Here’s a question for all of you dog trainers (and that should be just about everyone who has a dog)… how often should you train a dog? Many of us in this business would, of course, say, “as often as possible… please!” But that is not quite what I mean: from a scientific point-of-view, what is the optimum frequency of dog training… once a day, once a week, once a month? Again, many of us would answer, “as frequently as possible, within the attention span of our dog.” But surprisingly perhaps, there is very little information in the scientific literature about the optimum frequency for training, especially for dogs.
A few trainers like Bailey (1995) and Abrantes (2000) have provided some guidelines, generally “from once to several times a day” but provide no scientific evidence for such statements, which seem to me to fall into the “well, the more often, the better seems to make sense” category. And this issue would be a challenging one to test experimentally because most such research is done with dogs from widely varying rearing and training backgrounds, so that very large sample sizes would be required to determine accurately the answer to such a simple question.
In other species, like rats, horses and humans, there is some evidence that in almost all cases (with some interesting exceptions), more widely spaced training (say, weekly training sessions) are more effective (that is, fewer training sessions are required) than “massed” training sessions (multiple/day). The exceptions have to do with training certain behaviors in yearling horses, and in the decline versus the reoccurrence of fear behavior in humans, where prevention of reocccurrence is decreased by spaced training.
So what about dogs? In a recent paper in the journal Applied Animal Behavior Science, Dutch researchers Iben Meyer and Jan Ladewig tested laboratory beagles, all raised in a very similar fashion (thus removing the differential rearing confounding variable from the mix), on a simple standardized learning task. This task, broken into four stages, was to move one meter from the trainer and touch a pad on the floor with a front paw. The trainer used a clicker as a secondary reinforcer and a food treat as a primary reinforcer, very much as we do in a lot of our behavior counter-conditioning work. Eighteen beagles were divided into two equal groups: one group trained once per week, and the other trained for five days per week. So which group learned the task, to 80% accuracy, in the least amount of time, and which group learned the task in the fewest number of training sessions?
Well, somewhat surprisingly to many trainers and owners, the once-per-week trainees learned the task equally well but in signficantly fewer training sessions, an average of 6.7 sessions, than the 5-times-per-week group, who achieved proficiency in an average of 9.0 sessions. But if the difference in training rate is five-fold (once per week to 5-times-per-week), that should mean that the more-frequent-training group took fewer total days to learn their task: more training sessions but fewer days overall. That was indeed true: once-a-weekers took a total of 46.7 days and more frequent trainees took only 11.4 days, on average.
So this work reinforces, in dogs, some of the basic conclusions that have been demonstrated in rats and horses. It eliminates a huge potentially confounding rearing background problem by using one breed all reared in the same way, but of course, tells us nothing about possible genetic differences among breeds or the longterm implications: do dogs trained over a longer time period (in fewer training sessions) retain that learning better or worse than dogs trained more intensively? We would suggest from limited information from other species (but using a model of common evolutionary descent among mammals) that spaced training should produce longer, better retention of learned material. But we will have to wait for the research study to be done and published to know for sure.