A study entitled “Conspecific observational learning by adult dogs in a training context” published in Applied Animal Behaviour Science got me thinking about age as it relates to social or observational learning in dogs.
First, a brief description of the study that began with 50 adult Labradors (33 males and 17 females) from the Italian School of Water Rescue Dogs (SICS). Equal numbers of dogs had basic or advanced training, and none of the them had had any previous training for the tasks chosen by the researchers. Dogs with basic training could sit, lay down, stay, and come on command; those with advanced training could maintain those same behaviors longer in areas with more distractions. All of the dogs were only-dogs and lived apartments with their handlers who spent at least an hour daily walking, playing with, or training their animals.That all of those folks lacked any special training skills immediately caught my attention because this describes many dog owners.
The problem presented to the dogs consisted of 2 phases. In Phase 1, the handlers asked their dogs to perform one of two novel tasks in 15 seconds: jump on top of a trunk roughly 26.2″ high and wide, or climb the 4-step stairs to the top of a 43.3” high children’s slide. Tasks were assigned randomly and handlers told to use whatever method they found worked the best for them and their dogs. If they succeed, they were eliminated from Phase 2.
Phase 2 included a demonstration dog, a 7-year-old female Lab with expert water rescue skills who had been trained to perform both tasks when her handler approached the object and stood beside it. When the dog completed the task, the handler gave her a treat. For the test, the leashed test dog and handler first observed the demonstration dog and experimenter. Approximately 30 seconds after the demonstration, the test handlers used the same approaches they used in Phase 1 to entice their dogs to display the demonstrated behavior. Of the 33 dogs who made it to Phase 2, 16 were assigned to the demonstration group. The remaining 17 made up the control group. Those in the latter group spent 30 seconds leashed in the same location as the test dogs.
Or the Cliff Notes version: the only difference between Phase 1 and 2 relative to the test dogs and their handlers was the presence of the demonstration dog modeling the behavior. The only difference between the test dogs and the control dogs in Phase 2 was that the test dogs saw the demonstration dog before performing the exercise, but the control dogs did not.
While the study offers several areas for future research, I found two results particularly thought-provoking. First, the probability of the dogs’ repeating the desired behavior increased significantly with the presence of the demonstration dog. The extensive studies of learning in wild animals combined with a growing number of them in multiple species of domestic animals doesn’t make this headline news to those who enjoy observing animals’ behavior in their natural settings. (For companion dogs, the “natural setting” could be the owners’ home, backyard, local park, or daycare.) On the other hand, the unwillingness to accept that animals don’t need humans to formally teach them everything does persist, including among some in the behavioral and training communities.
Given the role context plays in ethology, it’s possible that the demonstration dog provided the context that made it easier for the test dogs to grasp what their handlers wanted them to do.To get an idea of this difference, imagine that a teacher speaking a language you don’t understand wants you to jump up on a trunk or go up 4-step steps to the top of a child’s slide and wait there. It would be a lot easier to learn to do this if that person actually did what they wanted you to do. If would be easier still if the person who modeled the behavior also spoke your language. Barring this, unless the teacher is willing and able to teach you what those foreign words mean first, you’d need to repeat the exercise until you figure it out by trial-and-error. That’s not a particularly energy-efficient way to learn–especially if your survival depended on it as it most surely did when the brain’s system of learning evolved.
In addition to modeling the physical mechanics necessary to accomplish the desired behavior, the canine demonstrator also provides the test dog with information regarding the emotional stability of the environment. Previously I’ve mentioned how animals can pick up on their owner’s anxiety and fear. This represents an extension of the communication that normally occurs between animals belonging to the same species. However, the likelihood of a canine observer/recipient perceiving and correctly interpreting that information will be greater if it comes from another dog than a person. This occurs because the two dogs share the same perceptual reality whereas the dog and handler do not.
Also keep in mind that this phenomenon works both ways: dogs in the company of a poorly behaved de facto demonstration dog may pick up the other dog’s problem behaviors more readily than any good ones the owner attempts to teach the dog in that same environment.
Another interesting study finding was that the younger test dogs (14-24 months) were less like to learn from the demonstration dog 
compared to the older ones (15-72 months). Shouldn’t a younger dog be more likely to follow the lead of another, more experienced dog? The more I thought about this, the more possible reasons why this might not be the case occurred to me. For example, it’s not uncommon for the young of multiple species, including humans, to pay attention to the behavior of adults they perceive as parental figures. Especially for a solo dog living with a person, this would be that person… Provided that person interacted with the dog in a way that reinforced this kind of canine-human relationship.
On the other hand, suppose that person inadvertently or deliberately established a more reactive peer-based relationship with the dog. In that case, if the dog assumed a more protective stance relative to the person at maturity, then I’d expect the dog to become more attuned to other dogs than that person. Although this shift can lead to all kinds of behavioral problems in pet dogs, it could be advantageous for a working dog expected to function independent of human input. For example, it would be dangerous for a rescue dog to obey commands given by a hysterical person the dog was trying to save. Similarly, herding dogs working with large groups of livestock may be working closer to other dogs than any person.
