fMRI in Dogs

We can learn a lot about animal cognition by using behavioral experiments, where we rely on the actions of an animal in a particular situation to infer what processes are occurring in the animal’s brain. But what about looking more directly at what the brain is doing?

One way to do that is to record the activity of neurons in the brain using electrodes. There are a few downsides to this method, though. First, neural recordings can only show us the activity of very small areas of the brain at a time. Additionally, the animals used in these experiments have usually been raised in a lab, rather than in their natural environment, which could affect their cognitive development.


MRI Machine

Another way to look at brain activity is through imaging such as functional Magnetic Resonance Imaging, or fMRI. The MRI machine uses a magnetic field to look at oxygen flow to different areas of the brain. More active brain areas need more oxygen, so by tracking the oxygen flow to the brain over time, we can determine the pattern of activity in the brain. And, unlike with neural recordings with electrodes, we can look at the activity of the entire brain at once.


Unfortunately, fMRI requires that the person (or animal) being scanned lie completely still inside the MRI scanner, which is a bit confining and noisy. While this isn’t a problem for anesthetized animals, it’s nearly impossible for awake animals. But recently, researchers have made incredible progress in using fMRI to study dogs.

In 2012, a group of researchers managed to train two dogs to lie still in an MRI machine long enough to have their brains scanned. They first got the dogs used to laying with their chins in a chin rest, then slowly acclimated them to the noise of the machine, ear muffs (to muffle some of the machine’s noise), and the MRI machine itself. While in the scanner, the dogs were shown two different hand signals: one was followed by a food reward (“reward”), while the other was not (“no reward”).

The researchers were interested in the response of an area of the brain called the caudate nucleus. Among other important functions including voluntary movement and learning, the caudate nucleus in involved in the reward system of the brain; previous studies have found that the caudate becomes more active when a reward is expected.

2241616932_107199f0fa_mSure enough, the researchers found that the dogs’ caudate nuclei were more active when the “reward” hand signal was given than when the “no reward” signal was given. They later replicated this study with 11 more dogs and found that the increase in activity in the caudate was comparable to that found in humans. Interestingly, some of the dogs studied were service dogs, and they had nearly significantly greater caudate activation than the non-service dogs when the “reward” hand signal was given. The researchers theorized that, due to their extensive training, the service dogs may have found the hand signal itself intrinsically more rewarding.

(Check out the Supporting Information for the first paper for an interesting video detailing the training that the dogs underwent!)

Before these studies, fMRI studies with animals required that they be restrained or anesthetized, both of which could greatly affect brain activity. Believe it or not, researchers actually ran studies investigating brain activity in response to odors…in anesthetized dogs! While the brain does respond to such stimuli even under the influence of anesthesia, it clearly does so in a different way than it would normally. (Of course researchers were aware of this, but there just wasn’t another option at the time.)

Once it had been shown that dogs could be trained to lie still in an MRI machine, some researchers decided to see just how different the brain’s response to odors is in anesthetized and awake dogs. They found that, in addition to the sensory brain areas active in anesthetized dogs, the frontal cortex was activated by odor in awake dogs. (The frontal cortex is implicated in complex cognitive functions like decision-making and planning.) The researchers suggested that the frontal cortex activity could be involved in understanding what an odor means in a particular context, and could inform how the dog behaves in response to that odor. Understanding this kind of brain activity could have major implications for how we train drug- and bomb-sniffing dogs.

443627085_4fe6f95f01_mThe first group of researchers also studied dogs’ brain activity in response to odors. They presented dogs with five different odors: a familiar dog, an unfamiliar dog, the dog itself, a familiar human, and an unfamiliar human. They found that, compared to the other four odors, there was greater activity in the caudate nucleus in response to the familiar human odor. This indicates that not only could the dogs discriminate between the different odors, but that they also associated that particular human with reward. And again, the researchers found that the response was stronger in service dogs, possibly due to their more extensive human contact during training.

Finally, another group of researchers compared the brain activity of humans and dogs as they were listening to human and dog vocalizations and natural sounds. They found that similar areas of the brain were more active when they heard the vocalizations of conspecifics (i.e. when humans heard humans and dogs heard dogs). This suggests that these brain areas may have evolved for that particular purpose more than 30 million years ago (although convergent evolution is a possibility). Additionally, some brain areas responded more strongly to more positive vocalizations (i.e. laughing versus crying). These brain areas were in similar locations in both dogs and humans, and responded to both species’ vocalizations. This suggests that dogs and humans may have comparable emotional processing of sounds.

