Special types of learning

Across animals, many learning processes seem to follow similar characteristics, such as temporal discounting, where the time elapsed since an event changes the strength of the memory update. Here, Lionetti and Cheng investigate whether an important navigational process also follows a similar process. By offsetting outward from return journeys on an ants foraging trip, you can induce a recalibration of Path Integration mechanisms. By varying the delay between these two legs of the journey, one might expect different degrees of re-calibration, but in fact no such difference is observed. This suggests that not all learning and updating processes follow the same general principles.

Abstract: “Desert ants navigate by using two chief strategies: path integration, keeping track of the straight‐line distance and direction to the starting point as they travel, and landmark guidance, orientation based on the visual panorama. Both Cataglyphis ants in North Africa and Melophorus bagoti in Central Australia are known to adjust their vectors derived from path integration to compensate for mismatches between their outbound direction of travel and (the reverse of) the inbound direction of travel that takes them home, a process known as vector calibration. We created mismatches of 90° between the outbound vector and the homebound direction by displacing ants from a feeder before their homebound run. We examined temporal factors in vector calibration by varying the delay (0, 1 or 3 hr) between the outbound run to the feeder and the homebound run from the displacement site. According to the temporal weighting rule, such a delay should decrease the weight given to the vector information obtained from the outbound run. This in turn should favour reliance on the visual panorama and thus speed up calibration. Results did not support this prediction. At the displacement site, a delay had little effect on the extent of calibration or the speed of calibration (the number of trials to reach maximum calibration). Just before being displaced, ants were also tested in a test ring surrounded by high walls that obliterated the visual scenery. In the test ring, a delay made the ants less likely to rely on their vector: ants were often not oriented as a group. Otherwise, the ants in the test ring also did not calibrate any more or any faster.”

 

Lionetti, V. A., & Cheng, K. Vector calibration in Australian desert ants, Melophorus bagoti: Effects of a delay after the acquisition of vector information. Ethology.

 

 

 

 

Back once again

“If you find yourself lost, then go back the way you just came” This sounds like the kind of advice one might get from a cautious aunt, but experiments have shown that it is a common and useful strategy for navigating ants. Freas et al extend this knowledge base, by looking at the behaviour in ants that had been using social as well as individual information prior to getting lost.

Abstract: “Diverse species may adopt behaviourally identical solutions to similar environmental challenges. However, the underlying mechanisms dictating these responses may be quite different and are often associated with the specific ecology or habitat of these species. Foraging desert ants use multiple strategies in order to successfully navigate. In individually foraging ants, these strategies are largely visually-based; this includes path integration and learned panorama cues, with systematic search and backtracking acting as backup mechanisms. Backtracking is believed to be controlled, at least in solitary foraging species, by three criteria: 1) foragers must have recent exposure to the nest panorama, 2) the path integrator must be near zero, and 3) the ant must be displaced to an unfamiliar location. Instead of searching for the nest, under these conditions, foragers head in the opposite compass direction of the one in which they were recently travelling. Here, we explore backtracking in the socially foraging desert harvester ant (Veromessor pergandei), which exhibits a foraging ecology consisting of a combination of social and individual cues in a column and fan structure. We find that backtracking in V. pergandei, similar to other solitary foraging species, is dependent on celestial cues, and in particular on the sun’s position. However, unlike solitary foraging species, backtracking in V. pergandei is not mediated by the same criteria. Instead the expression of this behaviour is dependent on the presence of the social cues of the column and the proportion of the column that foragers have completed prior to displacement.”

Freas, C. A., Congdon, J. V., Plowes, N. J., & Spetch, M. L. (2019). Same but Different: Socially foraging ants backtrack like individually foraging ants but use different mechanisms. Journal of Insect Physiology, 103944.

Social and individual navigation

Many ant species have public navigation information via pheromone trails, as well as private information from Path Integration and learning. The interaction between the two is fascinating. One nice result property of Harvester ant navigation is that individuals will travel along a trunk trail before then fanning out on individual routes, thus showing a public to private transition within a single route. Ants in this kind of foraging system actually show an interesting tweak in their PI system, whereby they aim for the head of the trunk trail rather than the nest. However, Freas et al show now that such ants are still sensitive to celestial compass information even when they are moving along the trunk trail. This suggests that either ants are Path Integrating, even when on the trail, or are perhaps using celestial compass information to disambiguate the two potential directions of the pheromone trail.

