Art and the science of canine consciousness

a painting consisting of a vertical stripe of blueThe frustration I feel when I see people look at abstract art and say, “I could do that,” is tinged with hypocrisy. Why? Because when I look at paintings like Blue Column, by Morris Louis, I say it too.

I say it even though I know that Blue Column is art and is important. Afterall, even though I could have produced that painting, I didn’t. And, I love color field paintings and can gaze at them for hours.

I have to remind myself that just because the bar might seem to be set very low when we call an abstract painting “art,” it doesn’t mean that the picture is unimportant or banal.

This attitude toward art helps me be more fairminded and respectful of science. Particularly of scientific research that seems to set the bar really low for canine emotion, consciousness, and cognition.

There is lots of scientific research that seems to do this. And my initial response to that research is generally pretty snarky.

For example, Juliane Bräuer et al just published a research study titled, “A ball is not a Kong: Odor representation and search behavior in domestic dogs (Canis familiaris) of different education” in the Journal of Comparative Psychology. The paper reported that when working and companion dogs followed a scent trail produced with one toy and encountered a different toy at the end of that trail, they hesitated.

My initial disrespectful response to the paper was, “No shit Sherlock, I could have told you that.”

But that response inhibits my curiosity about the research program that produced the paper and undermines my ability to think about why the researchers did that experiment and how they might have done it better.

This experiment was designed to explore whether dogs have a mental representation, a sort of olfactory picture in their minds, of objects in their world. The researchers are interested in canine consciousness, just like me.

The researchers found that pet dogs and working dogs, some of whom were trained in search and rescue, both hesitated the first time they encountered the wrong toy at the end of a scent trail, but didn’t hesitate in subsequent trials. They also found that the working dogs followed the trail faster than the pet dogs, but again, only in the early runs of the experiment.

This experiment brings some interesting things to mind:

I’m still bothered because the researchers set the bar so low for determining whether a dog has a mental representation of what she’s smelling. From my perspective, there was no need to do this experiment because the results are obvious:

  • If we think of this from an ecological or evolutionary perspective, it’s difficult to imagine a creature who uses scent to hunt for prey not hesitating when they find a surprising item at the end of a track.
  • I do tracking with my German Shepherd, Milo. When he comes to the end of a track, which is what happened in this experiment, he hesitates and sniffs around trying to find it again. That is what dogs do, and it is what the dogs in this study did. It seems that if the researchers had collaborated with an expert dog handler, they’d have seen that this was an unnecessary experiment.

But, I sometimes forget that my perspective isn’t the only one out there. What if the researchers weren’t trying to prove this point to someone like me?

Afterall, 25 working dogs took part in this study and presumably those dogs’ handlers knew what was going on and I bet the results weren’t surprising to many of those people either.

These researchers were speaking to a scientific community in which many members are skeptical that creatures other than humans and chimpanzees have rich inner lives. (I have one friend who describes rabbits as furry machines that turn carrots into poop.) When juxtaposed against background beliefs like these, this study’s results become surprising, interesting, and important.

Also, this study might be a necessary building block for more complicated investigations of canine consciousness and representation.

Finally, some people will give scientific knowledge more authority than the experience of expert dog handlers and trainers. For example, a scientific paper might convince policymakers who would not be moved by expert testimony that we need practices and laws that respect dogs as having rich inner lives.

I didn’t need this experiment to tell me that dogs have representations of what they are smelling. But that doesn’t make the research unimportant. This experiment is much more convincing than my blustering “I could have told you that” will ever be.



Are dogs smarter than cats? A new​ scientific​ study doesn’t say one way or the other

After the announcement of a neuroscience experiment on carnivore brains headlines like these leaped onto the internet:


Milo and I, as well as Hoss the Cat, share Grumpy Cat’s assessment of these headlines.

You see, the research paper that triggered this brouhaha about dogs being smarter than cats wasn’t about dogs being smarter than cats. This research was conducted in Suzana Herculano-Houzel’s lab at Vanderbilt University. These scientists took the brains of a ferret, banded mongoose, cat, raccoon, hyena, lion, brown bear, and two dogs, dissolved parts of those brains in a detergent solution, and used that solution to determine the number of neurons the brains contained. Researchers then compared the number of neurons in some parts of a creature’s brain to the size of the creature’s brain and the size of the creature itself. They did not do any intelligence testing of any sort. All of this ‘dogs are smarter than cats’ talk is based on the assumption that the number and density of neurons are correlated with intelligence (in this case it’s good to be dense).