Interestingly, part of the training of the dogs used in this study involved social learning – naïve dogs watched an experienced dog get in the MRI scanner and receive praise and treats, which motivated the naïve dogs to behave in the same way!

(Although the paper of this study isn’t freely available, the researchers did make a fascinating video summarizing their study, and NPR wrote an article about it.)

4840054871_b79aef16f2_mStudying canine cognition using fMRI is a relatively new application of the method, but the research so far looks promising. Not only will it allow us to get a direct look at what’s going on in dogs’ brains, but it could also improve the way we train service dogs and care for our best friends.


I’ll end with a thought-provoking quote from one of the above papers, about canine cognition research:

“…While the study of the canine mind is fascinating for its own sake, it also provides a unique mirror into the human mind. Because humans, in effect, created dogs through domestication, the canine mind reflects back to us how we see ourselves through the eyes, ears, and noses of another species.”

-Berns et al. (2012)


Sources Cited:

Andics, Attila, et al. “Voice-Sensitive Regions in the Dog and Human Brain Are Revealed by Comparative fMRI.” Current Biology (2014).

Berns, Gregory S., Andrew M. Brooks, and Mark Spivak. “Functional MRI in awake unrestrained dogs.” PloS one 7.5 (2012): e38027.

Berns, Gregory S., Andrew Brooks, and Mark Spivak. “Replicability and Heterogeneity of Awake Unrestrained Canine fMRI Responses.” PloS one 8.12 (2013): e81698.

Berns, Gregory S., Andrew M. Brooks, and Mark Spivak. “Scent of the Familiar: An fMRI Study of Canine Brain Responses to Familiar and Unfamiliar Human and Dog Odors.” Behavioural Processes (2014).

Jia, Hao, et al. “Functional MRI of the Olfactory System in Conscious Dogs.” PloS one 9.1 (2014): e86362.


Letting the Cat Out of the Bag

I’ve talked a lot on this blog about canine social cognition, but what about feline social cognition? That may sound a bit oxymoronic to some, but cats, like other domesticated species, live closely alongside humans, so it makes sense that they could have social cognitive abilities.

However, an important difference between cats and other domesticated species lies in the domestication process itself: whereas most domesticated species were actively domesticated by humans through artificial selection, researchers hypothesize that cats were domesticated through natural selection. Unlike dogs, which could be trained to aid humans with hunting, herding, etc., cats had little usefulness to humans (except maybe killing rodents, which could also be accomplished by dogs). Cats were actually probably first attracted not to humans, but to the mice and rats that infested their fields and grain stores. The cats that were least averse to humans were more likely to hang around humans in order to have greater access to rodents. These less-averse cats then bred with each other, causing successive generations of cats to be less and less averse to humans. Cats essentially selectively bred themselves the same way Dmitri Belyaev selectively bred silver foxes.

The upshot of this difference in domestication between cats and other species is that cats weren’t bred to work with humans; rather, they were bred to be less averse to humans. This is basically the difference between cooperating and tolerating. And although both characteristics allow cohabitation with humans, being tolerant doesn’t entail the possession of any social cognitive abilities.

Fortunately for us, some researchers have investigated the influence of the differential domestication in cats and dogs by directly comparing their performance on social cognition tasks. Miklosi et al. (2005) first tested cats and dogs on the hidden food task, where a human indicates the location of food by pointing at it. Cats, like dogs, were able to understand the human social cue of pointing to find the food.

CatMiklosi et al. then observed cats’ and dogs’ behavior in a task where the food was impossible to access (very similar to this study). While dogs looked to humans for help, cats rarely did so.  Taken together, these results suggest that cats can understand human social cues, but are unable to communicate with humans.


A more recent study investigated whether cats could recognize the voice of their owner. Saito & Shinozuka (2013) played recordings of people saying a cat’s name and examined the cat’s response. In order to gauge recognition, the researchers used what’s called a habituation task. The basic principle behind a habituation task is that if you experience a stimulus enough (many times or over a prolonged period), your response to that stimulus will decrease. For example, when you first hear a fire alarm, you instantly become alert, adrenaline rushes through your body, etc. However, if the alarm goes on long enough and nothing else happens, your body will return to normal. Similarly, when you first put on clothes in the morning, you’re probably aware of how they feel against your skin for a minute or two, but then you stop noticing. Habituation is the brain’s way of filtering out unimportant stimuli so you can focus on the important stuff. In a habituation task, experimenters habituate a subject to a stimulus, then change the stimulus in some way and see how the subject responds. If the subject perceives the stimulus as different, then her response will be comparable to the first time she heard the habituation stimulus. If she doesn’t perceive the stimulus as different, then she will have little or no response.

