We present a protocol for the classical conditioning of harnessed ants that permits researchers to study visual learning and memory formation at a level of analysis not possible with freely moving individuals.
Several species of insects have become model systems for studying learning and memory formation. Although many studies focus on freely moving animals, studies implementing classical conditioning paradigms with harnessed insects have been important for investigating the exact cues that individuals learn and the neural mechanisms underlying learning and memory formation. Here we present a protocol for evoking visual associative learning in wood ants through classical conditioning. In this paradigm, ants are harnessed and presented with a visual cue (a blue cardboard), the conditional stimulus (CS), paired with an appetitive sugar reward, the unconditional stimulus (US). Ants perform a Maxilla-Labium Extension Reflex (MaLER), the unconditional response (UR), which can be used as a readout for learning. Training consists of 10 trials, separated by a 5-minute intertrial interval (ITI). Ants are also tested for memory retention 10 minutes or 1 hour after training. This protocol has the potential to allow researchers to analyze, in a precise and controlled manner, the details of visual memory formation and the neural basis of learning and memory formation in wood ants.
Insects have been extensively used as models for studying learning and memory formation1. A particularly successful vein of research involves using classical conditioning of restrained animals, which allows precise control over the cues being learned and permits researchers to investigate the neural mechanisms underpinning learning and memory. The majority of studies have focused on the appetitive classical conditioning of honeybee workers, Apis mellifera. The honeybee workers are trained to associate a CS with a US that evokes the UR. In this paradigm, originally developed by Takeda2 and Bitterman et al.3, the UR is the Proboscis Extension Reflex (PER), the US is sugar, and the CS is an odor. Bees learn the association between the CS and the US and can form a long-term memory of this association.
The original paradigm using the PER as the UR has been used to unravel the neural pathways underpinning olfactory learning in bees4. It has been modified in several ways to test learning of different sensory stimuli, including visual stimuli5,6,7, to incorporate aversive learning using suppression of the PER8 or the Sting Extension Reflex (SER)9 as the UR, and to test learning in other species, such as bumblebees10 and fruit flies11. Although memory formation with several modalities has been studied through classical conditioning, visual learning is still difficult to observe using this approach, even in species that show a high degree of visual learning ability during foraging, such as honeybees.
Recent studies have applied a similar approach to insects that do not have a proboscis, such as locusts, which perform the Palp Opening Response (POR)12, and ants, which perform the Maxilla-Labium Extension Reflex (MaLER)13. This has already revealed phase-specific learning abilities that match the phase-specific feeding strategies of the two different desert locust phenotypes, the gregarious and the solitarious forms, re-enforcing the idea that memory formation needs to match ecological needs14. Furthermore, studies on olfactory learning in ants has shown similarities between ants and honeybees in memory formation and retention, with long-term memory retention, 72 hours after training, being dependent on protein synthesis15.
These adaptations of the original paradigm have allowed learning and memory formation to be studied with many modalities and in several model species. This is essential for identifying mechanisms of memory formation common to the insects but also for identifying differences that reflect the particular ecology of learning and memory in different species. The main goal of the protocol we describe here is to provide a way to perform classical conditioning experiments using a visual conditional stimulus with a widely studied ant species, Formica rufa. It is issued from our study of visual learning in wood ants16, which is also an adaptation of visual classical conditioning paradigms.
Classical conditioning is one of the most well-established paradigms to study learning and memory. The protocol we presented here is an adaptation of the paradigm designed for honey bee workers2,3 and subsequently used with several other species, such as bumblebees, fruit flies, which also use PER as a readout for learning10,11, and locusts and ants, which use POR and MaLER, respectively12,13. Using this protocol, it is possible to train harnessed wood ants to learn the association between a visual cue and a sugar reward and analyze the retention of this short- (10 min) and middle-term (1 hour) memory16.
In any behavioral experiment, it is necessary to take into account critical steps that can minimize the variability in the animals' responses. In the protocol presented here, several measures are taken to minimize variability before and during training. Prior to the start of experiments, the colony needs to be starved for at least two days and ants should be selected based on their willingness to eat from a sugar drop in the holding box. Selecting ants in this way is intended to maximize the chance of training ants that are motivated to feed. Careful handling is also an important consideration because it can help to reduce stress levels, which disrupts learning if it is too intense21. To this end, ants should be anesthetized with cold to stay motionless whilst being harnessed, because any movement (for escaping) during this procedure could be a source of stress. Furthermore, the contact between the ant and the wax should be minimal, avoiding contact between the antennae and the hot wax or wire, which could cause damage. Although these observations have not been analyzed formally, the antennae seemed to move with a specific pattern during learning.