Of course there is the potential that private and public information might conflict. Czaczkes et al. here analyse the way that preferences for private information can be switched to public information when the trail has information about resource quality. Abstract: “When personally gathered and socially acquired information conflict, animals often prioritize private information. We propose that this is because private information often contains details that social information lacks. We test this idea in an ant model. Ants using a food source learn its location and quality rapidly (private information), whereas pheromone trails (social information) provide good directional information, but lack reliable information about food quality. If this lack is indeed responsible for the choice of memory over pheromone trails, adding information that better food is available should cause foragers to switch their priority to social information. We show it does: while ants follow memory rather than pheromones when they conflict, adding unambiguous information about a better potential food source (a 2 µl droplet of good food) reverses this pattern, from 60% of ants following their memory to 75% following the pheromone trail. Using fluorescence microscopy, we demonstrate that food (and thus information) flows from fed workers to outgoing foragers, explaining the frequent contacts of ants on trails. Ants trained to poor food that contact nest-mates fed with good food are more likely to follow a trail than ants which received information about poor food. We conclude that social information may often be ignored because it lacks certain crucial dimensions, suggesting that information content is crucial for understanding how and when animals prioritize social and private information.”

Freas, C. A., Plowes, N. J., & Spetch, M. L. (2019). Not just going with the flow: foraging ants attend to polarised light even while on the pheromone trail. Journal of Comparative Physiology A, 1-13.

Czaczkes, T. J., Beckwith, J. J., Horsch, A. L., & Hartig, F. (2019). The multi-dimensional nature of information drives prioritization of private over social information in ants. Proceedings of the Royal Society B, 286(1909), 20191136.

Keeping to the straight and narrow

Here we have a really nice example of the productive interaction that can arise from modelling and behavioural experiments. Looking at different size beetles, that use the same compass information, allows an understanding of how beetles account for errors coming from motors (stepping) and compass sensors as they try to keep to a straight line course.

Abstract: Moving along a straight path is a surprisingly difficult task. This is because, with each ensuing step, noise is generated in the motor and sensory systems, causing the animal to deviate from its intended route. When relying solely on internal sensory information to correct for this noise, the directional error generated with each stride accumulates, ultimately leading to a curved path. In contrast, external compass cues effectively allow the animal to correct for errors in its bearing. Here, we studied straight-line orientation in two different sized dung beetles. This allowed us to characterize and model the size of the directional error generated with each step, in the absence of external visual compass cues (motor error) as well as in the presence of these cues (compass and motor errors). In addition, we model how dung beetles balance the influence of internal and external orientation cues as they orient along straight paths under the open sky. We conclude that the directional error that unavoidably accumulates as the beetle travels is inversely proportional to the step size of the insect, and that both beetle species weigh the two sources of directional information in a similar fashion.

Khaldy, L., Peleg, O., Tocco, C., Mahadevan, L., Byrne, M., & Dacke, M. (2019). The effect of step size on straight-line orientation. Journal of the Royal Society Interface, 16(157), 20190181.

Complex spatial memory in flies

Many aspects of spatial orientation have been studied in flies, where genetic tools have enabled a deep level of understanding of the neural circuits involved in direction setting. However, navigation, in the broadest sense, is a complex behaviour. Here, Stern et al use optogenetic reward in a spatial assay, showing flies will learn a location within an arena that is “rewarding”. Probing of these flies shows that both blind and sighted flies can learn the task. With the Central Complex more important for sighted flies and the Mushroom Bodies for blind flies, who seem to be relying on tactile and chemical features of the arena. Overall this highlights that navigation is not the product of a single brain area and that the information from different modalities, that together contribute to navigation, may be processed in a distributed manner.

