Fig 1.
Pheromone depositions were recorded on the T maze stems (shaded grey). Ants were considered to have chosen left or right at a bifurcation when they crossed the relevant decision line (dashed lines). A syrup feeder (1M sucrose) was placed on a platform at either the left-right or right-left end of the maze (left-right treatment shown). In the confined maze treatments, 1 cm gaps prevented ants from making incorrect turns (hashed sections). Adapted from Czaczkes et al. [28].
Fig 2.
The effect of pheromone and maze treatment on learning.
Ants were trained either on an open maze or on a confined maze, and pheromone was either removed or allowed to remain. Note that on the final (7th) visit shown, all ants are tested on an open maze with no pheromone information. The presence of pheromone during the previous training visits had no effect on learning, and confinement caused ants to make more mistakes at the first bifurcation. Symbols represent means, whiskers 95% Cis.
Fig 3.
The effect of four treatments and number of previous visits on pheromone deposition (A&B), and the link between decision-making and pheromone deposition (C).
Fewer pheromone depositions are performed on the outward journey in the open treatment (A). On the return journey pheromone deposition was lower if pheromone was allowed to build up, and was higher if pheromone was continually removed (B). Outgoing ants deposited less pheromone if they would go on to make a mistake on their upcoming decision (C). Returning ants deposited more pheromone if they had just made a mistake on the current visit (C). This is driven by the likelihood of depositing pheromone at all, rather than by pheromone deposition intensity (fig A in S1 File). Data from both ants which did and did not deposit pheromone are merged in this figure. Data from all open treatments, and both bifurcations, have been merged for clarity. Symbols represent means, whiskers 95% CIs. Whiskers have been omitted in A and B for clarity.