A Dog’s Perspective: Generalization
I am often asked why dogs behave well in some situations but poorly in others, and what exactly can be done to allow them to demonstrate their good manners more consistently. You have undoubtedly heard me talk about the importance of taking the dog’s perspective into account when training. This is precisely what makes training simple, effective, and permanent. If we want to reliably teach a dog to behave a certain way or perform a particular behavior on cue, it behooves us to first look at how they view the world and exactly how they learn.
From the perspective of learning, generalization can be described as an extension of a concept (or behavior) from a familiar situation to a less familiar situation. To give an exaggerated example, a dog may understand the cue “Sit” and do so diligently when it is requested of him at home. Given that he has had lots of practice and has been rewarded for sitting at home in the past, this is a familiar situation. If this same dog were asked to sit outdoors, and had never been asked to do so in this environment before – there is a good chance he would not understand what is expected of him. The reason for this is that his behavior of sitting has not been generalized to the unfamiliar outdoor situation appropriately.
So why does this happen? Dogs are associative learners and the way that they acquire new knowledge depends on making connections between events and sensory cues. These cues are most often visual (hand signals, body language) and auditory (verbal). In addition, learning is driven by cause-and-effect, hence why creating a positive consequence through rewards is conducive to training. However, as dogs make decisions in their everyday life, such as whether or not to respond to a given cue – they take into consideration how similar or different the current situation is in comparison to what their store of past experience contains. By expanding this bank of experience – we increase the likelihood that we get a positive response, and thus elevate the dog’s generalization of the learned concept.
There are a myriad of things to consider when something isn’t working in training. In fact, I have a checklist that I like to run through which often points to the area that needs attention. Sometimes the issue is with setting the criteria correctly or working with the right type of reinforcement. However, quite often I have noticed that improper generalization is to blame. The mistake is common and quite reasonable. Dogs are incredibly intelligent creatures and when given consistent training become avid learners. It is therefore sensible and easy to underestimate how much generalization is truly needed.
Generalizing behaviors thoroughly is a critical aspect of training, and one that is easily overlooked by even the most experienced trainers. Since much of how well a dog learns on a given day can depend on unknowable factors (like how hungry or tired the dog may be), there is no shame in miscalculating this piece of the equation. In addition, I would like to point out that generalization does not necessarily pertain to ‘environment’, and can actually embody a number of different variations in the training situation. Here are some of them:
- Location: what room the training takes place in, the set-up of the furniture, the presence of other objects or distractions that define the environment. It is essential to practice in a number of different locations, preferably increasing the difficulty level incrementally. A busy park with all of its enticing sights, sounds, and smells is going to be far more challenging than the very familiar and unexciting living room. However, the driveway may be the perfect middle ground between those two extremes for some dogs. Also, keep in mind that you may need to lower the dog’s reinforcement criteria when you make the location more challenging to help him ‘catch up’. For example, if a dog has trouble focusing and responding to “come here” in a distracting park – you may need to make it easier for them and reward them at first for simply making eye contact (an example of this can be found in this video).
- The dog and trainer’s physical position relative to one another: dogs learn features of their training environment very specifically, and need to be taught to expand what they’ve learned. This is why it’s important to practice from different angles, body positions, and so on. Asking a dog to ‘sit’ while you stand in front of him or while siting next to them may be two very different things to a dog. In addition, make sure that you are giving your dog an adequate number of repetitions in each variation (and for some dogs this may be as much as 20-30 trials).
- Presence of training equipment: Whether you are using a clicker or a treat pouch is important information that a dog will remember. It is important to properly fade those tools out of the training, and/or practice with them set aside to prevent your dog from assuming they need to be present at all times and becoming dependent upon them. You will see examples of how to do this in almost all of my videos during the ‘fading’ portion of training.
Ultimately, generalization will become easier and more natural for a dog over time – assuming that they are given the opportunity to practice what they have been taught with a sufficient amount of variability and repetition. The real challenge of training a dog is committing to practice in real world settings, but if you want to have solid reliability, there is no substitute for it. Perhaps it is most important to be systematic and patient in your approach. Most trainers (including myself) desire to progress much faster than is optimal – and if there is one major lesson I’d like you to take away from this, it would be to work from the perspective of making small and gradual advances and taking the time to let an idea fully absorb before moving forward. After all, the proverbial journey matters most in developing a good relationship than simply arriving at your destination.