Another possibly age-related question is related to the treat. What role does it play, if any? We know from other studies that dogs know when other dogs get something they don’t. This raises the possibility that the only reason the test dog is modeling the demonstration dog’s behavior is to get the treat. The only way to determine whether the older dogs were willing to learn from the demonstration dog vs willing to learn only for food would be to repeat the same experiment without food. Maybe someday…
First, a brief description of the study that began with 50 adult Labradors (33 males and 17 females) from the Italian School of Water Rescue Dogs (SICS). Equal numbers of dogs had basic or advanced training, and none of the them had had any previous training for the tasks chosen by the researchers. Dogs with basic training could sit, lay down, stay, and come on command; those with advanced training could maintain those same behaviors longer in areas with more distractions. All of the dogs were only-dogs and lived apartments with their handlers who spent at least an hour daily walking, playing with, or training their animals.That all of those folks lacked any special training skills immediately caught my attention because this describes many dog owners.
The problem presented to the dogs consisted of 2 phases. In Phase 1, the handlers asked their dogs to perform one of two novel tasks in 15 seconds: jump on top of a trunk roughly 26.2″ high and wide, or climb the 4-step stairs to the top of a 43.3” high children’s slide. Tasks were assigned randomly and handlers told to use whatever method they found worked the best for them and their dogs. If they succeed, they were eliminated from Phase 2.
Phase 2 included a demonstration dog, a 7-year-old female Lab with expert water rescue skills who had been trained to perform both tasks when her handler approached the object and stood beside it. When the dog completed the task, the handler gave her a treat. For the test, the leashed test dog and handler first observed the demonstration dog and experimenter. Approximately 30 seconds after the demonstration, the test handlers used the same approaches they used in Phase 1 to entice their dogs to display the demonstrated behavior. Of the 33 dogs who made it to Phase 2, 16 were assigned to the demonstration group. The remaining 17 made up the control group. Those in the latter group spent 30 seconds leashed in the same location as the test dogs.
Or the Cliff Notes version: the only difference between Phase 1 and 2 relative to the test dogs and their handlers was the presence of the demonstration dog modeling the behavior. The only difference between the test dogs and the control dogs in Phase 2 was that the test dogs saw the demonstration dog before performing the exercise, but the control dogs did not.
While the study offers several areas for future research, I found two results particularly thought-provoking. First, the probability of the dogs’ repeating the desired behavior increased significantly with the presence of the demonstration dog. The extensive studies of learning in wild animals combined with a growing number of them in multiple species of domestic animals doesn’t make this headline news to those who enjoy observing animals’ behavior in their natural settings. (For companion dogs, the “natural setting” could be the owners’ home, backyard, local park, or daycare.) On the other hand, the unwillingness to accept that animals don’t need humans to formally teach them everything does persist, including among some in the behavioral and training communities.
Given the role context plays in ethology, it’s possible that the demonstration dog provided the context that made it easier for the test dogs to grasp what their handlers wanted them to do.To get an idea of this difference, imagine that a teacher speaking a language you don’t understand wants you to jump up on a trunk or go up 4-step steps to the top of a child’s slide and wait there. It would be a lot easier to learn to do this if that person actually did what they wanted you to do. If would be easier still if the person who modeled the behavior also spoke your language. Barring this, unless the teacher is willing and able to teach you what those foreign words mean first, you’d need to repeat the exercise until you figure it out by trial-and-error. That’s not a particularly energy-efficient way to learn–especially if your survival depended on it as it most surely did when the brain’s system of learning evolved.
In addition to modeling the physical mechanics necessary to accomplish the desired behavior, the canine demonstrator also provides the test dog with information regarding the emotional stability of the environment. Previously I’ve mentioned how animals can pick up on their owner’s anxiety and fear. This represents an extension of the communication that normally occurs between animals belonging to the same species. However, the likelihood of a canine observer/recipient perceiving and correctly interpreting that information will be greater if it comes from another dog than a person. This occurs because the two dogs share the same perceptual reality whereas the dog and handler do not.
Also keep in mind that this phenomenon works both ways: dogs in the company of a poorly behaved de facto demonstration dog may pick up the other dog’s problem behaviors more readily than any good ones the owner attempts to teach the dog in that same environment.
Another interesting study finding was that the younger test dogs (14-24 months) were less like to learn from the demonstration dog
compared to the older ones (15-72 months). Shouldn’t a younger dog be more likely to follow the lead of another, more experienced dog? The more I thought about this, the more possible reasons why this might not be the case occurred to me. For example, it’s not uncommon for the young of multiple species, including humans, to pay attention to the behavior of adults they perceive as parental figures. Especially for a solo dog living with a person, this would be that person… Provided that person interacted with the dog in a way that reinforced this kind of canine-human relationship.
On the other hand, suppose that person inadvertently or deliberately established a more
reactive peer-based relationship with the dog. In that case, if the dog assumed a more protective stance relative to the person at maturity, then I’d expect the dog to become more attuned to other dogs than that person. Although this shift can lead to all kinds of behavioral problems in pet dogs, it could be advantageous for a working dog expected to function independent of human input. For example, it would be dangerous for a rescue dog to obey commands given by a hysterical person the dog was trying to save. Similarly, herding dogs working with large groups of livestock may be working closer to other dogs than any person.
Another possibly age-related question is related to the treat. What role does it play, if any? We know from other studies that dogs know when other dogs get something they don’t. This raises the possibility that the only reason the test dog is modeling the demonstration dog’s behavior is to get the treat. The only way to determine whether the older dogs were willing to learn from the demonstration dog vs willing to learn only for food would be to repeat the same experiment without food. Maybe someday…