During experiments, careful delivery of the sugar is also important to keep ants motivated. Again, whilst this has not been analyzed formally, abrupt food delivery seemed to cause additional stress to the ant, which in turn led to a lack of motivation and learning. Furthermore, the sucrose delivered during the training should be of a reduced concentration (200 g/L) to avoid satiation before the end of the training and testing. This allows MaLER to be a good candidate unconditional response because, together with a low spontaneous performance of this response to the visual cue, it also does not saturate over trials. Lastly, contrary to most classical conditioning studies2,3,5,6,7,8,9,10,11,12,13, we trained one ant at the time until the end of the experiment, leaving it in place between trials rather than removing it to test another ant. Training several ants together seemed to produce more variable results, which may be due to an increase in stress and/or conflict between visual information caused by the complete change of the scenery. To reduce the duration of each experiment, we used a 5-minute ITI instead of the 10-minute ITI used in most classical conditioning studies16. Although all these considerations should help ensure that the ants are motivated to feed and learn during training, some variability cannot be avoided. We recommend using ants that seem to have normal social, appetitive, and locomotion behavior and excluding ants from the analysis the moment they fail to feed on a training trial or a test.
The nature of the CS was not tested in this study. Although we have used a blue visual stimulus because ants of the same genus are sensitive to these wavelengths18, other colors might also be learned in association with a reward. Further experiments would be required to fully characterize the colors being seen and learned in this set-up. This is also true for different shapes and sizes of the visual cue. We have not tested if ants' spatial resolution would be sufficient for distinguishing the visual stimulus presented here at the distance from the ants' eyes it was presented at. Although wood ants' compound eyes have been described in terms of size and number of facets22, to our knowledge, their spatial resolution has not been fully described yet. However, this has been calculated for Melophorus magoti23. A similar characterization of the wood ants', or other tested insects' eyes would contribute to a clear investigation of the features of the visual cue being observed and learned by the animals. Furthermore, we included motion when presenting the visual stimulus to the ant because it has been shown to play a role in honeybee associative learning during classical conditioning6. However, this was also not tested in this study and, due to the different movement nature of flying insects compared to walking insects, differences between honeybee and ant visual classical conditioning could be observed.
On a final note, we were unable to examine long-term memory retention because ants did not survive being harnessed for such long periods after training. However, in subsequent sets of experiments, we have kept ants alive and motivated to eat and learn when harnessed and left them overnight in a dark and humid environment (placing a box over them). Therefore, this paradigm could be used to unravel long-term memory retention of wood ants, in addition to short- and middle-term memory.
With this simple procedure adapted from general classical conditioning paradigms, it is possible to study the acquisition and retention of visual memories in harnessed wood ants, which have been studied widely in paradigms using free-moving animals. This paradigm has the potential to be used for analyzing the neural basis of visual learning in a very well-established model for insect navigation.
The authors have nothing to disclose.
The authors thank Tom Collett and Cornelia Buehlmann for sharing information regarding collection and maintenance of wood ant colonies. The authors also thank Justine Crevel for commenting on previous versions of this article, and Nora Nevala for measuring the spectrum intensity of the visual stimulus. This work was supported by a BBSRC grant to JEN (grant number BB/R005036/1). All the data pertaining to this manuscript are published in the University of Sussex Research Data Repository online database (10.25377/sussex.5794386).
Fluon | Blades Biological Ltd, Edenbridge, UK | ACS 109; ACS 112; ACS 114 | For preventing insects from scaping |
Crickets | Blades Biological Ltd, Edenbridge, UK | LZJ 217 | Given to the ant colonies as protein source |
Natural Pine Rosin/Resin | Minerals-water Ltd, Rainham, UK | 500g | Given to the ant colonies for sanitation |
Austerlitz Insect Pin | Fine Science Tools GmbH, Heidelberg, Germany | 26000-40 | For harnessing ants |
High speed camera | Edmund Optics Inc., Barrington, USA | eo-13122M | MaLER recordings during training and testing |
Macrolens | Cannon, Surrey, UK | EF 100 mm f/2.8 L Macro IS USM | MaLER recordings during training and testing |
Software | IDS Imaging Development Systems GmbH | uEye64 | MaLER recordings during training and testing |
Blue Cardboard | john smith's at Union Store, University of Sussex | JACK-PJM41358 | Constitutes de conditional stimulus |
Syringe | Fisher Scientific LTD, Loughborough, UK | BD Plastipak 300185case; Product Code.12369289 | US and CS (attached) presentation |
Needle (0.5 x 16 mm) | Fisher Scientific LTD, Loughborough, UK | BD Microlance 300600; Product Code:10442204 | US and CS (attached) presentation |