Meerkat teaching

It's all meerkats all over the place today. Here's an AP article by Randolph Schmid:

Researchers from the University of Cambridge in England observed meerkats gradually introducing cubs to prey, showing them how to handle captured insects and even removing the stingers from scorpions before giving them to youngsters.
"Although there are anecdotal reports of teaching in species from chimpanzees to killer whales, until this year solid evidence was really lacking," said Alex Thornton, co-author of the report appearing in Friday's issue of the journal Science.

Here's the article by Thornton and McAuliffe. Much hangs on the definition of "teaching", in this paper that definition really comes at the end:

The results of this study provide strong evidence that the provisioning behavior of meerkat helpers constitutes a form of "opportunity teaching," in which teachers provide pupils with opportunities to practice skills, thus facilitating learning (3, 7). Helpers modified their behavior in the presence of pups, gradually introducing them to live prey, monitoring their handling behavior, nudging prey, and retrieving and further modifying prey if necessary. Dangerous items were more likely to be killed or disabled than other mobile prey. Helpers gained no direct benefits from their provisioning behavior and incurred costs through giving pups prey that was difficult to handle and might escape. Finally, there was strong evidence that helper provisioning behavior plays an important role in promoting the development of pup handling skills.
It is often assumed that teaching requires awareness of the ignorance of pupils and a deliberate attempt to correct that ignorance (5, 6, 20), but viewed from a functional perspective (3), teaching can be based on simple mechanisms without the need for intentionality and the attribution of mental states. By differentially responding to the calls of pups of different ages, helpers may accelerate pups' learning of handling skills without the need for complex cognitive processes. Additional post-provisioning behavior, such as nudging and retrieving prey, may then further enhance skill acquisition.

In the last paragraph, Thornton and McAuliffe place the meerkats in a broader context:

Evidence from ants (10) and meerkats suggests that teaching, as defined by Caro and Hauser (3), may have evolved independently in many unrelated taxa. Where individuals must acquire critical skills or information but individual learning is costly or opportunities to practice are lacking, selection may favor mechanisms whereby experienced individuals actively facilitate learning by naïve conspecifics. The paucity of evidence for teaching is likely to reflect difficulties in producing unequivocal support for strict criteria rather than an absence of teaching per se. As evidence for teaching in nonhuman animals emerges, research will be in a position to look in more detail at the conditions under which teaching is likely to evolve and to relate forms of teaching found in humans and other animals in a broad framework.

Personally, I think that "theory of mind", insofar as it exists, must consist precisely of the sorts of knowledge that allow teaching to happen. In other words, you have to be able to assess whether another individual "gets" the information you are trying to send with your efforts. And from that perspective, the only distinction between teaching and communication is the kind of effort that goes into it -- which in the case of teaching is generally either progressive or repetitive. Those kinds of cases are the ones that fulfill "strict criteria", at least; a simple communication may convey just as much information or more (to a suitably primed individual), but has less oomph to it in examining the behavior of the teacher.

But then, there are the ants. In case you missed that paper, here's the abstract:

The ant Temnothorax albipennis uses a technique known as tandem running to lead another ant from the nest to food--with signals between the two ants controlling both the speed and course of the run. Here we analyse the results of this communication and show that tandem running is an example of teaching, to our knowledge the first in a non-human animal, that involves bidirectional feedback between teacher and pupil. This behaviour indicates that it could be the value of information, rather than the constraint of brain size, that has influenced the evolution of teaching.

The key element in that definition of teaching is the bidirectional feedback. And bidirectional feedback does at least involve the concept of being able to read the signs that a learner is responding appropriately. For the ants, there is little information being exchanged to maintain the system. In a real sense, the teacher ant doesn't have to have very much information about the learner to assess that the learner is responding appropriately. We could imagine how a robot might be built to "teach" this kind of task. It's essentially what Lassie does whenever Timmy gets stuck in trouble somewhere.

But by the same token, the meerkat teachers may not need much information to judge their learners' responses. They just repeatedly present food to the learners; and the learners' eating, playing or whatever give the feedback. Now, there is something more complicated here -- the learner is learning signs. The desired outcome is that certain animals will be recognized by the learners as food, which requires the involvement of specialized visual, olfactory, and other cognitive subsystems. But the teachers aren't creating this entire system; they are only presenting a consistent set of associations. Nor do the teachers have to judge the workings of every stage of cognition -- "theory of mind" need not go anywhere near that far. They just have to rachet the teaching when they perceive appropriate responses by the learners.

So I think this is hardly less robotic than the ants. The additional complexity is not in the communication between teacher and learner, but in the perceptual systems necessary to carry out the communication.

And I think that's the point -- the feedback communication that enables the evolution of teaching (at least, this kind of teaching) is pretty simple (that is, it doesn't require a large channel). The story is that mammal brains (and primate brains in particular) do a very good job of abstracting this narrow channel of communciation out of the very broad sensory inputs available (vastly moreso than the chemical and tactile system used by ants). And that they are then so successful at using the limited communications to model other individuals.


Thornton A, McAuliffe K. 2006. Teaching in wild meerkats. Science 313:227-229. DOI link

Franks NR, Richardson T. 2006. Teaching in tandem-running ants. Nature 439:153. PubMed