Sunday, November 27, 2011

Patricia McConnell Seminar: Canine Cognition

Historically, there has been a lot of animal research, but not much on dogs. I've always found this puzzling, but I guess working with non-human primates or rare birds is a bit more exotic. Thankfully, research on man's best friend has exploded in recent years, so today I'm going to share what Patricia had to say on the subject of canine cognition.

As always, we must start at the beginning. Can dogs think? For those of us who live with them, the question seems silly. Of course they can. But science requires that we prove our assertions. So what is “thinking”? Patricia shared that the scientific community generally defines it as: the ability to formulate an abstract mental representation of an event or object external to the self and the ability to manipulate that representation to solve a problem.

That's a pretty big definition, so let's break it down. First: can dogs understand abstract concepts? It seems that they can. Ken Ramirez has done quite a bit of what he calls “concept training”- teaching dogs (and other animals) things like left vs. right, high vs. low, and big vs. small. He's even taught them how to copy the actions of other animals!

Imitation is actually a pretty complex cognitive process. In order to perform a novel behavior after observing someone else do it, the observer needs to have self-awareness, some measure of empathy (as in, his left front leg is like my left frong leg), and an ability to translate seeing into doing. Ramirez isn't the only person to have taught mimicry- Adam Miklosi from the Family Dog Project in Budapest has done it, too.

No discussion of smart dogs would be complete without talking about Rico the border collie who has demonstrated the ability to “fast map.” This is a mental process in which children (and apparently Rico) learn the meaning of a new word after hearing it only once. Rico learned the names of over 200 different objects, and could retrieve them for a different room when requested. That's pretty impressive on its own, but when Rico was told to find the name of something new, something he'd never heard of before, he could correctly choose the novel object through the process of elimination.

In a similar vein, European researchers have taught dogs the concept of “match to sample.” Using touch screens, they have the dog match two identical images. Dogs are quite adept at this, and so researchers began assigning “value” to the images. Some images resulted in treats, and some didn't. Dogs learned to choose the “positive” images and avoid the “negative” ones. When two images were placed next to each other, the dogs would not only choose the positive ones, but would also infer that the other image- even if they'd never seen it before- must be negative. After that, they would avoid the negative item, even when it was paired with another novel image. Pretty neat.

These are pretty complicated ideas, so I'm really glad that somone uploaded this video demonstrating them:

So, it seems pretty clear that dogs can think abstractly. But can they solve problems? Again, we dog people would agree this is a no brainer. Thankfully, science supports us: dogs can solve problems. What's interesting, though, is the way they do it.

Dogs, dingoes, and wolves have all been tested on what's called the “detour test.” The subject is on one side of a see-through fence, and there's food on the other. The task is to go around the ends of the fence in order to get to the food. It sounds simple, and the dingoes would agree: 100% of them could complete this task within 60 seconds. Dogs, though? Depending on the study, only 60-80% could do it.

Why is this? Well, scientists have found that wild canids seem to be more adept at solving non-social problems than dogs. Man's best friend truly believes in his role, and is more likely to look to his owner than to try to solve the problem himself. Which, let's be fair, is a way of the solving the problem. After all, why not get the being with opposable thumbs to do the hard work?

Another study set out to see if dogs could recognize if their owners needed help and alert a nearby person. They set up two situations, in which the owners either fell to the ground, faking a heart attack, or where the dog entered the room to find their owner trapped under a bookcase. Lassie they were not: not a single dog sought help from a bystander. The authors concluded that the dogs could not recognize an emergency, but Patrica criticized this, sharing that she thought it possible that the dogs knew their humans weren't actually in danger. I tend to agree with her- after all, dogs have an excellent sense of smell, and their people probably didn't have any of the chemical indicators of distress. Add to that all of the hundreds of anecdotal stories of dogs saving their owner's lives... well, I think further research is needed.

The good news is that there will be further research. Canine cognition labs have sprung up all over the world, and there is a lot of really interesting stuff being studied. Check out the links below not only for references to the specific studies Patricia discussed, but also to the labs' sites in general- there is tons to explore there, and the science geeks out there will be in heaven. (There are also some links to non-canine animal cognition studies that Patricia shared with us, but that just don't fit in with this post.)

In the meantime, I'd love to hear stories about your dog's ability to think. Has he ever copied you or another dog's actions? Does he have a unique method of solving a problem? What amazing feats has he accomplished? Tell us! Anecdotes may not be science, but they sure are interesting... and maybe someone will read your story and decide to study it.

