Robot teaches tandem running to ants

It’s not just us humans that sit in class and pick up wisdom from a teacher: ants do it too. As ant expert Nigel Franks explains to Will Tingle, they learn their way to better food and nest sites through a process called ‘tandem running’ where a teacher ant with the know-how shows a pupil the way. To explore how this works, they built an undercover robotic ant.…

Nigel - Our demonstration was the first rigorous evidence of teaching in non-human animals, but we used the definition of an individual as a teacher who, firstly, modifies its behaviour in the presence of a naive observer. Secondly, it's some initial cost to itself. Thirdly, in order to set an example so that, fourthly, the other individual can learn more quickly. So they had these four clauses and essentially, in tandem running in ants, the tandem leader qualifies under all four criteria. So we were able to demonstrate according to this rigorous definition that tandem running qualified as teaching among non-human animals.

Will - So to better understand this behaviour, for all intents and purposes, a robot ant was made. Why was this deemed necessary for the experiment?

Nigel - I think if you can build something that works, it does show that you've mostly understood what is crucial to the process. The robotic ant is not an ant that any one of us would recognise, but thankfully the ant does recognise it. So essentially it was a pinhead attached to quite a large gantry, a sort of overhead device to move things in two dimensions across a surface. And the pinhead was a dawned with the appropriate pheromones from a donor ant. And those are the pheromones that the tandem leader would normally deploy to get a naive individual to follow it. And the crucial thing was that the very artificial robot was able to teach the ant the route. So not only were we able to lead an ant from the old nest to a much better one, but we were able to observe that ant having surveyed the new nest and probably decided that it really is a valuable resource. We could track it on its return route and show that it had really learned a very efficient path home.

Will - What was significantly discovered about how ants teach one another?

Nigel - The thing with all these experiments is that the thing that most enlightens one is the unexpected, and when real ants lead a tandem run, they tapped their gaster, and the gaster is their bottom for want of a better expression, they tap their gaster periodically on the substrate and almost certainly leave a trail. Now, the intriguing thing is that neither the leader nor the followers slavishly followed that trail on the way home, but maybe it's a sort of safety line. But we didn't have the robotic leader doing this. What we noticed was that almost certainly the following ant in the absence of this trail, laid by the leader, started laying a trail itself. And I think the wonderful thing about that is that normally a tandem follower, if it really likes a new nest, will do a tandem run itself and will become a leader and will lay a trail. So if it was deploying this behaviour, to put it very anthropomorphically, blimey, this robot isn't doing a very good job and therefore I'll take over laying a trail myself. At this point, I would love to believe that it was evidence of yet another fail safe mechanism among the ants. When I was running the lab at Bristol, my chief role in life was to try and confuse ants and get them to exhibit odd behaviours. In a sense, every day of our experiment lives we were setting them puzzles to solve. And I can only remember one or two occasions when they were unable to solve those puzzles. And the reason they were so good at it is that they have backup mechanisms that enabled them to solve problems. They are marvellous creatures. They really are.