Dogs are well known to be chowhounds. The idea that they love food more than anything else is practically (excuse the expression) dogma in the fields of canine behavior and dog training. The trouble is, recent research suggests that it is not true for all dogs.
In a study called “Awake Canine fMRI Predicts Dogs’ Preference for Praise Versus Food” scientists investigated whether dogs prefer treats or praise, and whether their choice can be predicted by their brains’ response to both stimuli. In one experiment, they measured the level of activation of the brain’s ventral caudate, an area known to function as a reward center, in response to items that predicted various outcomes. A toy car predicted that verbal praise was coming, a toy horse predicted that food on its way and a hairbrush was associated with nothing. Dogs were trained to make these associations with a series of 40 pairings of each object with what it predicted. The activation of the specific region of the brain was measured with functional Magnetic Resonance Imaging (fMRI), which is possible because the dogs in the study have all been trained to remain motionless while in the scanner.
The average activation of the reward center of the brain was higher in the food and praise conditions than in the neutral condition, which shows that the dogs did learn the associations between the objects and what the objects predicted. (Each dog’s responses in the brain to seeing the toy horse and NOT receiving the expected praise was also measured.) There were 15 dogs in this experiment, and most of them had a similar response in the reward center to the food or to the praise. Four showed a stronger response to praise and two showed a stronger response to food. The average response to praise and to food did not differ.
In another experiment, dogs were placed in a Y-maze and given the opportunity to choose which arm of the maze to go to. One arm led to a food bowl with treats and the other arm led to the dog’s guardian, who provided petting and praise. Each dog was tested in the Y-maze 20 times. Seven dogs in the study chose the guardian the more times than the food, and seven dogs chose the food more often. One dog chose the guardian and the food an equal number of times.
The relative value of praise versus food in the first experiment was highly predictive of the choices that dogs made in the Y-maze experiment. Dogs whose ventral caudate showed a strong response to praise were more likely to choose their guardian over food but dogs who did not show such a strong response to praise relative to food were more likely to head for the food when given a choice.
Regrettably, the results of this study have erroneously been reported in many places as proof that dogs prefer praise and belly rubs to treats, and suggested that using treats in training is therefore unnecessary. It has been written in many places discussing this study that 13 of 15 dogs prefer praise to food, and that’s not correct. What the researchers actually wrote is that in 13 of the 15 dogs, the ventral caudate showed either roughly equal activation to food and to praise or greater activation to praise than to food.
It’s quite interesting that roughly half of the dogs chose their guardian over food. For those dogs, social interaction such as praise and belly rubs may be more effective than treats in training. However, caution is important when acting on the findings in this study because the research may overestimate the response of dogs to their guardians relative to food in situations outside the laboratory setting.
The lab may have been stressful, causing a bias in dogs towards an increased interest in their guardians when compared with food. They may have been seeking comfort from their guardians in a way that they might not be during typical training situations. The scientists do point out that these dogs have been trained to stay still in the scanner and that the lab is a familiar environment. That does not mean the dogs are as comfortable as they are at home or in other areas such as on neighborhood walks, at the park or at the training center where they attend classes. It’s important to know what dogs choose in the actual training setting before changing what reinforcement to use based on lab research.
Additionally, although dogs may value social connections over food when the social interaction is with their guardian, not all training occurs between guardian and pet. I do a lot of training with dogs who I adore, but I don’t share quite the same bond with them as they do with their own guardian. So, just because dogs may prefer affection from their guardian over food does not mean that they prefer affection from just anyone over food. Finally, in many training scenarios, dogs receive praise in addition to food during training, and that may be more effective than either one alone.
Many people swear that their dogs prefer praise and petting to treats, and others are just as certain that food wins out every time with their dogs. Perhaps the most important lesson from this study is that individual variation in preferences is huge. If you feel strongly about what matters most to dogs, there’s a good chance you’re right—when it comes to your dog, anyway.
Do you think your dog would go for food or for praise and affection if given the choice?