing to do with training, it’s just love. 
Yes, I trained Bottlenosed dolphins, Orcas and Sea Lions for 5 years, but I don’t want to use the same conditioning techniques I used to train those animals on my lowly canine. I had very structured training sessions and the animals had to perform in shows… it was my full time job… so it was very rigorous and everything was planned out well in advance
I’m trying to train my dog to behave well and stay close to me when we are outside and off leash. I will try RIR and see how he responds to that. He’s a very smart JRT and is always trying to outsmart me… which isn’t all that difficult.
Some variation of the same basic techniques should work just fine. Standard behavior shaping techniques; RIR; some Aristotelian symbolic logic, and a dose of Benjamin Spock. Or was it Skinner? Yeah, there’s the ticket! Build your doggie a Skinner box!
Dogs have a tendency to follow the Masterperson closely, so shaping that to teach a dog to heel tends to come easily. But to make sure doggy heels all the time and not just most of the time, you need to add some negative reinforcement every time he wanders off. ETA: Oh, and I was going to mention, this looks like a good case for target training, like we did with Zonker the epileptic sea lion (Post #17 above).
At the annual week-long IMATA conference some years ago (circa 1983 or so?), a young female trainer gave a presentation on training their walrus. I think she was from one of the Sea World locations. She may have been a good trainer, but she was definitely a novice at giving a presentation, and was noticeably nervous and unconfident at public speaking. At the end of the presentation, during questions, somebody asked her if the walrus ever gets belligerent and how they handle that. Suddenly, she was clear and confident and unwavering in her answer: “Oh, we beat her senseless!” – to much applause and laughter.
Other neeto-keeno sightings at IMATA conferences:
[ol][li] Somebody trained a dolphin to balance a ball on the end of its snout, sea lion style. Damn! For a dolphin, that’s a real world-class trick! (Sea lions cheat.) They won behavior-of-the-year for that one.[/li][*] Octopus trained to go through a little hole in a little barrier, on some specific signal. Problem for octopuses is that the tentacles all tend to act more-or-less autonomously, so it gives the appearance that each tentacle is trying to go through the hole independently of the others. The barrier had several holes in it too, and the separate tentacles didn’t all agree on which hole to go through. But once the head went through, the rest of the tentacles eventually followed.[/ol]
Cool! When I was an undergraduate, I worked in a neurophysiology lab. We “taught” invertebrate nudibranchs (*Hermissenda crassicornis) * to avoid light using a noxious stimulus. Then we dissected out their CNS and recorded action potentials from their photoreceptors, comparing results to normal controls. Fun times!
And? Were you actually able to detect measurable physiological changes in your educated nudibranchs?
We showed (in a paper I coauthored) that serotonin had an effect similar to training in the one cell we were recording from. Basic science. It was all pretty dry stuff. Hours of recording from photoreceptors cells, with most of it blinded (experimenter had no idea what treatment, if any, animal received). It was similar to the work Eric Kandel was doing on a different sea slug at the time.
In fact, once Kandel called to ask my boss a question about a paper of ours he was peer reviewing. Kandel was the man in neurophysiology at the time. I wrote “Eric Kandel” and his number on the dry erase board by the phone, since boss was at lunch. Boss asked me not to erase the note. It was still there years later when I stopped to visit.
In essence avoiding a scenario of negative punishment (complete removal of the treat) when the dog sees the conditioned stimulus and performs the behavior until the conditioned stimulus takes on it’s own level of habituation (possibly via pavlovian-instrumental transfer) independent of the reward? Have I got that right?
I haven’t seen this said quite like this, so I’m piling on:
A truly excellent vet of ours once said that dogs have no sense of “volume,” a pea-sized treat was the same as a basketball-sized treat to a K9. We used to give Milk-bone large biscuits for treats and the one dog ballooned up to 20% over her target weight. TEVet “prescribed” Iams puppy biscuits and the weight dropped right off.
With RIR, don’t be surprised if your (intelligent) dog is really thinking That idiot owner of mine forgot the treat this time; I shall remind him, instead of Shoot, I didn’t get the treat this time. If/when you do the “withholding,” don’t be surprised if the mutt has picked up on a pattern you didn’t realize you were doing.
Supposedly, a Border Collie has the same intelligence level as a five-year-old child. Puts her out of my league.
You also have the conditioned vs. the unconditioned consequent (reward or punishment). You blow the whistle then toss the dolphin a fish. Keep doing this, and the dolphin associates whistle with fish. The result is that the “conditioned consequent” (the whistle) becomes a proxy for the “unconditioned consequent” (the fish), and can be thereafter be used as the reward for further behavior shaping.
Dog minds are more complex and emotional than pure behaviorists would like to believe, but behavioral conditioning is very well-established as a way to elicit and fix desired behaviors. It works on people too. Works on almost anything that is self-propelling. Behavior mod encompasses a lot more than just the “positive reinforcement” of treats for desired behavior though.
As far as the OP question goes, among behaviorist dog trainers the norm is to give tiny treats to keep the dog working for more, and because treats are essentially just communication, not a part of their regular diet (although greedier dogs will work with enthusiasm for pieces of their dinner kibble).
Another reason to feed tiny treats is that dogs easily recognize what are known in the biz as “jackpots” --using higher value treats (fried chicken instead of the usual dried liver, for example) and a larger portion, for a behavior breakthrough, will most quickly solidify that breakthrough. You need to create enough difference between jackpot treats and regular treats so it is noticeable to the dog.
Or, alternately, behavioral conditioning is more complex and emotional than non-behaviorists realize. There are an awful lot of extremely complicated and subtle psychological phenomena even in humans which can ultimately be traced to very simple principles.
Agree with this.
Some Buddhists describe emotion as simply the combination of a thought and a physical sensation, to which we put a name.
B. F. Skinner was fond of saying that, given all the right behavior-shaping contingencies, he could train two pigeons to play ping-pong – BUT, he was always careful to explain, that they would just be going through the motions of playing ping-pong (however detailed and elaborately), but would have no actual concept or understanding that they were playing ping-pong.
One of the directors at the dolphin lab had an issue with this: He claimed that this is exactly, no more and no less, what humans do when playing ping-pong, except that humans are going through the shaped behavior motions based on a much more elaborate set of shaping contingencies. (For example, the rules about who serves each play, and whether the winner of a given play gets a point if he didn’t serve that play, or just gets to serve the next play, or if you simply change servers after every fifth play, or whatever the ping-pong rules are.) And if you could get pigeons to learn an equally complex set of contingencies, they would then be playing bona-fide ping-pong too!
Heh, and to think; I wasn’t even able to teach my ex to drive a stick. *(desired smilie currently unavailable.)