Can dogs recognise their owners' faces?

We delve into some of the latest neuroscience research with our local experts...
21 October 2020

Interview with 

Duncan Astle, Cambridge University; Helen Keyes, Anglia Ruskin University


Brain schematic


First up is our usual neuroscience news segment, and cognitive neuroscientist Duncan Astle has been looking into a paper asking whether man’s best friend can recognise faces in the same way that humans can...

Duncan - So they've recruited 40 participants - 20 humans, and 20 family dogs. For the humans their average age was 32, and 47% had a Master's degree or equivalent and 37% had a Bachelor's degree. For the dogs their average age was five and to date, none had yet completed any higher qualifications.


So all of the subjects were put in the fMRI scanner - the functional magnetic resonance imaging scanner. And they were shown videos with human faces or human backs of heads, and dog faces or dog backs of heads. So we've got a two by two design. Face versus back of head, and species - human versus dog. And the stimuli were really carefully controlled. So for example, the dogs and humans would never look directly at the camera and that's because apparently that can be triggering for dogs and can make them anxious or aggressive. So they had to think very carefully about how they delivered the stimuli. What they wanted to explore is whether or not both species are sensitive to faces versus backs of heads, and whether they discriminate the kind of species that it is.

The headline result is that in humans, there are lots of visual areas that are very sensitive to faces, but not particularly sensitive to species. Whereas in dogs, there were lots of visual areas that were highly sensitive to species. So those brain areas seem to distinguish another dog from a human being. But not to faces ie they don't seem to distinguish the front of the head from the back of the head.

Now, amongst their analyses, one that they ran was called an MVPA analysis or a multi-voxel pattern analysis. What it's doing is testing what information is being represented in different brain regions. So that if we read out the activity in that brain region, could we predict whether the person was currently looking at a face, a back of the head, a dog, or a human?

In the dogs they found that the left medial and right caudal supra sylvian gyrus showed really good decoding for species. So the dogs had brain areas that seem to be quite specialized for distinguishing dogs from humans, but there are no significant areas that seem to be specialised for distinguishing faces versus not faces i.e. backs of heads. Contrast that with the humans, so in one particular area, the right middle temporal gyrus, you can decode species, so they can distinguish humans from dogs. And then in loads of areas, including the fusiform gyrus, sometimes called the fusiform face area, you see selectivity for faces, so a classic face response. In essence, it seems to be that while both species can decode the species of the person they're looking at, only the humans seem to have this kind of selective face processing set of areas.

And a final analysis they did was called a representational dissimilarity analysis, a look at different brain areas to see how well it distinguishes all of the different videos that people were shown. And then they can test whether there are parts of a dog's brain that has a kind of profile across those video clips that's very similar to a part of the human's brain and vice versa. So they can see whether there are analogous regions in the dog that seem to have a similar kind of representational profile of faces and species as a human would have. And they find that they do get some analogous regions for species, but not for faces. Again, supporting the idea that dogs don't really recognise the human faces or the dog faces. And that is that.

Katie - So overall, is what the papers saying that dogs are really good at telling that's a human that's a dog, perhaps not individuals so well. Whereas humans are really good at going “that's Duncan, that's Helen, that's Katie”.

Duncan - Precisely. There are all sorts of explanations as to why that might be. So for example, most dog owners tell me that their dogs recognise them. But if the dogs can't really recognise human faces, then how can they recognise their owners? So my guess is that they're using other things. So it may be that for human beings because of the way that our visual systems work, because of our language systems, needing to look at lips and so on, it may be that we've become highly specialised for decoding faces and distinguishing one person's face from another. Whereas for dogs, someone's face is just one of many useful cues for distinguishing an individual from another. And because they're not really looking at lips for language, the face is perhaps less vital than it would be for a human.

So it's obviously a really interesting paper. I mean the very fact they've managed to get people's dogs into an fMRI scanner, I think is itself quite impressive. But behind, you know, kind of the novelty factor, they're asking a kind of interesting question, which is, where does this special processing for faces come from? At what point in our kind of evolutionary past does it develop or does it emerge? And, you know, we know that higher order primates have similar face processing areas to human beings, but it's obviously not ubiquitous to all mammals because the dogs don't show it. So the question is, when does it crop up and why? So it may be that for instance, humans and other primates have a kind of uniquely sort of social interaction and social structure, where being able to distinguish one person's face from another becomes a really useful tool that perhaps was not there previously for other kinds of mammals.

Now if you listened to the last few episodes of Naked Neuroscience, you’ll know we’ve been pondering the subject of play. And perceptual psychologist Helen Keyes’ paper of choice this month - asking whether playing with dolls can impact social development - fits rather nicely into that theme...

Helen - Engaging in pretend play is really helpful for a child's development. It helps them to develop both cognitively and socially. And this study was looking at the neuroscience of this. So what's happening in the brain when children are engaging in this pretend play?