For each cat, Saito & Shinozuka played three recordings of different strangers saying the cat’s name, followed by a recording of the owner saying the cat’s name. They were interested in comparing the cat’s habituated reaction to the third recording (a stranger) and the potentially different fourth recording (the owner). If the cat has a greater response to the owner’s recording than the third stranger’s recording, then he probably recognizes his owner’s voice.

The researchers found that overall, the cats had a greater response to their owners’ recordings than to the third strangers’ recordings, demonstrating that they recognized their owners’ voices. Interestingly, the responses were limited mostly to ear and head movements – no cats actually stood up and moved toward the sound. (The media have had some fun with this result – here’s an informative but slightly biased video about the study and its results.)

So it seems that although cats are domesticated and can understand some human social cues, they may not have all the social cognitive abilities of dogs, and this is likely due to their domestication through natural selection.

I’ll leave you with a video about an interesting study comparing the relationships between humans and cats, dogs, and babies. Do the results fit with our hypotheses about dog and cat domestication?

Sources Cited:

Miklósi, Áam, et al. “A comparative study of the use of visual communicative signals in interactions between dogs (Canis familiaris) and humans and cats (Felis catus) and humans.” Journal of Comparative Psychology 119.2 (2005): 179.

Saito, Atsuko, and Kazutaka Shinozuka. “Vocal recognition of owners by domestic cats (Felis catus).” Animal cognition (2013): 1-6.

If you’re interested in reading more about domestication, here’s an interesting paper:

Driscoll, Carlos A., David W. Macdonald, and Stephen J. O’Brien. “From wild animals to domestic pets, an evolutionary view of domestication.” Proceedings of the National Academy of Sciences 106.Supplement 1 (2009): 9971-9978.

This Little Piggy…

160px-Cute_PigletIn my very first post, I talked about how dogs can understand human social cues such as pointing to find hidden food. An interesting study published this month investigates whether domesticated pigs can understand human social cues, too.

The leading hypothesis of how dogs gained the ability to understand human social cues suggests that the ability may develop along with domestication (which I discussed here). It therefore seems reasonable that domesticated pigs may also possess the ability to understand human social cues.

Nawroth et al. (2013) conducted the study investigating the ability of young domesticated pigs to understand human social cues. Something I particularly like about this study is that they really tested the limitations of this ability in pigs by examining many factors, including the distance of the experimenter to the food and the length of time the experimenter pointed to the food.

The researchers used the same task throughout their entire experiment: an experimenter positioned between two bowls (only one of which contains food) makes some sort of cue towards the baited bowl, and then the pig chooses a bowl.

The first part of the study investigated two factors using what’s called a 2 x 2 factorial design. This type of study looks at two factors (independent variables), where each factor has two levels (a 2 x 2 x 2 factorial design would look at three factors with 2 levels each; a 3 x 3 factorial design would look at two factors with 3 levels each).

1519121063_0f075b7265_mThe first factor was the position of the experimenter: kneeling or standing. The second factor was the length of time the experimenter pointed to the food: one second (momentary) or until the pig chose a bowl (sustained). In a 2 x 2 factorial design, there will be four experimental conditions comprising all possible combinations of the factors; in this case, the conditions were: standing with momentary pointing, standing with sustained pointing, kneeling with momentary pointing, and kneeling with sustained pointing. The benefit to using a factorial design study is that it makes it easy to compare the effects of different factors (and combinations of factors) on the results (in this case, the pigs’ performance).

Nawroth et al. found that the pigs performed better than chance when the experimenter was kneeling, regardless of how long the experimenter pointed at the baited bowl. When the experimenter stood, however, performance was no better than chance. The researchers suggested two reasons why the pigs understood the kneeling experimenter better than the standing experimenter: one, the experimenter’s pointing hand was closer to the food in the kneeling condition than in the standing condition, so the pigs could rely on a close proximity between the experimenter’s hand and the bowl to make their choice. Alternatively, since pigs spend much of their time with their heads close to the ground, foraging, the standing experimenter may be outside the pigs’ field of attention (the pigs may not pay attention to things that far above the ground).

8076878857_2b39837f52_mTo investigate these two possibilities, Nawroth et al. tested the pigs on the two kneeling conditions from the first part of the experiment (momentary and sustained pointing), but this time, the bowls were farther apart from each other. Thus, in this part of the study, the distance between the kneeling experimenter’s pointing hand and the bowl was the same as the distance between the standing experimenter’s pointing hand and the bowl in the first part of the study.