This is a very cool experiment, and the researchers found some surprising results, which by the way, had nothing to do with comparing dog and cat intelligence. Here are some of the things they found:

  1. A lot of people tend to assume that predators are smarter than prey animals. From this assumption, a person might predict that predator species will have more cortical neurons than prey species. If there’s a correlation between neuronal density and intelligence, the results of this study contradict this prediction. This study found that predators had about the same neuronal density as their prey.
  2. Scientists expected to find a correlation between how big an animal is and how many neurons it has —bigger animals have bigger brains. This study found that the correlation generally held for small and medium-sized animals but not for big animals. Animals larger than a Golden Retriever had fewer neurons than these scientists expected. Perhaps, these researchers speculate, because neurons are so energetically costly. Think of neurons as race cars in that they take expensive fuel and burn it really fast. It might be the case that large animals cannot afford to have lots and lots of neurons because they lack sufficient fuel.
  3. There is a general belief that domestication makes animals stupider. These scientists did not find decreased numbers of cortical neurons in the domestic species they looked at. So again, if there’s a correlation between neuronal density and intelligence, this study contradicts the idea that domestication results in decreased intelligence.

When scientists find unexpected results or results that contradict things lots of people assume are true, is when science gets exciting and important. These three findings are the best parts of the study, but you would never know it from the headlines racing around the internet.

To be fair, these scientists did find about twice as many cortical neurons in the brains of a Golden Retriever and a small mixed-breed dog than they did in a cat. This is indeed a striking difference. They also found that bears had way fewer and raccoons had way more cortical neurons than expected for their body size. These findings about the bear and the raccoon are way more scientifically interesting than the cat-dog business.

Even though the headlines miss these exciting bits of the research, I am concerned, cranky even, other reasons.

The study did not measure intelligence. It measured the number and density of neurons in one or two individuals of several species. There is a big difference between the number of neurons a species has and how smart or brainy or intelligent that species is. There are many questions left to answer before we start casting aspersions on the intelligence of cats, or on making any inferences about intelligence at all. I would like to know if there are differences in how well those different neurons work, how big they are, and how they are connected to one another. And even if I had the answer to those questions, I would still feel uncomfortable using that data to draw conclusions about the intelligence of a species. Many news stories about this experiment contain buried disclaimers warning us about making assumptions about intelligence. However, they make the very assumption they warn against throughout the story and rely on it to create catchy headlines to interest people in this research in the first place.

Intelligence is not a thing, it is many things. Or more accurately many different kinds of capacities or skills. Some creatures are good at living in social groups, some are good at making tools, some are good at solving puzzles, and some have great spatial memory. There are many different ways to be intelligent. If you are going to talk about intelligence, you have to be very clear about what you mean.

The reporting of this research bugs me not only because the reasoning is sloppy or because Hoss the Cat is getting an unfair deal, although both of those things are true. It bugs me because it has the potential to hurt people.

There are lots of interactions between human and nonhuman research in areas like medicine, public health, psychology, and neuroscience. Research papers about human neurobiology, psychology, and intelligence draw on animal studies all the time. Sloppy reasoning about cats and dogs can lead to and support sloppy reasoning about people. Sloppy reasoning about people has been used and is still used in some circles, to justify sexist and racist conclusions about humans.

So, let’s be more careful. One way to do this is to distinguish between what the data clearly says and what we interpret the data to mean. Once we do that we can start figuring out where we are making assumptions and deciding whether those assumptions are justified. In the case of the density of neurons in cat’s and dog’s brains, the data tells us that, in the sample of two dogs and one cat, the dogs had more neurons in some parts of their brains than the cat did. It does not measure intelligence. The media stories about this research assume a connection between numbers of neurons and intelligence. It might turn out to be true that the number of neurons in some parts of the brain is, for example, related to a species’ ability to solve spatial problems. But this study does not give us data that supports this assumption.

In fact, exciting things can happen when we take the same data and pair it with a different hypothesis and a different set of assumptions. Let’s start by considering one of the interesting conclusions this experiment suggests. It appears to disprove the long-held hypothesis that predator animals are smarter than prey animals. Here’s the argument against that hypothesis:

  • If predators are smarter than prey, [our hypothesis]
  • and the number of neurons is correlated with smartness, [an assumption]
  • then we would expect predator species to have more neurons than prey species. [a prediction]
  • In the experiment, we did not find that predators have more neurons than prey species. [the data]
  • So, our hypothesis about predators being smarter than prey is false.

But, look what happens when we flip things around and use the data to test the assumption I made in the above description of the experiment:

  • If the number of neurons is correlated with smartness, [a new hypothesis]
  • and predators are smarter than prey, [a different assumption]
  • then we would expect predator species to have more neurons than prey species. [the same prediction]
  • In the experiment, we did not find that predators have more neurons than prey species. [the same data]
  • So, our hypothesis about the number of neurons being correlated with smartness is false.

The point is that an experiment and a set of data can be objective and accurate, but can still be used to prove many different things depending on the assumptions that go along with it. We can use the same data to disprove the assumption, OR to disprove the hypothesis. Pay attention to the assumptions and look for independent ways to test them!

And, give your cat a break, she still might be a super genius.


