Dog Training is Not Rocket-Science; But it is Science
But it is science.
However certainly not the Century-old, laboratory-generated learning theory from studies of computers dispensing food pellets and shock to caged rats. Obviously, those studies have little relevance to people training dogs and other animals off-leash at home or in parks. The reinforcement schedules are lame or difficult to implement or calculate, and punishment never deserved equal billing with rewards.
Shock was incorporated because computers couldn't give verbal guidance, "Now Ratty, just hop on the shelf when the buzzer sounds and then the electrified floor won't shock your paws." However, we can offer verbal guidance when dogs err. "Ratty, Jump".
As researchers disappeared down the rabbit hole of using shock on captive rats (that couldn't escape the shock), their creativity was blunted, and they were distracted from a more vibrant and productive quest for much more powerful rewards (life rewards) and considerably more effective reinforcement schedules.
Consequently, way back in 1972, in preparation for teaching a 10-week course on dog behavior (and training) for the UC Berkeley Extension, I took all that I thought was valuable from laboratory learning theory and combined it into a single Basic Training Sequence for teaching cued-behaviors to dogs, cats, horses, and other animals. See, How to Teach a New Dog Old Tricks.
The science of dog training is as simple as 1-2-3-4:
- REQUEST
- Lure (Handsignal)
- Response
- PRAISE (+maybe Reward)
This sequence combines the classical associative learning relationship between the verbal Request and following Lure (Handsignal), plus, the operant conditioning effect of copious Praise and occasional, unpredictable rewards reinforcing both the appropriate Response and the associative relationship.
Unlike most other reward-training techniques, Lure-Reward Training, as I called it, is a comprehensive training methodology for teaching animals the meaning of verbal cues and testing their comprehension plus motivation to respond, to produce an index of response-reliability of cued behaviors to evaluate the effectiveness of training.
Lure-Reward Training comprises three distinct stages:
1. To phase out food lures completely once the dog has learned the meaning of handsignals and verbal cues;
2. To drastically decrease the number of food rewards to increase their reinforcing power and to make room to use infinitely more powerful life rewards; and
3. To quantify the progressive increase of response-reliability percentages (compliance), using only non-aversive means.
1. Teaching Dogs What to Do
By watching and following the lure-movements, dogs quickly learn the meaning of different Handsignals, a language they naturally understand — body language. Handsignals may then be used as lures to quickly teach the dog the meaning of the words we use, which otherwise would be difficult and time-consuming. By effectively teaching ESL, our words may be used as instructions prior to task and especially, as verbal guidance when dogs err.
Essentially, Stage 1 involves using food lures to teach puppies and dogs which specific response is required following each verbal cue or handsignal, so that we can reinforce desired responses, but then going cold-turkey on food lures during the very first training session, once we have confirmed the dog's comprehension of cues and signals. Once dogs begin to grasp the meaning of handsignals and then verbal cues, food lures are unnecessary.
Progress, i.e., the effectiveness and speed of training is monitored by calculating Response-Reliability Percentages — a combined index of 'cue-comprehension' and 'motivation to respond', that indicates the percentage likelihood that the dog will respond following a single cue, in that scenario, i.e., the owner's control over their dog.
Reliability-testing is concomitant with training throughout all three stages. Puppies/dogs are tested with three repetitions of a Come-Sit-Down-Sit-Stand-Down-Stand position-change sequence (21 behaviors total), using a) Handsignals only, and b) Verbal cues only, in a variety of scenarios: off-leash in class, the kitchen, other rooms, the yard, and the dog park, and on-leash on sidewalks, and with different handlers. The variations in RR%s are truly enlightening and offer the best motivation for owners to train to surpass their personal bests.
2. Teaching Dogs Why Comply
Drastically reducing the number of food rewards substantially increases their reinforcing power. Or, to put it another way, offering too many food rewards decreases their reinforcing power. Once we escape from our knee-jerk, instant-food-reward-for any-and-every-response, reducing the frequency of food rewards is falling-off-a-log easy, by simply asking more for less, more and more short-duration responses (e.g., body position changes) for fewer and fewer food rewards. Reward after two responses, then after three, then four, five, seven, ten etc. For longer duration behaviors, we ask longer for less — longer and longer Watches or Stays, for fewer and fewer food rewards.
Progress is tracked by calculating Response:Reward Ratios for each session. Count out the number of food rewards beforehand, video the session, and then review the video to count the number of short-duration position changes. To give you an idea what to aim for: In Week 1 of puppy classes, most puppies can complete the position-change test sequence for a single food reward. (R:R Ratio = 7:1). Or, one of my favorite demonstrations, the average Golden puppy can perform 20 puppy-pushups (Down-Sits) for the prospect of a single food reward. (R:R Ratio = 40:1). Owners are motivated by competing with themselves (surpassing personal bests) and against other family members, or others in class. Are we offering far too many food rewards!?
The BIG benefit of reducing the number of food rewards is being able to replace them with much more powerful life rewards. For example, by integrating extremely short training interludes into dog-dog play, sniffing, walks, or interactive games, such as, Fetch, Tug, and Hide 'n Seek. Most training interludes may be as short as a single sit, others may be of longer durations, such as stays, or one-step Come-Sit and Heel-Sit sequences. Integrate a training interlude every 20 seconds or so, or every 25 yards when walking. Every time you interrupt an enjoyable activity for a short training session, you may use resumption of the activity as a reward to reinforce training. An added benefit is that each training interlude is performed in a unique scenario, so that your dog eventually generalizes comprehension and responsiveness to any and every scenario and all family members.
3. Enforcing Compliance Without Force
Phasing out food lures and reducing the number of food reward are essential to achieve high levels of response-reliability, otherwise, a dog's compliance might become contingent on the owner having food in their hand or on their person and then, off we go down the slippery slope of bribing, which as parents and politicians know, seldom works.
Whereas luring teaches a willing dog what you would like them to do, so you may reward them afterwards, bribing comprises trying to coerce an unwilling or unfocused dog to act against their will by offering the promise of a food reward upfront. (Duh!? Confuses me too; offering a Consequence as an Antecedent? I thought it was ABC. Behavior is changed by the Consequences that follow the behavior, not precede it.)
Once a dog has mastered even a small understandable vocabulary, non-compliance and misbehavior become easy to resolve. Non-compliance is dealt with by Repetitive Reinstruction until Compliance and then, Repeating the Entire Exercise, (as many times as needed), until the Dog Responds Following a Single Verbal Cue. Yes, if done right, 'repeating the command' causes learned relevance (not learned irrelevance), as evidenced by a gradual increase in response-reliability percentages from trial to trial, reflecting the dog's progressive increase in comprehension of verbal cues that are delivered in distracting circumstances, novel scenarios, and at a distance.
Misbehaviors are quickly resolved using Single-Word Instructions. Unlike leash-jerks or shocks, a single word, such as Sit, Settle, Shush, Outside, or Chewtoy, conveys two, (and maybe three) pieces of information (guidance): 1. Stop what you're doing; 2. Do this instead; and 3. This instruction is reeeally important right now.
The above two non-aversive procedures both 'act' as 'punishment' by quickly reducing the frequency of, and eventually eliminating undesirable behavior. They are the quintessence of Barking Up the Right Tree and will be the subject of several blogs and vlogs in this series. Simply put, Lure-Reward Training is the BEST! — The easiest, most effective, and absolutely, quickest way to teach ESL to a willing and engaged dog.
When dogs understand our language, the dog-human relationship improves in leaps and bounds.