top of page

Key components of any proofing or generalization program

Proofing and Generalization are among the most important training disciplines every trainer needs to know, practice and teach to both clients and dogs.

Proofing is the process of undergoing repetitions of training exercises in a way that communicates the details of how, when, and where you want your dog to behave - particularly, which aspects of the behaviour or skill you want generalized and certain, and which aspects may have more flexibility. The key to good proofing is to hold some aspects of your training consistent, and other parts varied over time.

Training is good preparatory learning, but to be sure your dog will behave in a certain way in a certain situation, you need "proof" -- in this case we are looking for evidence that would suggest the dog will behave as planned -- and the best evidence to look for in most cases is exposing the dog to increasingly accurate simulations of the situation in question. In my mind, this is why it's called "proofing". First you prepare by training in a simulated situation and then you "prove" that prep training by gradually exposing the dog to the real-life situation. You don't put strict performance expectations in place until the dog proves it can perform them reliably in various situations with varying degrees of management by the owner and freedom for the dog.

What should be proofed first?

I generally recommend the skill-lure, hand-signal and vocal cue be generalized first. For example, let's assume we want to teach a dog 'sit' and 'down'. Please note, this is also related to one of the most frequent questions I get, "why isn't it recommended that you say 'sit' the first time you want the dog to sit?" The answer is, because the dog doesn't know what you want it to do yet. This is why you teach and generalize the skill-lure first, before teaching the vocal cue.


SIT - The skill-lure for 'sit' is to pinch a kibble, show it to the dog's nose, and then lift the treat up and over the dog's head, to hovering about 10cm above the back of the dog's head. Most often, dogs will sit to try to maintain a view of the kibble. Then you reinforce with voice ("good dog!") and give the kibble.

DOWN - The skill-lure for the down is to pinch the kibble, reach towards the dog's nose as if to give the treat, but then as you get 25% of the way, you lower the treat down to the ground and hold it down under your finger. Once the dog lays down, you release the kibble.

In essence, the skill-lure IS the hand signal.

Hand Signals:

SIT - The hand signal for the 'sit' is the same motion you made with the kibble, but without the kibble. Once the dog sits, you grab a kibble from the pouch and give it to the dog.

DOWN - The hand signal for the 'down' is the same motion you made with the kibble, but without the kibble. Once the dog downs, you grab a kibble from the pouch and give it to the dog.

Vocal Cues:

I find it is important for me to try to teach a dog where, when and how to do things silently using skill-lures, hand signals, body language, and known cues only. That means I teach vocal cues last -- only after the dog knows exactly what I expect. By working this way, it helps the vocal cue to stand out loud and clear once you do start to use it.

I find the best way to condition a vocal cue is to say the cue shortly before cueing the desired behaviour with the hand signal. Note, this technique uses prediction error to help teach the vocal cue. Usually, no more than 50 reinforced repetitions are needed for the dog to establish a nascent conditioned response, although a further couple of hundred reinforced repetitions (as well as further reinforced repetitions in the future) on voice cue only are generally needed to ensure a lasting conditioned response to the vocal cue.

Further proofing than this is certainly necessary for these skills, some of the most important your dogs will ever learn -- at least in terms of frequency of use, if not in many other aspects. So what would that further proofing look like?

The most common types of proofing include:

Treat value, treat frequency,

Cue presence ("will the dog do the behaviour un-cued?"), cue volume, cue type, cue distance, cue discrimination ("can the dog distinguish one cue from another and behave appropriately?")

On/off leash, leash length, holding/not holding the leash, distance from end of leash to owner

Distraction type, distraction volume, distraction distance,

Handler type, handler volume, handler distance, handler interactivity, handler assertiveness, handler intervention

Handler-initiation, dog-initiation

Time, place, context, environment, objects present, individuals present or interacting, the state of the dog (hungry?, exercised?, anxious?, having a good week?)

I recommend covering off most or all of these proofs before signing off on any professional dog like a service dog or scent-detection dog, or any dog you want to make certain will behave in the way you trained it to, but the details of each of the individual proofs mentioned above would also depend heavily on the application the dog is being trained for. NB, that the industry is moving more and more towards covering this list as a minimum standard.

The more different proofing types you use, the greater the generalizability of the behaviour. NB any conditioning or training program must change and evolve over time. There is no such thing as absolute consistency in routine, nor in leadership, nor in obedience, nor in behaviour.

If you don't know what conditioning, or how much conditioning your dog has around any one or more of these factors, then you risk running into problems eventually. It's far better to know exactly how many hours of training and a ballpark number of repetitions for any given skill -- and better to know that the quality of the conditioning is both high and reliable, than to wonder if you and your dog will run into a situation you won't be able to handle as a team. Once you start trying to whittle away at these minimum standards, you start to let variability in the behaviour and conditioning start to creep in.

If you have any questions about proofing or generalization, please contact us!


Featured Posts
Recent Posts
Search By Tags
No tags yet.
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square
bottom of page