How Do You Figure Out How Chimps Learn? Peanuts.

What is culture? One simple definition might be: a distinctive behavior shared by two or more individuals, which persists over time, and that ignorant individuals acquire through socially-aided learning.

There are at least four different ways to learn a particular behavior or problem-solving strategy. That is to say, there are four different ways to learn. The first is social facilitation, in which one individual does the same thing as the demonstrator at the same time. Essentially this is a situation of on-line matching of motor actions. For example, I might learn the steps to a complicated dance by watching somebody else doing it and replicating the motor actions at the same time. The second is emulation, in which the learner replicates the demonstrator's goal, without replication of the precise motor actions. If you see me climb a tree to retrieve a candy-filled pinata, you might figure out that an easier way to retrieve the candy would be to smack the pinata with a baseball bat. Importantly, you have deduced my goal from my behavior (I want candy), as well as the results (I get candy). The third is imitation in which after observing the demonstrator solve a task, the learner precisely replicates the novel motor actions, in the absence of the demonstrator. Finally, the fourth is teaching, or pedagogy in which a knowledgeable individual provides active instruction to an ignorant individual. It was through teaching that I learned how to tie a tie, for example, or to cook a steak to a perfect medium done-ness.

Since learning underlies culture, it is important to understand how different species learn, if we wish to understand the evolution of culture.

Chimpanzees, for example, appear as if they have at least some sort of culture. Among six groups of chimpanzees in Africa, for example, the different groups have various methods for throwing, inspecting wounds, inspecting parasites, and so forth. Some groups don't show evidence of those behaviors at all. These behaviors are not species-specific, and instead appear to be group-specific. In other words, a basic form of culture.

However, when you see a population of individuals doing the same thing, it is a mistake to assume that they acquired the behavior in the same way. For an analogy from biology, consider eyes. Both mammals and cephalopods have eyes, but they evolved separately, and are morphologically distinct. Likewise, just because two groups of chimps (or people) show similar behavior does not mean that they came about the same way, or are constructed of the same cognitive building blocks.

How do you figure out how individuals acquire novel behaviors? A paper that was published today in the journal PLoS ONE describes one method: Peanuts.

Chimpanzees really like peanuts. Here's the plan: create a long clear tube, and drop a peanut into it. Bolt it to a chimp cage, so that the ape will have to get creative to retrieve the tasty snack. Have some chimpanzees spontaneously figure out how to get the peanut. The only possible solution is to suck water into your mouth, and spit it into the tube. Over and over again, until the peanut floats to the top. Success. Reap your nutty reward. Ignore the fact that your nuts are soggy.

The researchers did this with thirty-one socially housed chimps, in two different locations. Some of them came from the Ngamba Island Chimpanzee Sanctuary in Uganda, and some lived in the Wolfgang Kohler Primate Research Center in Leipzig Zoo.

Most of the chimps were unable to spontaneously come up with the correct solution, but a few were.

Now comes the experiment.

In one condition, subjects watched four to six successful demonstrations of the solution by another chimp. The subject and demonstrator were always in the same cage, allowing the subject to approach and closely inspect the process. The demonstrator was always dominant to the subject, so that the subject would watch but not interfere. In this condition, subjects could invent the solution spontaneously, they could copy the actions of water spitting (imitation), or they could copy only the results of the demonstrator's actions (emulation).

The second condition was identical, except for the following: A solution-naive but dominant conspecific "stooge" chimp was used as the social partner in this condition. The experimenter demonstrated an alternative solution to water-spitting, by pouring water into the tube from a bottle from outside the cage. The dominant "stooge," by virtue of being dominant, still gets the peanut. In this way, the conditions are identical except for the method by which the dominant chimp gets the nut.

In this condition, the subject was able to witness the results producing the solution - adding water to the tube - but without any specific actions that the subject could mimic to solve the task. They could spontaneously come up with the solution, or they could emulate (but not imitate). Therefore, the key difference between conditions 1 and 2 is whether or not the chimp can use imitation.

Both experimental conditions showed significantly more successes than in the baseline condition (in which the only way to solve the problem was spontaneously inventing the solution). Five subjects solved the problem in condition 1, and three in condition 2. Subsequent analyses indicated that successful subjects were not successful by virtue of being more motivated, though successful individuals were younger than unsuccessful subjects. There were no differences between the two groups from Uganda and Leipzig. Further, there were no significant differences between the two experimental conditions. So, the demonstrations of condition 1, which included three types of information - actions, goals, and results - offered no advantage over the demonstrations of condition 2, which only included two types of information - results and goals.

Since action information did not provide an advantage, the best interpretation of these results is that the successful chimps used emulation learning to solve the task. Is it therefore possible that copying of goals and results, and not action, could underlie cultural traditions in apes? That does seem to be the best explanation, given these findings, which is also consistent with previous findings in which apes ignored superfluous actions and only imitated relevant actions to retrieve a reward. This also provides a reasonable explanation for the variation among chimp groups in Africa, each with slightly different solutions to their problems.

The finding that younger chimps were more successful than older individuals is also consistent with previous findings. The age of the successful learners (4-6 years) coincides with the age at which chimps are observed using tools in the wild. In fact, primate researcher Michael Tomasello (who was one of the authors of this paper), has previously suggested that there may be a sensitive period underlying the ability of chimps and other apes to use tools, much like the sensitive period for learning language in human children. This would explain why the older subjects in this and other studies failed to successfully solve the peanut problem, even after social demonstrations.

One other implication of this paper is important to mention, and that is the social nature of this learning. Previous studies have done experiments very similar to condition 2, in which the learner watched a human fill the tube with water to make the peanut float, but without another conspecific present, and in those experiments, apes tend to be unsuccessful in learning the solution to the problem. While in this experiment, results and goals were sufficient to learn the task, it may be that the results themselves are not what is driving the pattern. Instead, perhaps it is the addition of the goal-related information from the social partner (getting to eat the reward) that allows for successful emulation, making social information critical to emulation learning.

Tennie, C., Call, J., & Tomasello, M. (2010). Evidence for Emulation in Chimpanzees in Social Settings Using the Floating Peanut Task PLoS ONE, 5 (5). DOI: 10.1371/journal.pone.0010544