A blog post highlighting the article by S. Popp, P. Buckham-Bonnett, S.E.F. Evison, E.J.H. Robinson and T.J. Czaczkes in Insectes Sociaux

Written by Tomer Czaczkes and Sophie Evison

Anyone who has spent a few minutes watching ants running along a trail will have noticed that when two ants meet, they often interact for a second or two. It seems obvious to anyone watching that some sort of information transfer is occurring. But what could they be communicating? One of the most tempting hypotheses is that the returning ant is telling the outgoing ant where to go: “take the next left to a great patch of aphids!” in the way that the honey bee waggle dance conveys information about the location of food patches to other workers. However, as inviting as this hypothesis is, multiple investigations over the centuries have found no evidence to support an antennal language in ants. The widely accepted view is summed up by Hölldobler and Wilson in their bible of myrmecology (1990): “‘ants antennate nestmates in order to smell them, not to inform them’’.

However, from the mid 1990s, a Russian scientist named Zhanna Reznikova has been researching the communication skills and numerical competency of wood ants (family: Formicidae). Reznikova and colleagues report finding astounding physical communication abilities: not only could ants communicate a complex series of turns to nestmates by physical contact, they could also encode numbers (i.e. take the 27^th turn to the food”) (Reznikova and Ryabko 1994). Sophie Evison first heard about this during her doctoral studies from Reznikova herself, at a Central European Workshop on Myrmecology. Of course, she had to try this herself, so using the ants Evison had available at the time – Lasius niger – she carried out a (fairly rudimentary) replication of Reznikova’s experiment. Evison’s experiment simply tested whether returning L. niger foragers could tell outgoing foragers the correct direction to go at an upcoming T-maze. Amazingly, initial experiments appeared to show that these ants were communicating a form of directional information about a T shaped maze simply via antennation, with no other cues. However, to be candid, these results had similar impact to those of Reznikova’s, and the findings only ever appeared in Evison’s doctoral thesis.

Reznikova’s discoveries are incredible. Why were they not picked up? Simply put, no one believed them. However, incredulity is not a basis for scientific discourse. Many incredible scientific findings – in both the current and original meaning of the word – have gone on to be proven right, and changed the face of science forever. It seems incredible that we live on a spherical ball of rock, but it is true. Although before it was widely accepted, this fact had to be independently verified by repeated observations. The appropriate response from a modern but incredulous scientific community should be replication, not dismissal.

Why then did no one try to replicate these results? Why didn’t we, in our recent Insectes Sociaux study, try to replicate it exactly? The answer is simple: No incentives. Replication – the backbone of the scientific method – has no career rewards – especially for low traction ideas. We all remember the STAP cell fiasco, but do you remember who tried to replicate it and failed? Indeed, when we take into account the time lost performing such replications, they can reasonably be considered harmful to a career. This is especially true in the specific Reznikova case, as their method requires weeks of patient observation to define stable working ‘teams’ before tests can even begin.

Our recent study in Insectes Sociaux has convinced us that, at least in two Lasius species, on-trail physical interactions do not communicate direction, but these are only two species. This increases the burden of proof for such physical communication, but does not rule it out. Of course, even though we set out to find evidence for such communication, we also find the results of Reznikova incredible. And while incredulity alone is not a basis for scientific discourse, it is far from meaningless. The collective intuition of our research community should not be ignored. Perhaps we should be taking a Bayesian approach, adjusting our demands for evidence by our level of incredulity. There an important part of this story that helped to keep our collective investigation going all these years: during a Royal Society event, the artificial intelligence expert Donald Michie mentioned to Elva Robinson that he had spent some time with Reznikova, and was very interested by her results. He had planned to investigate the claims further, but tragically lost his life in a car accident. Donald’s perspective on the work of Reznikova was important; it provided an externally driven impetus to resolve our contrasting findings.

So what would it take to convince us of the Reznikova findings? We propose replication by an unaffiliated research team, with meticulous video documentation. But let’s face it – for the reasons stated above, we don’t think this is going to happen. We hope to be proven wrong.