Evidence Suggests Dogs Prefer Free Food to Contrafreeloading

There is no experimental evidence to support the claim that dogs prefer to work for their food. In fact, there is evidence that they don’t.

Contrafreeloading is the behavior shown when an animal prefers to work for food (“earned food”) rather than take identical food that is freely available from another, nearby source (“free food”). — Inglis, 2022

A recent study found that statistically, the group of dogs they tested did not prefer contrafreeloading. They were willing to do it. I’ll discuss the results of the study below, including their definitions of “prefer” and “willing.”

But it’s just as important to learn about contrafreeloading findings in general. Studies on contrafreeloading have been done with many species. The studies have had large variability in results (Lentz & Cohen, 1980). Contrafreeloading is far from a universal preference; it appears more to be situational. And the many studies have yielded patterns that can inform us of situations in which this interesting phenomenon is more or less likely to happen.

Highlights of the Research

The seminal Jensen study (1963) is often cited to support contrafreeloading generally, including in dogs. In that study, rats preferred to press a bar to get food rather than eat it out of a dish. What many people miss are the full results of the study and the experimental follow-ups. Jensen’s rats contrafreeloaded in proportion to their reinforcement history with bar pressing—they worked for food more when they had built a habit. In 1972, Taylor performed an experiment similar to Jensen’s, but the rats in his study showed a strong preference for the free food, and an even stronger preference for free water. Taylor had more preference tests after the initial bar pressing training (Jensen had only one preference test), and contrafreeloading decreased over those sessions as time passed from the initial training. Lentz and Cohen (1980) worked with pigeons to investigate whether previous training with the contrafreeloading method of obtaining food had an effect. They also found that response rates in the presence of free food increased as a function of previous training.

“The data suggest that amount of training affects contrafreeloading and that this variable may be partly responsible for the large variability in responding observed among studies examining responding in the presence of free food.” — Lentz & Cohen, 1980, p. 48

The research continued, and scientists discovered many other factors that could influence whether animals chose to contrafreeload. A review paper in 1997 (Inglis et al.) compiled from previous studies seven items that could increase or decrease contrafreeloading . They used them to create a fuzzy logic model of contrafreeloading. Here are the seven items they considered. Supporting references for each of these are in the Inglis paper.

• Prior training. A reinforcement history with the method of accessing the food increases contrafreeloading.
• Deprivation level. This has both increased and decreased contrafreeloading in studies, depending on other variables.
• Effort required. The larger the effort needed to access the food, the less likely animals are to contrafreeload.
• Stimulus change. A change in the environment that correlates with the presentation of the earned food increases contrafreeloading. This might be a light or a sound. (Hmm, clicker training, anyone?)
• Environmental uncertainty. Novelty, ambiguity, and surprisingness increase contrafreeloading. For instance, animals are more likely to contrafreeload if that food source appears at an unpredictable location.
• Rearing conditions. Being reared under conditions of sensory deprivation increases contrafreeloading.
• Manipulation of the environment. Being able to control the environment may be a factor that increases contrafreeloading.

The list above comprises factors that can affect whether animals contrafreeload in different situations. But they do not explain why. One well-supported hypothesis is that the food itself is not the only reinforcer for food seeking behavior and eating. This “information hypothesis” notes that animals gain information about their environment when they contrafreeload. Such information could be helpful when seeking food in the future. This hypothesis has been borne out in studies.

An earlier review (Osborne, 1977) looked at contrafreeloading experiments through a behavior analysis lens and analyzed the reinforcement schedules identified in the studies. Osborne was apparently the first to note the effects of “stimulus change,” one factor listed above. After reviewing the research up to that point, Osborne states, ” . . . when the consequences of responding for food or obtaining it freely are made equally reinforcing (e.g., by equating stimulus change conditions for both alternatives), animals show a near exclusive preference for obtaining their food freely.”

Scientists have noted one other trend that supports the information hypothesis: domestic species tend to contrafreeload less. This is thought to be because their needs are taken care of by humans, and they don’t have a pressing need to seek information as wild animals do.

The Dog Study

The study of contrafreeloading in dogs (Rothkoff et al., 2024) is freely available online and very readable. Since the results were that dogs were “willing” to contrafreeload but didn’t “prefer” it, it’s important to learn how the scientists defined these terms.

Here’s how the experiment was set up. There were four “acclimation” sessions where the dogs gained experience with eating from a snuffle mat and became familiar with a setup of a snuffle mat and a tray of “free” food. After the acclimation sessions, there were 10 experimental sessions. In these sessions, half of the dogs’ meal was in the snuffle mat, and half in the tray of food. All acclimation and experimental sessions were video recorded.

A preference for contrafreeloading was determined by measuring how often the dog went to the snuffle mat first. A willingness to contrafreeload was determined by measuring how often the dog approached and interacted with the snuffle mat before finishing the food in the tray.

Results under these definitions were:

Preference. Only one of 38 dogs preferred to contrafreeload, going first to the snuffle mat 80% of the time. Twenty-two dogs went first to the tray for a large majority of the time, and fifteen dogs did not show a strong preference.

Willingness. Thirty of 38 dogs were “willing” to contrafreeload. Eight were not willing.

We can consider the characteristics of snuffle mats in view of Inglis et al.’s seven factors affecting contrafreeloading. Compared to many food-based enrichment toys for dogs, snuffle mats are low effort. They do not require the skill of puzzle feeders or frozen Kongs. I find it interesting that even with a comparatively easy “earned food” opportunity, most dogs preferred the free food. But mats also lack a stimulus change—there is no bell or click when the dog grabs a piece of food from the mat. That absence could make contrafreeloading less reinforcing.