If You Want to Know More:

Canine Cognition Labs:


Tegan said...

Phew! So much to think about and comment on.

Recently (ish) I went to a seminar with Paul McGreevy. He spoke about dog science, and he believed we perhaps have little data on dogs because dogs are always interconnected to people... That is, science is interested in 'natural' dogs, but any dogs interacting with people, are tarnished and no longer unnatural. Of course, science is now hoping to control for the human element, which is good. There is also no science field for 'dog science', so you can't really train in a field that doesn't exist (at least in concise terms). [I have more from Paul McGreevy on my blog, with a post about his thoughts on dog science.]

I still can't quite believe in imitation, despite anecdotal evidence. But I'll follow your links and think about it. Certainly, the study on 'jealousy' (as you recently posted) suggests that dogs must have self awareness to realise they are 'the same' or 'not the same' as the other dog.

I soooo believe that dogs are very social and very reliant on people for cues. I describe my bitch's mothering experience as, "Tell me what to do next." This switched on dog, who I had trained from day dot to be attentive and responsive to me, seemed to be undeterred by mothering hormones and still sought my input on what she should do with her puppies. Somewhat frustrating! But an interesting phenomena.

I agree with you - the distress situations seem bogus. I think dogs would smell distress and respond differently to any staged scenario.

As for your questions:

I don't think I have seem very explicit imitation from dogs. The only scenarios I can think of is when you call one dog and get all dogs. A nervous dog can sometimes overcome fears by responding with this pack mentality. I don't think it's truly imitation, though, just the dog seeking comfort in its companion dogs - and if they happen to be close to a person, then so be it.

The clever / creepy thing that Clover does is use mirrors to her advantage. I trained her to make eye contact, for use in obedience, and countless times I have seen her making eye contact - through a mirror! She clearly doesn't think my reflection is me... As soon as I make eye contact with her through the mirror, she gets excited, and then engages with my physical self.

I keep meaning to train her a hand cue for a behaviour, to see if I can cue her to do something through the mirror. (All my current cues are verbals only.)

On the other hand, it's highly possible that I rewarded her 'eye contact through mirror' behaviour excessively the first couple of times (in surprise) and she has, in turn, thought it'd be a pretty good trick. She has learnt other behaviours in 2-3 repetitions, so perhaps I trained this a little bit better than I wanted to.

Ci Da said...

I had a recent interaction with Cohen that really impressed me. She understands the cue "show me something" which means for her to choose a trick from her repertoire for her to earn a treat. It's already a pretty advanced cue as far as cognition goes.

So I was sitting down with her and asked her to "show me something", which she did. I asked again, and she showed me another trick. And again and again. All told, she showed me 6 distinct tricks before I ended the session. What I found so impressive that she was not only going through her list of tricks in her head, but she was tracking the ones she'd done already and was careful not to repeat them.

Granted this isn't of the same level as Rio, but it impressed me and gave me some insight into canine cognition and intelligence.

Crystal Thompson said...

Tegan, that's a great point about science and dogs- that by living with us they are somehow less natural. Patricia mentioned that it used to be that dogs would be raised in isolation for research (unhandled, sterile environments), but that wasn't good for their long-term prospects. She told a story about how when she was doing some research for her PhD (I think) that she snuck toys into the kennels and would have to pick them up when people came to visit, lest she got caught breaking the rules.

As for imitation... Maisy once did something curious. I don't know if it would be imitation or coincidence, but it was interesting! I was trying to shape her to push different objects with her nose (for this video: ) One of the things I was trying to get her to push was different cylinders, including a play tunnel. She would NOT touch the play tunnel. Refused, no matter how small I broke down the steps (looking at, moving towards, etc.).

Finally, I poked the tunnel with my own finger, and clicked myself, and gave her the treat. I have no idea why, but immediately after that, she poked the tunnel with her nose. It's not imitation exactly, since I didn't use my nose, but... it was interesting!

Very, very cool that your dog uses mirrors. PLEASE tell me if you ever do the hand cue through a mirror thing. I'd love to hear what happens! :)

Ci Da- that's pretty cool that your dog didn't repeat any tricks. Most dogs repeat things when they do the "show me something new" game... at least in my limited experience.

Tegan said...

Found this abstract today regarding dogs using mirrors... Going to try to download later!

Crystal Thompson said...

Tegan, you ROCK.