My method is scientific 5: Responsible use of science

There is a very high bar for claiming that a dog training method is scientific. Not only does there have to be peer-reviewed research studies, but there also has to be a bunch of them, and they need to directly refer to the training method in question. It is a difficult standard to meet.

Just because it is difficult to meet this standard, it most certainly doesn’t mean that we should abandon scientific research on dog training. What the high standard does mean is that we need to be careful, responsible, about how we use scientific information.

In the rest of this post, I’ll explore just how high that bar is, and give you an alternative way, a responsible way, to think about and use scientific information about dog training methods.

What the heck is a body of evidence anyway?

The idea is that there needs to be a bunch of studies and that you need to be on the lookout for studies that support your training method AND studies that don’t. It’s not immediately clear how many studies make a bunch, or how to balance the studies that support and that don’t support a particular method. Most of the time a person has to rely on scientific experts to get an overall idea of the degree of support for particular methods.

Philosopher Heidi Grasswick has argued that it should be part of the job of science to sift through the evidence it generates and provide us with usable, significant information. However, there is not much motivation for individual scientists to do this time-consuming work. Thankfully, when it comes to veterinary and human medicine and public health, there are professional and government organisations we can turn to for this sort of expertise.

For example, the Canadian Veterinary Medical Association has a Humane Training Methods for Dogs position statement. In this statement, they recommend reward-based methods, and discourage aversive methods, and do not recommend aversive methods unless reward-based methods have failed and unless the aversive methods are used by a skilled person. They provide citations to the scientific literature supporting their claims, and you can use those citations to track the research papers down for yourself. However, even this position statement is not especially specific about particular training methods. It pretty much advises us to be as humane as possible.

Is the evidence ever directly about what you do with your dog?

If a person wants to call a dog training method scientific, then the scientific research should be about the dog training method they are actually talking about. This seems obvious, but in fact, there is rarely, some might even say never, a direct connection between the research and dog training practice. One reason for this mismatch is that studies are conducted under controlled experimental conditions, and dog training happens ‘in the wild,’ or at least in the more complicated everyday situations that we find ourselves in. The context the dog-handler team is working in, the skill of the trainer, the temperament and experience of the dog, all make a difference. Also, sometimes, a lack of communication between scientists and dog trainers can result in them using the same name to describe what are, in practice, different training methods.

The bottom line is that it is difficult to be justified in claiming that a training method is scientific. But don’t give up on scientific approaches to dog training.

A matter of degree

Instead of thinking of the scientific nature of a training method as an all or nothing sort of thing, we need to think of it as a matter of degree. Some training methods are supported by more evidence than others, and the scientific research can be more or less like what we actually do with our dogs.

When we point to a method as being scientific, our usual goal is to justify our use of it or recommend it to others. More evidence that is closer to our training method, justifies us in giving a stronger endorsement or recommendation of the method as scientific. It’s not an on/off switch; it’s a volume control knob.

Take care though, because scientific evidence can’t give us 100% certainty that a method is the best possible method. In addition to the tricky nature of evidence and the fraught relation between science and practice, science can only test what we have dreamed up in the first place. It is completely possible that someone will come up with a new method that is better than anything we have right now. Also, science is trustworthy because it depends on the experiments we’ve actually done and the measurements we’ve already taken. This means that we always need to be open to changing our minds in the face of new experiments and new evidence.

Responsible use of scientific information means calibrating our endorsement of a method as scientific to the quality of the evidence supporting that method.


Note: this is my fifth post in a series on scientific dog training methods.

  1. “My method is scientific” 1: “That’s right, I said ‘SCIENTIFIC’!”
  2. “My method is scientific” 2: What does this even mean?
  3. “My method is scientific” 3: The trouble with clicker training
  4. “My Method is scientific” 4: Science, goodness, and goals
  5. “My method is scientific” 5: Responsible use of science

“There is no Nobel Prize for fact checking.”

A particular sort of delightful demoralization washed over me when I saw comedian John Oliver sum up years of my research in 19 hilarious minutes. Here are some things that you will learn from this segment:

  • science is difficult,John_Oliver_November_2016
  • not all scientific research is fabulous,
  • scientists are pushed to publish new and striking results, and are not rewarding for double-checking other peoples’ research,
  • scientific research can be manipulated in ways that sacrifice its integrity,
  • science is often distorted by the media to make it more newsworthy,
  • people tend to be more interested in easily-digested, exaggerated, glamorized science, than in the nuts and bolts of how science actually works,
  • and, farts don’t cure cancer.

These things undermine the credibility and authority of scientific research.  We need excellent, credible, and authoritative scientific research to address public health challenges, environmental problems, climate change, and a host of other significant issues. This is one reason why it is important to think critically about how we produce, share, and use scientific knowledge.

John Oliver says all of this in a way that might not be safe for all workplaces, but is really funny.