Monday Pets: Cold Blooded Cognition

She: "What are you writing about?"

Me: "Cognition in cold-blooded animals."

She: "Hot."

Most people who study cognition focus on mammals or birds. But I hope I've convinced you that other animals are important to investigate as well. One research group at the University of Vienna likes cold-blooded critters. Turtles and lizards and such. They argue:

Reptiles, birds and mammals have all evolved from a common amniotic ancestor, and as such they are likely to share both behavioural and morphological traits. However, this common ancestor lived around 280 million years ago and so it is equally as likely that different traits and abilities may have emerged. Despite their clear importance for the study of cognitive evolution, very little research has investigated the learning abilities of reptiles. The few studies that have been conducted with reptiles found little evidence of impressive cognitive skills. However, many of these studies took place in unsuitable environments for the species tested (e.g. a cold room for a tropical reptile). As reptiles are ectothermic (cold-blooded) it is essential to provide them with an environmental temperature similar to that which they would experience in their natural habitat. Only then can their true cognitive abilities be tested.

The ability to learn from the actions of another conspecific (a member of the same species) is adaptive. Animals who live in social groups, such as mammals, birds, fishes, and insects, can learn how to solve certain problems by watching other group members solving those problems. Nobody, according to these authors, has studied this form of social learning in reptiles. But the evolutionary origins of this trait - the ability to learn by observation - are unclear. Given the right circumstances, could the red-footed tortoise (Geochelone carbonaria) show evidence of this form of social learning?

An implicit assumption often made is that living in social groups promoted the evolution of social learning. This leads to the hypothesis that social learning is an adaptation borne out of social living. But research that has tried to test this hypothesis has run into an important confounding factor: some species are simply better at learning than others. The species that are better in social tasks also perform better in non-social tasks.

Wilkinson and colleagues think that they have found a better way to address this question. They took individuals from a non-social species (red-footed tortoises) and asked whether or not they could solve a task by observing the actions of a conspecific. If social learning is an adaptation to social living, then they should not demonstrate successful learning by observation. However, if the ability to learn socially is simply just a specific effect of a more general ability to learn, then any animal that can learn should be able to use social cues just as they would use any other environmental cue.

The red-footed tortoise is a solitary species that is found in the tropical forests of Central and South America. They do not live in permanent groups (unlike a particular group of sewer-dwelling, pizza-eating ninjas that I adore), and there is no parental care. So, you're a tortoise and you spend your entire life alone. One day, you're walking around and you find another tortoise. You do a headbob display and figure out if the other tortoise is of the opposite sex. If so, game on. Make some little tortoises. Lay some eggs. If not, keeping moving. That's pretty much how it goes.

The eight tortoises were housed together for two months, allowing plenty of exposure to other conspecifics. This would obviously not have allowed any adaptation to evolve in response to the social environment. Even though they were living together, they never saw anyone do the experimental task except under strict conditions. For the experiment, they were separated into two age- and sex-matched groups.

The goal of the game was to get a piece of food, but there was a transparent V-shaped fence separating the turtle from the food. One group (the "non-observers") had to try to figure out how to get the food on their own. The second group (the "observers") watched a trained individual walk around the fence to retrieve the food before attempting it themselves.

Why should this task be hard? Tortoises probably don't have very good advance planning skills. Let's be honest: adolescent humans don't have very good advance planning skills. And in order to reach the food, first you have to increase the distance between you and the food before you can turn the corner and then reduce the distance between you and the food. Without any real ability to plan for the future, increasing the distance between you and your goal probably doesn't make any real sense. Why would you fly from Los Angeles to Las Vegas with a layover in Atlanta? It only makes sense if that is the only way to get to Vegas.

First, the four non-observer tortoises (Alexandra, Wilhelmina, Esme, and Emily) were tested to see if they could figure it out on their own. Each animal was tested once per day, twelve days in a row. How did they do?

"But where are the bars?" you ask. I asked the same thing. I wondered if there had been some sort of error in the paper. No error. The data is all there. None of the tortoises succeeded. They all eagerly approached the fence, through which they could visually see the reward, but none of them EVER navigated the detour. Some of them tried to force their way THROUGH the barrier. Most just gave up and went to sleep.

Then, Wilhelmina was painstakingly trained to navigate around the right-hand side of the fence to get the food. It took over 150 trials for Wilhelmina to attain reliable performance. I would not want that job.

Then, the observer group had its turn. Each of the observer tortoises was placed into a cage and watched Wilhelmina successfully get her reward. Then, Wilhelmina was removed, the reward bowl was re-filled, and the bark that covered the ground was replaced (to make sure that the observers couldn't track her scent). Then the observer was released and allowed to retrieve the food.

All four tortoises in this condition had at least some success. The shaded bars represent successful retrieval by going around the fence on the right-hand side (as Wilhelmina did), and the white bars represent success by going around the fence on the left-hand side. In all, success was achieved 21 times by going to the right, and 8 times by going to the left. All the participants but Quinn had their first success by going right. Statistical tests showed that they went right significantly more than would have been expected by chance.

I think this is a great study because it showed that the red-footed tortoise, which do not live in social groups in the wild, are able to use social information to solve a problem. The fact that sometimes the observer turtles went around the fence on the left side suggests that they were not relying on some other environmental cue not seen by the experimenters - they had more generally learned a problem-solving strategy for that specific problem by watching Wilhelmina. Quinn even went left on his very first successful attempt, and both Moses and Aldous used both ways around.

This paper, as far as I know, is the first evidence of social learning in a non-social reptile, and it provides solid evidence that social living is not a prerequisite for social learning. Instead, it suggests that social learning may be the result of a more general ability to learn. The observer tortoises may simply have used Wilhelmina as just another source of information in the environment.

I think (and the researchers note as well) that the next steps are to see whether the turtles can also learn in this was from other species or from objects. What if the observers watched a person, or a dog, solve this problem? Or what if the observers watched a remote control submarine on wheels navigate around the fence? Would they figure it out? I think if they could do it under those conditions, then there may not actually be anything social to this, it may just be that Wilhelmina was another source of information in the environment. However, if they can only learn, or at least learn much better or faster, from other conspecifics, then I think that suggests a fundamentally social mechanism.

Wilkinson, A., Kuenstner, K., Mueller, J., & Huber, L. (2010). Social learning in a non-social reptile (Geochelone carbonaria) Biology Letters. DOI: 10.1098/rsbl.2010.0092