Category: Training People to Train Their Dogs ¤ Author: Deb Jones ¤ Title: Tricks and Clicks Class Notes, Week 4 ¤ > From: Deborah Jones > Subject: CT4 homework > > > CLICKS & TRICKS - WEEK 4 > > Thank you for attending this session of Clicks & Tricks. I hope > you have enjoyed it as much as I have. I also hope that you will > continue to use what you've learned in this class to work on > refining the tricks you've taught and on teaching new ones. You > may also want to transfer this method to teaching household > manners and obedience as well. Any behavior can be taught using > the principles and ideas you've learned in this class. > > Continuous & partial reinforcement > > Our goal in training is to be able to move away from reinforcing > for every behavior, towards a more variable schedule of > reinforcement. We use continuous reinforcement whenever we're > teaching something new. By reinforcing every attempt at the > behavior, we're giving the dog information that he's doing the right > thing. However, once a behavior is well-learned, we can move to > partial reinforcement. Using partial reinforcement actually > encourages the dog to work harder because he never knows when > the reinforcer might be coming. > > We start moving from continuous to partial reinforcement by > doing 2fers and 3fers. We will now require the dog to perform the > behavior 2 or 3 times in order to be reinforced. Start by asking the > dog to perform a behavior he knows well, give light verbal praise > and release. Now ask for the behavior again, C&T. This is a > single exercise 2fer. Do not click without treating. Simply use > some verbal praise and a verbal release. > > You can also do this with different behaviors, as long as the dog > knows them well. For example, ask your dog to sit, release, ask > your dog to down, release, ask your dog to stand, C&T. This is a > combined exercise 3fer. > > By using 2fers and 3fers we're slowly getting the dog used to not > being reinforced for every single behavior. Remember, never go to > this type of partial reinforcement until your dog is very reliably > offering the behavior on command. If you do this too soon, before > your dog understands the command and the behavior, you'll > confuse your dog. > > When your dog will work well on 2fers and 3fers you can move to > more variable reinforcement. You may reinforce on the 2nd > behavior, then on the 4th, then on the 1st, then on the 3rd. Your > dog will never be sure exactly when reinforcement will be > available, so this keeps the dog guessing. > > This type of reinforcement is actually called variable ratio. It > leads > to the best level of response from an animal. The animal gets > reinforced after a number of responses (ratio), but that number > keeps changing (variable). Therefore, the animal cannot predict > when reinforcement will happen. It's important to avoid falling > into patterns. The more surprising the reinforcement, the better. > > When people gamble or play slot machines they're being > reinforced on a variable ratio schedule. They keep performing the > behavior based on the hope that they will get reinforced (win) at > some point. They will play for a very, very long time between > reinforcements. Compare this to a continuous schedule of > reinforcement: getting a Coke from a machine. You expect to be > reinforced every single time you put in your money. If you're not, > you usually give up and go away, you don't endlessly continue the > behavior. > > > Alternative reinforcers > > Up to now we've been using food as our primary reinforcer. It > helps to have a number of different primary reinforcers and to vary > their use. Reinforcers differ from dog to dog. Some will work > hard for praise, others could care less. Some will work for a chest > rub or butt scratch. Others may work for the chance to play a > quick game of fetch. You may have a training toy that your dog > will work hard to obtain. > > Try to be unpredictable in your use of reinforcers. Keep a toy > hidden in your pocket and when your dog does something very > nicely pull it out for a quick game. Some dogs would love > permission to go sniff in the corner. Make that a reinforcer. Give > the dog what she really wants. On walks you might ask your dog > to stop and obey a command, then reinforce by allowing the dog > freedom to go back to the walk. Any event, activity, or object that > the dog likes can be a reinforcer. Observe what your dog naturally > does if left to her own devices, then use it. > > Chaining behaviors > > More complicated tricks may be taught by chaining together 2 or > more behaviors. For a behavior chain you first teach each part > separately, then put them together. A retrieve is a good example of > a behavior chain. It has a number of small parts which must be > taught separately. Then the parts are put together to be performed > on one command. Some examples of possible chained tricks > include: picking up toys and putting them in a box, going to and > closing a door, opening the refrigerator and getting out an object, > sitting up and begging, etc. > > A method called backchaining is often used to put together the > parts of a chain. In backchaining you ask the dog to perform the > very last behavior in the chain and you reinforce that. Then you > ask for the next to last behavior, then the last one, then reinforce. > Then you ask for the second to last behavior, the next to last one, > the last one, then reinforce. This way the dog is always working > towards the reinforcement, which only comes after the last > behavior. Remember, you've already taught each behavior in the > chain first. Backchaining involves putting them all together. > > When you want to train a behavior chain, sit down first and figure > out each small component of the chain. Break it down as far as > possible and train each part separately. > > > > > ------------------------------ > > Date: Wed, 20 Nov 1996 07:48:55 -0500 (EST) > From: Deborah Jones > Subject: inst. CT4 > > > INSTRUCTOR'S GUIDE: CLICKS & TRICKS - WEEK 4 > > Have students demonstrate their progress on 'retrieve' and 'paw'. > Make comments on how to refine those tricks. > > Have students demonstrate their 'target' tricks. > > Talk about moving to partial reinforcement. > > Talk about finding and using alternative reinforcers. > > Introduce the concept of a behavior chain. Show & analyze the > parts of a retrieve. > > Brainstorm chained trick ideas. > > Talk about backchaining. > > Thank everyone for participating. Ask if anyone wants to continue > a few weeks doing some obedience exercises using the clicker. > > Give out certificates and doggie bags.