There are two other interesting findings in the study: dogs’ activity level did not correlate with a preference for free food or contrafreeloading, but dogs with higher body condition scores (more overweight) were more likely to contrafreeload.

Finally, there is a thoughtful listing of the limitations of the study at the end of the writeup.

Enrichment

Where does this leave us with enrichment? In dozens of online articles on enrichment for dogs, you will see contrafreeloading referenced. You will see such absolute and non-evidence-based statements as:

• Dogs have a behavioral drive to earn their keep
• All dogs are contrafreeloaders
• Dogs will work for their food while passing up the same food for free
• Dogs would rather work for their food than not

The message is sometimes more nuanced, especially in the last couple of years. But you can still find plenty of articles that present contrafreeloading as an unquestioned attribute of dogs. Moreover, contrafreeloading is presented as a justification for enrichment activities that involve dogs working for food.

How this information about dogs and contrafreeloading might affect our enrichment choices is beyond the scope of this post. But here are some resources.

I have written about assessing whether a dog really enjoys a food toy, rather than assuming they do because they are “supposed to.”

Linda Case at The Science Dog has a comprehensive blog post about studies that assess whether food toys have beneficial qualities for dogs.

And here are three studies that might inform our choices about enrichment. The Breakfast Effect study (Miller & Bender, 2012) showed evidence that dogs, just like people, probably learn better when their stomachs are not empty. The Eureka Effect study (McGowan et al., 2014) provided evidence that solving problems can make dogs feel good. Finally, a fascinating paper in 2024 (Veissier et al.) suggested that opportunities for animals to gain information should be central to designing enrichment activities. This paper has a wealth of information that can help us think about enrichment in new ways.

A Personal Note

I use food toys. This study and the rest of the contrafreeloading research do not discourage me from offering them in moderation. Here are some things I do and don’t do with food toys.

• I do use food toys with dogs with cognitive dysfunction (easy ones).
• I do look for toys and activities where the dog’s enthusiasm seems much greater than the value of the food involved, as in searching games.
• I do use food toys with easy-to-access, high-value food for delay counterconditioning or just to sweeten difficult situations.
• I do help my dogs with those last stubborn pieces at the end if they ask me to.
• I don’t give a dog a challenging food toy on an empty stomach, nor do I train in that situation.
• I don’t ask a dog to work for food when they are sick or have a low appetite.
• I don’t ask a dog to work for food in a way that stresses them out. For instance, treat scatters can stress some intense resource guarders. Similarly, I don’t do food chase games with dogs who gulp it frantically.
• I don’t ask a dog to interact with a toy they find un-fun. Lewis has told me he finds Kongs, especially frozen ones, pretty “meh.” I give him Westpaw toys instead.

The research with dogs is only one study, with identified limitations. More research needs to be done. But with the power of the rest of the research behind it, the study is compelling. Scientists have been studying the “when” and the “why” of contrafreeloading for decades. It’s time for us to consider those findings in the dog training world.

Copyright 2025 Eileen Anderson

Related Posts

References

• Inglis, I. R., Forkman, B., & Lazarus, J. (1997). Free food or earned food? A review and fuzzy model of contrafreeloading. Animal behaviour, 53(6), 1171-1191.
• Inglis, I. R. (2022). Contrafreeloading. In Encyclopedia of Animal Cognition and Behavior (pp. 1665-1670). Cham: Springer International Publishing.
• Jensen, G. D. (1963). Preference for bar pressing over “freeloading” as a function of number of rewarded presses. Journal of Experimental Psychology, 65(5), 451.
• Lentz, B. E., & Cohen, S. L. (1980). The effect of prior training on the contrafreeloading phenomenon. Bulletin of the Psychonomic Society, 15(1), 48-50.
• Lindqvist, C., Schütz, K., & Jensen, P. (2002). Red jungle fowl have more contrafreeloading than white leghorn layers: Effect of food deprivation and consequences for information gain. Behaviour, 139(9), 1195-1209.
• McGowan, R. T., Rehn, T., Norling, Y., & Keeling, L. J. (2014). Positive affect and learning: exploring the “Eureka Effect” in dogs. Animal cognition, 17, 577-587.
• Miller, H. C., & Bender, C. (2012). The breakfast effect: Dogs (Canis familiaris) search more accurately when they are less hungry. Behavioural processes, 91(3), 313-317.
• Osborne, S. R. (1977). The free food (contrafreeloading) phenomenon: A review and analysis. Animal Learning & Behavior, 5(3), 221-235.
• Rothkoff, L., Feng, L., & Byosiere, S. E. (2024). Domestic pet dogs (Canis lupus familiaris) do not show a preference to contrafreeload, but are willing. Scientific Reports, 14(1), 1314.
• Taylor, G. T. (1972). A limitation of the contrafreeloading phenomenon. Psychonomic Science, 29(3), 173-174.
• Veissier, I., Lesimple, C., Brunet, V., Aubé, L., & Botreau, R. (2024). Rethinking environmental enrichment as providing opportunities to acquire information. animal, 101251.

Copyright 2025 Eileen Anderson

Image Credits

• Maned wolf from Wikimedia Commons, user Jonathan Wilkins , under this license.
• Red junglefowl from Wikimedia Commons, user Charles J. Sharp , under this license. Modification: I cropped the photo.
• Images of Zani on the snuffle mat and Clara playing with the flirt pole copyright Eileen Anderson.