A number of studies have shown that dogs possess the ability to learn new behaviors and even problem solve by observing the actions of other dogs or humans. Several forms of social (aka observational) learning are defined by researchers. Discussions regarding dogs’ proficiency, the types of tasks that can be learned, and the relative cognitive complexity of different levels of social learning are ongoing (see “Missing the Point“, “Doggie See Doggie Do?“, and most recently “Wait, You Can Eat That?“).
When we look at feeding behavior and food preferences, however, there is not much available to suggest that dogs learn what to eat by observing other dogs (or humans). There is some evidence that puppies tend to like the same foods as their moms, but those studies did not specifically target the importance of observation. Although not technically a food preference (ick), the recent coprophagy study suggested that dogs may develop a preference for eating poop by observing this behavior in other dogs. This type of observational learning, called social facilitation, involves direct proximity and visual access between the dog who is learning to poop-eat and the other dog (the poop-eating instructor, if you will).
Social Transmission of Food Preferences
A related type of observational learning that animals use to decide what foods they should consume is called social transmission of food preferences (STFP). This type of learning refers to the transmission of information about foods solely through olfactory cues, rather than through proximity and visual cues. In other words, Stanley may learn what is good to eat by smelling Alice’s face and breath, rather than by actually witnessing Alice consume the yummy food. This differs from what we know about other forms of social learning in dogs, in which visual observation is the primary conduit to learning the new behavior or preference.
How does STFP Work?
The study of STFP has an interesting history. In the early 1980’s, some innovative (or perhaps bored) researchers decided to find out if Norway rats could transmit information to one another about sources of food via scent alone. To test this, they added a novel flavor, in this case either cinnamon or cocoa, to the rat’s food. One rat, who they deemed the Demonstrator (we will call him Dave), was separated from his partner rat and offered one of the foods.
After enjoying his flavored treat, Dave was reunited with his partner. The researchers labeled this rat the Observer (we shall call him Oli). Dave and Oli were allowed to have a happy reunion for 15 minutes. Oli was then removed and offered both foods. And guess what! Oli consistently chose the flavor of food that his pal Dave had previously enjoyed – even though this was the first time that Oli was presented with that particular flavor and even though he had not directly witnessed Dave eating the food. In repeated rounds of Dave and Oli experiments, the Oli rats consistently showed a preference for the flavor that their particular Dave had previously consumed.
Over the years, these results have been found to be robust, showing that Norway rats are quite proficient at STFP. It is speculated that this form of social learning benefits rats because it provides an effective way for individuals to learn from one another about safe vs. unsafe foods to eat in their environment.
What about Dogs?
In addition to rats, STFP has been demonstrated in mice, gerbils, rabbits, voles, and even in several primate species. However, only a single study, published in 2007, reported that dogs were capable of learning food preferences from other dogs via scent alone and in the absence of visual cues (1). In that study, the researchers selected the novel flavors of basil and thyme, which they added to the dogs’ food. Using the same protocol as the Dave and Oli tests, they reported similar results to the rat studies.
Recently, a new study attempted to replicate that work and also asked the additional question “Can dogs learn food preferences from their owners via STFP?” In other words, if you come home smelling like the burger that you just enjoyed at your local pub, would your dog prefer a burger as his meal that night?
Drs. Nat Hall and Armando Mendez, of Texas Tech University in the United States, conducted a set of three experiments (2):
- Experiment 1 – Owners and their dogs: A group of 24 dog and owner pairs were tested. Each owner consumed either blueberry- or strawberry-flavored oatmeal, out of sight of their dog. When reunited, the dogs and owners interacted for five minutes, during which time owners encouraged their dogs to sniff their face and mouth (Note: Owners who dislike dog kisses need not apply). Following kiss-time, the dogs were allowed to smell both foods in a standard two-bowl preference test. After “smell-time”, they were allowed to consume as much of each food as they desired. Results: In a nutshell, unlike rats, the dogs did not care which flavor of oatmeal their owners had previously enjoyed. They consumed amounts of each flavor that did not differ from chance (i.e. no preference). STFP was not demonstrated in this experiment.