So pretend play or what you might call, make believe play, typically involves using toys or dolls to act out a pretend scene. And so you might be really putting yourself in the position of your toys or taking the perspective of your dolls as you're playing with them. And this type of pretend play usually emerges around the age of two. It can happen alone, or it can happen with a play partner, but even when it's happening alone, there's a lot of evidence that this is still social in nature. So you're kind of imagining an audience in this type of play or you're taking the perspective of others in this type of play. So it is particularly interesting to us as psychologists to ask whether this type of perspective-taking in pretend play does help to develop childrens' social brains, essentially.

So the scientists in this study used fNIRS, which is functional near infrared spectroscopy. And that essentially measures blood flow in different regions of the brain. An increased blood flow would be a good indication that that part of the brain is particularly activated. And here the scientists used this fNIRS to measure blood flow in the brains of 33 children who are aged between four and eight, while they engaged in different types of play, Either open-ended creative play on a tablet - so this would be where you are cutting hair or building towns on the tablets, open-ended play - or if you're playing with dolls and kind of doll sets. Both of these activities were recorded when the children were playing alone with these activities or when they were playing socially. So the research assistant was engaging in this play with them.

And the researchers were particularly interested in the parts of the brain that are strongly involved with social cognition. So empathy and perspective taking, and the particular part of the brain they focused on was the posterior superior temporal sulcus. What they found, interestingly, was that there was no effect of the age of the child or of the gender of the child here. So the findings I'm going to talk to you about applied equally for boys and girls.

They found that the social regions of the brain were really activated during joint play in general. However, during solo play, when you're playing with a tablet that activation dropped off. So that social part of the brain wasn't engaged, but that activation remained for solo play when the children were playing with dolls. So it looks like imaginative play using dolls engages the social regions of your brain. These regions involved with empathy and perspective taking, even during solo play.

Katie - That's so interesting Helen, because anecdotally, I've sometimes heard, you know, of parents who have a little one and then are expecting another baby might say things like, "We'll get our little one a doll", in some way kind of to prepare them for having a sibling. Do you think this kind of speaks to that in any way?

Helen - I think it does speak to that. It seems that playing with dolls or engaging in this type of imaginative play is a way of rehearsing your social skills and in a way of, of really thinking, encouraging our children to think about other people and that perspective taking. So absolutely I would recommend it to be a good idea to ask your child to engage in this type of doll play, to develop them socially.

Katie - But do we actually know if it makes a difference? I guess it's one thing to say these areas of the brain are highlighted, but has anyone followed kids up and thought, "Ah, these kids are more empathetic adults", or anything like that?

Helen - Well, engaging with this type of pretend play, yes, we do know that it's good for different areas of the brain. In general, engaging in pretend play is good for your cognitive skills and your social skills. So that's quite well established. What was interesting about this study, was it was showing that even in the absence of a playmate or that direct social stimulation, at a neural level we can see that engaging in imaginative play with dolls is, you know, recruiting those areas. In a way, making those areas practice your social skills. So yeah, we do know that engaging in pretend play is good for your social skills, but this is such a direct neural demonstration of that. It's just really strong evidence.

Katie - Do you think the tablet is a good comparison? I was just wondering about, you know, what, if you just give a kid some blocks or a stick or, or nothing, and just ask them to make up a game? Would that be more work for the brain?

Helen - The reason they didn't do that is if you give a child some blocks or some sticks, they will often engage in the type of pretend play that we would engage with when we're using dolls. They often won't, sometimes they will just stack the bricks or make rules for a game for themselves, but they really wanted to disentangle those social elements here by just giving a child a tablet and asking them to engage in a creative free play, not a rule based game, but it wasn't a social game.

The other point I wanted to make was that this study was funded by Mattel who make Barbie. So while I'm very confident looking at the research methods and looking at the researchers involved in this study, that it was, you know, a proper legit study, it is important to say that it was funded by the makers of Barbie.

Duncan - fNIRS is great in that it's portable. The kids can play. We haven't got to slide them in a scanner. But also one of the challenges is that you don't necessarily get the same coverage. So one possibility is that whilst the tablets seem to be less engaging social areas, it may be that what kids can do on them is more cognitively demanding in other ways than playing with a doll. So for example, if they're building something on Minecraft on the screen, it may be that they're engaged in all sorts of other areas to a greater extent than playing with a doll. Different toys for different types of play, yielding different types of benefit. Do you think that's possible?

Helen - Absolutely. I think that is certainly the case. If we were asking a child to engage in rule-based play, we would almost certainly see frontal areas, the prefrontal cortex, recruited, much more than during doll play. So absolutely I would recommend that we should be encouraging our children into all these different types of play. So not, you know, doll play alone might be great for social skills, but not so great for cognitive skills and vice versa. I think what it really tells us, when we put all this research together, is it's a really bad idea to have children just playing with one type of stimulus or one type of game all the time.


Add a comment