The pigs performed above chance in both conditions, indicating that the height at which the social cue was presented, and not necessarily the distance between the experimenter’s hand and the bowl, affected the pigs’ performance.

Nawroth et al. further investigated the effect of the experimenter’s proximity to a bowl by testing the pigs with the experimenter either kneeling behind the baited bowl, or kneeling behind the empty bowl and pointing to the other (baited) bowl. They found that the pigs performed better than chance when the experimenter kneeled behind the baited bowl, but not when the experimenter kneeled behind the empty bowl and pointed to the baited bowl. So, while the pigs were unaffected by the distance between the experimenter’s pointing hand and the bowl in the second part of the experiment, the much closer distance between the experimenter and the bowl in this part of the experiment served as a strong cue for the pigs.

4902999050_85ab6bb35f_nNawroth et al. tested a few more factors with additional experiments – I don’t have space to detail them here, but you can read about them in the original paper, if you’re interested.

The gist of their study, though, is that pigs can understand human social cues including pointing (and also head and body orientation – part of the study I didn’t discuss here). There are some limitations to this ability (very close human proximity trumps the pointing gesture as a cue). But the fact that pigs are generally able to use these cues lends support to the hypothesis that this ability develops through domestication.

(As Nawroth et al. discuss in the introduction and discussion sections of their paper, the ability to understand pointing cues has also been found in other domesticated species, including cats, goats, and horses. However, none of these previous studies investigated head and body orientation cues.)


Nawroth, Christian, Mirjam Ebersbach, and Eberhard von Borell. “Juvenile domestic pigs (Sus scrofa domestica) use human-given cues in an object choice task.” Animal Cognition (2013): 1-13.

Experience or Evolution?

Last week’s post talked about how dogs can use human social cues like pointing to find hidden food, and that chimps, though more closely related to humans evolutionarily, cannot. How do we explain this?

3517862378_43be39f2f6_mA major difference between dogs and chimps that could explain their difference in this skill is that dogs just have more experience with humans than chimps do. All the dogs tested on their use of human social cues had interacted with humans since birth. The chimps, though they had exposure to the humans who took care of them in the sanctuary where they lived, arguably had much less interaction with humans than the dogs. So, the factor influencing performance on the food-finding task (and understanding of human social cues) could be the amount of experience an animal has with humans. This theory is called the “human exposure hypothesis”.

One way to evaluate this theory is to test very young puppies on the food-finding task. The puppies, due to their young age, would necessarily have less experience interacting with humans. If the human exposure hypothesis is true, we would expect the puppies’ performance on the task to be worse than older dogs’ performance.

When researchers tested puppies between 9 and 24 weeks of age, they found that they were also able to understand human social cues. Moreover, age (i.e. amount of experience with humans) was not correlated with performance on the task — younger puppies were just as good at the task as older puppies. These findings indicate that the human exposure hypothesis is incorrect, and suggests that the ability to understand human social cues is innate in dogs.

5532096313_cfaf563c7a_mAnother possible explanation for the ability of dogs to understand human social cues when chimps cannot is that dogs have been domesticated — they’ve been selectively bred to have characteristics desirable to humans. We don’t know the specific details of how dogs were domesticated, but we do know that they split from their ancestors, gray wolves, about 100,000 years ago.

Researchers tested wolves on the food-finding task and found that they were unable to use human social cues to find the food, indicating that dogs acquired this ability sometime after their split with wolves. This finding supports the theory that dogs evolved the ability to understand human social cues through domestication (this theory is called the “domestication hypothesis”).

So, unlike chimps, it seems that dogs are able to understand human social cues because they’ve been bred by humans to do so. But a couple questions remain:

1. Was the evolution of this skill due to humans specifically breeding wolves that understood human social cues, or was it due to humans breeding wolves based on a more general trait, like behaving friendly towards humans (or, on the other hand, NOT breeding wolves that showed aggression towards humans)?

2. How would you even test the domestication hypothesis, besides spending thousands of years selectively breeding wolves (again)?

Next week’s post will explore the answers to these questions and conclude this series on dogs and human social cues!

Until then, here’s an interesting video that talks a bit about how dogs evolved from wolves and suggests that dogs played an important part in human evolution.


Hare et al. (2002) presents data from both the studies mentioned above, and also briefly discusses a few hypotheses of why dogs can understand human social cues.

Hare, Brian, et al. “The domestication of social cognition in dogs.” Science 298.5598 (2002): 1634-1636.