Summary: ” The ability to use memory to return to specific locations for foraging is advantageous for survival. Although recent reports have demonstrated that the fruit flies Drosophila melanogaster are capable of visual cue-driven place learning and idiothetic path integration [1, 2, 3, 4], the depth and flexibility of Drosophila’s ability to solve spatial tasks and the underlying neural substrate and genetic basis have not been extensively explored. Here, we show that Drosophila can remember a reward-baited location through reinforcement learning and do so quickly and without requiring vision. After gaining genetic access to neurons (through 0273-GAL4) with properties reminiscent of the vertebrate medial forebrain bundle (MFB) and developing a high-throughput closed-loop stimulation system, we found that both sighted and blind flies can learn—by trial and error—to repeatedly return to an unmarked location (in a rectangularly shaped arena) where a brief stimulation of the 0273-GAL4 neurons was available for each visit. We found that optogenetic stimulation of these neurons enabled learning by employing both a cholinergic pathway that promoted self-stimulation and a dopaminergic pathway that likely promoted association of relevant cues with reward. Lastly, inhibiting activities of specific neurons time-locked with stimulation of 0273-GAL4 neurons showed that mushroom bodies (MB) and central complex (CX) both play a role in promoting learning of our task. Our work uncovered new depth in flies’ ability to learn a spatial task and established an assay with a level of throughput that permits a systematic genetic interrogation of flies’ ability to learn spatial tasks.  ”

Stern, U., Srivastava, H., Chen, H. L., Mohammad, F., Claridge-Chang, A., & Yang, C. H. (2019). Learning a Spatial Task by Trial and Error in Drosophila. Current Biology.

Learning Walks

I have always been a little confused as to why we knew more about the learning flights of wasps and bees than the learning walks of ants. Not least because the latter seem easier to record. However the imbalance has been reduced recently by lots of exciting work from the authors of this new review. In describing recent findings and comparing it to established findings on flying insects, this review represents a thorough and very useful manuscript. I particularly like the ‘open questions’ section at the end of the paper: a wish-list for future experiments!

Zeil, J., & Fleischmann, P. N. (2019). The learning walks of ants (Hymenoptera: Formicidae). Myrmecological News, 29.

How do old bees learn new tricks?

In the insect navigation literature there is a common story of the naive individual forager who explores the world and eventually, after finding rewarding locations, fixes on a particular idiosyncratic foraging route that reflects their history of reward. Here, Kembro et al. analyse the foraging histories of many individual bees in an artificial meadow. Unsupervised (automatic) statistical analyses of these bee’s visits to feeders show that they transition between local behaviours, fixed routes and new exploration. In an example of the ‘ecological fallacy’, the expected pattern of bees drifting from local behaviour to explorative bouts to fixed foraging routes is only true at the population level, but all individuals show much more flexibility in the structure of their transitions between the types of foraging route. These Machine Learning methods, when allied to very rich data sets of individual behaviour promise to be very important in moving towards a richer understanding of individual decision making within foraging.

Abstract: “How animals explore and acquire knowledge from the environment is a key question in movement ecology. For pollinators that feed on multiple small replenishing nectar resources, the challenge is to learn efficient foraging routes while dynamically acquiring spatial information about new resource locations. Here, we use the behavioural mapping t-Stochastic Neighbouring Embedding algorithm and Shannon entropy to statistically analyse previously published sampling patterns of bumblebees feeding on artificial flowers in the field. We show that bumblebees modulate foraging excursions into distinctive behavioural strategies, characterizing the trade-off dynamics between (i) visiting and exploiting flowers close to the nest, (ii) searching for new routes and resources, and (iii) exploiting learned flower visitation sequences. Experienced bees combine these behavioural strategies even after they find an optimal route minimizing travel distances between flowers. This behavioural variability may help balancing energy costs–benefits and facilitate rapid adaptation to changing environments and the integration of more profitable resources in their routes.”

Kembro, J. M., Lihoreau, M., Garriga, J., Raposo, E. P. & Bartumeus, F. 2019 Bumblebees learn foraging routes through exploitation-exploration cycles. Journal of The Royal Society Interface 16, 20190103. doi: doi:10.1098/rsif.2019.0103. Kembro1 2019