I had an interesting argument with Dr. Ian Dunbar last week. I highly respect Dr. Dunbar, but I disagreed with several statements in his latest videoblog and posted my comments on his Facebook page. Thought you might find it interesting. Please watch the video first.
Morten Egtvedt: Hi Ian! I respect you highly, but on this subject I disagree strongly with you. I train much like a computer, and I´m proud of it :-)
Dr. Ian Dunbar: @Morten I'm not saying that we shouldn't compute. I objectively quantify response-reliability more than any other trainer that I have met. I just said that the average person could not compute an ongoing VI and train dog at the same time. Try this ... let's consider a VI5 - give me ten numbers that average out to 5. I do this in every lecture, but the only problem is that I have to concentrate and couldn't train a dog at the same time.
However, I ruthlessly count number of commands given to each correct response and then calculate the Response Reliability % = # of correct responses divided by number of verbal and signal commands given times 100. Walk in a straight line and every three steps, stop and ask you dog to "Down" (no handsignals and definitely no food) for a total of 10 Downs and then calculate your Response reliability %. The results will blow you mind.
I try to work people (and dogs) up to 95% reliability on rewards and motivation alone before introducing non-aversive instructive reprimands.
Morten Egtvedt: Hi Ian! Well, I´m not talking about schedules, really. I like to keep it simple and usually only do continuous reinforcement until the behavior is really fluent.
When you say "like a computer" I think in terms of how much we talk to the dog during training. And generally I like to keep it quiet. I might talk to and praise the dog while playing with it, but before the click (or the treat) I keep my mouth shut and let the dog do the job. BTW, I´m a capturing/shaping-nazi, but I guess you already know that.
These are my comments to your vlog. Please excuse my English, hope you still understand most of it:
BTW, I disagree with those who say that you shouldn´t waste your time arguing with other positive trainers. I think you definitely should waste much more time doing just that! :-) Saying that lure/reward training is more effective than capturing is a fair opinion, and a great starting point for a debate!
We should be able to do two things at the same time: fighting punishment based trainers in the morning, and discussing training details with other positive trainers in the evening. That´s what brings us forward! I mean, how much do we evolve arguing with traditional trainers...?
I criticize other clicker trainers on a daily basis :-) And ironically, what I usually criticize is that they are NOT training like computers - they talk to much, explain too much, help the dog to much etc - instead of giving the dog clear information.
Looking forward to your next vlog!
I have not got any response to my latest answer yet, so I guess we agree on disagreeing on this particular topic. Fair enough :-)
You can read more about our style of clicker training in "Clickertraining: The 4 Secrets of Becoming a Supertrainer".
(Chief Instructor, Canis Clickertraining Academy)
Copyright © Canis Clickertraining Academy 2010
Address: CANIS, Vestre Rosten 78, NO-7075 Tiller, NORWAY