- Experiment 2 – Shelter dog pairs: A group of 12 shelter dogs (6 pairs) were tested. Each pair was housed together, so the dogs were familiar to each other (i.e. similar to Dave and Oli). The jam flavored oatmeal test was repeated, using a dog rather than an owner as the demonstrator. Kiss-time and two-bowl preference tests were the same. Results: Once again, the dogs did not show a preference for the flavor that was just consumed by their dog friend. The observer dogs consumed amounts of food that were not significantly different from what they would have consumed by chance (or for my dogs, whichever oatmeal bowl was closer).
- Experiment 3 – Replication of the 2007 study: Given the lack of evidence for STFP provided by the first two studies, the researchers decided to carefully replicate the 2007 study that had reported food preference transmission between dogs. (For a discussion of the importance of replication in science, see “Thyroid on Trial“). Instead of using fruit-flavored oatmeal, they used dog food flavored with either basil or thyme and carefully repeated the methods used in the earlier trial. (Differences are addressed in the paper). Results: (Ruh roh). The Olis in this experiment did not prefer the flavor of the Daves. They basically chose in proportions that were not statistically different from chance, as in the first two experiments.
Take Away for Dog Folks
The researchers of this study concluded that dogs did not acquire a food preference (at least for strawberry versus blueberry flavored oatmeal) from either their owner or from another dog. Similarly, the researchers were unable to replicate a previous study that reported food preference acquisition between pairs of dog.
Practically speaking, if you eat sushi for dinner, it is probable that your dog will not be hankering for the same.
But wait! There is more!
The Value of Null Results
This set of three studies showed a lack of support for a specific type of social learning in dogs – developing a food preference based not upon seeing what another dog or human is eating, but rather based upon smelling what the dog or person had recently consumed.
This is an important distinction because we have substantial evidence of social learning in dogs when they are provided with visual cues. Examples include various forms of pointing, eye gaze, and observing another dog or human as they engage in a particular task. We know that dogs are proficient at attending to and understanding visual cues such as body postures, emotional expressions and eye contact. However, we also know that dogs have a highly developed and acute sense of smell and that they rely heavily upon olfactory cues as they go about their daily lives. For this reason, I was personally surprised by the results of these studies.
There are several explanations for dogs’ failings at STFP. One of the most salient is that the species in which STFP is most consistently demonstrated (mice and rats) usually live in groups, eat the same foods, and have experienced strong selective pressure for distinguishing between foods that are safe versus those that are unsafe. Perhaps this type of pressure has not existed for dogs to the same degree, or perhaps domestication-related changes have led dogs to rely more upon visual cues during observational learning.
Another explanation, and one that the authors address, has to do with the attraction of the novel flavors to dogs. It is possible that the novel flavors that were used in these experiments were not yummy enough (or perhaps did not differ enough) to make an impression on the dogs. My husband and I currently live with two Goldens and a Toller. The Goldens, Cooper and Alice, like many of their clan, live by the motto “Eat it first, hope that it stays down, adjust behavior accordingly if it does not“. If they see Mike or I eating, they always, always, always, want a bite (actually, they want all of the food). So, perhaps a difference was not found simply because many dogs do not have very discriminating (refined?) tastes and in this way may differ from those species for whom food selection was evolutionarily more important for ultimate survival.
Regardless, this set of studies gets high marks for science – for reporting null results (which some researchers are loathe to do), for a well-designed and well-controlled set of experiments, and for adding to our collective knowledge regarding how dogs learn.
What about your dog(s)? Do they pay attention to what you are eating (or recently ate)?
Do you think your food choices influence what your dog does or does not like to eat?
Share your thoughts!
- Lupfer-Johnson G, Ross J. Dogs acquire food preferences from interacting with recently fed conspecifics. Behavioural Processes 2007; 74:104-106.
- Mendez A, Hall NJ. Evaluating and re-evaluating intra- and inter-species social transmission of food preferences in domestic dogs. Behavioural Processes 2021; 191; doi.org/10.1016/j.be