Is your dog 100% reliable? Does he always come when called, no matter what? Could you train a dog to avoid a snake reliably? In Part 1 [Array] of this article series we looked at what is necessary to maintain or increase operant behavior - namely, a schedule of reinforcement. It is relatively easy to train a behavior under a set of unchanging circumstances, put it on a schedule of reinforcement and keep it "reliable". If you only ever ask your dog to do something at home and you use a schedule of reinforcement, then your dog will probably do it 95% of the time. Never 100%, but 95% is a realistic goal in an unchanging environment. ...Dogs can be off their game. They might feel sick, or tired. They might be injured. They might have learned something new recently and become confused. ...We might be off our game. Our cues might be inconsistent. We might have been upset earlier in the day and our dog is avoiding us or trying to appease us. ...Something in the environment might have changed. A new dog might have moved in next door, or a new cat. The weather might be different, there might be a really low pressure system moving over, or it might be ridiculously hot (this happens more often than you might think in Tasmania, the coldest place in Australia). Any time the picture changes, our "stimulus conditions" change. We tend to think of cues as being the single cue we use to elicit behavior, but cues are more involved than that. Dogs pay attention to things like our posture, our tone of voice, and other things in the environment. Your cues should be "cookie cutter" consistent to achieve maximum reliability. For life-saving behaviors or competition behaviors it's worth practising your cues. It's worth actually recording yourself either on audio tape or video to review how consistent your cues are. The more you practise using consistent cues, the more likely it is you will use a consistent cue when the chips are down and you're shocked or frightened! "Generalisation" is the phenomenon of behavior occurring despite changes in the environment. If we teach a dog to sit in the kitchen, then move to the dining room it is a miracle if the dog still responds correctly to the cue. That miracle is called "generalisation" and we mustn't take it for granted. If our dog can sit in every room of the house and we go outside and ask him to sit, we mustn't expect him to understand. Being outside is a completely new "stimulus condition" not associated with sitting on cue. We need to take a step or two back and set our dog up for success. Generalisation can be helped along by setting our dogs up for success in a wide variety of likely scenarios. Some people call this "proofing". In The Positive Petzine Collection [1] there is a chapter devoted to dealing with distractions and proofing with positive reinforcement. If you want your dog to come when called, even when a strange cat runs across his path, you need to train for that. Even dogs who don't chase cats are likely to fail their recall if a cat suddenly dashes across their path and they haven't been systematically prepared for it, continuously being set up for success and reinforced along the way. When can we call a behavior "reliable"? That depends upon our expectations or our requirements. I expect that my dogs will settle themselves down to bed every night so long as I only ask them in the kitchen at home with no visitors. They are reliable under those circumstances. If someone was kicking down my back door and a brass band was playing dixie by my front door, I wouldn't expect they would be very reliable at all! Thankfully where I live snakes only come out at certain times of year. We avoid walking off-lead in areas we know there will be snakes. I do not expect my dogs to be reliable around snakes. My requirement is that they don't get bitten and killed. If I had ready access to snakes and a lot more time on my hands I'm certain that I could train my dogs to reliably avoid snakes, but there is a big gap between the reliability I require and the time it would take to train this. More on that next time...