Skip to main content
Dryad logo

Lateralisation of short- and long-term visual memories in an insect


David Fernandes, Ana Sofia; Niven, Jeremy (2020), Lateralisation of short- and long-term visual memories in an insect, Dryad, Dataset,


The formation of memories within the vertebrate brain is lateralised between hemispheres across multiple modalities, however, in invertebrates evidence for lateralisation is restricted to olfactory memories, primarily from social bees. Here we use a classical conditioning paradigm with a visual conditioned stimulus to show that visual memories are lateralised in the wood ant, Formica rufa. We show that a brief contact between a sugar reward and either the right or left antenna (reinforcement) is sufficient to produce a lateralised memory, even though the visual cue is visible to both eyes throughout training and testing. Reinforcement given to the right antenna induced short-term memories whereas reinforcement given to the left antenna induced long-term memories. Thus, short- and long-term visual memories are lateralised in wood ants. This extends the modalities across which visual memories are lateralised in insects and suggests that such memory lateralisation may have evolved multiple times, possibly linked to the evolution of eusociality in the Hymenoptera.


(a) Animals and preparation
Red wood ant workers (Formica rufa L.) from three different colonies were used for the experiments. These colonies were collected from Ashdown Forest, Sussex, UK (N 51 4.680, E 0 1.800) in June 2017 and June and August 2018. They were maintained indoors for at least 2 months prior to the experiments commencing, during which they were kept at 26°C, under a 12 h light:12 h dark cycle and fed with sucrose (333 g/L). Prior to training, colonies were starved for at least 2 days to increase motivation for feeding. Ants were selected from the colony on the day of training and harnessed to a custom-made holder (see Fernandes et al., 2018).

(b) Training
Ants were trained to associate a visual cue with a sucrose reward as described in Fernandes et al. (2018). We performed six different types of training to investigate the role of the antennae during visual associative learning. Paired Right (PR) and Paired Left (PL) ants experienced right or left antenna touching by the US prior to contact with the mouthparts, respectively. Paired Bilateral (PB) ants were trained with a sucrose reward to both antennae prior to feeding, and data pooled with that from Fernandes et al. (2018). No touch to the antennae (NTA) ants had their antenna attached to the insect pin with low melting temperature wax to prevent antennal contact with the reward (US) during training.
Unpaired controls were carried out as described in Fernandes et al. (2018). Unpaired Right (UPR) and Unpaired Left (UPL) ants experienced right or left antenna touching by the US prior to contact with the mouthparts, respectively. An additional unpaired group, Randomized Unpaired Left (RUPL), was trained identically to the UPL cohort except CS and US were presented in a randomized order. A CS-only (CSO) control group had the CS (plus syringe) presentation for the same number of trials, but without the US to antennae or mouthparts. A US-only (USO) control was given paired training but with the CS removed and replaced with a white circular cue (10 cm diameter) to remove all potential visual cues generated during CS delivery except the syringe tip and sucrose drop. The USO did not differ in memory formation from the CSO (supplementary table 1) and is not further discussed here. For all types of training, half of the ants were trained with their right side facing the experimenter and the other half with the left, to account for potential environmental asymmetries that could be learnt during training. In both cases the visual stimulus faced the ant and was seen by both eyes. 

(c) Testing
We tested all groups in the same set-up with a 10 second CS presentation either 10 minutes, 1 hour or 24 hours after training. Ants tested at 10 minutes were not tested at any other time points, but all ants tested at 1 hour were also tested at 24 hours. After testing, ants were briefly presented with the sugar reward to assess whether they were still motivated to feed. Ants that did not feed after testing were excluded from analysis. All ants were kept harnessed until the last test ended. Ants harnessed overnight for 24 hour testing were kept in a dark and humid environment. 

(d) Scoring and statistical analysis
We scored whether ants performed the maxilla-labium extension response (MaLER) during the 10 second presentation of the conditioned stimulus (CS) for each training trial and test (see Fernandes et al., 2018). All statistical analysis was performed using R (RStudio v1.0.143). To analyse whether the propensity to perform MaLER increased during training and whether it differed among different training groups, we ran a logistic regression model with mixed effects (Bates, 2010), using the ‘lme4’ package. Models were implemented using the glmer command incorporating fixed (trial number, training type) and random (individual) effects (see supplementary materials). We used Barnard’s test (Barnard, 1989) to compare the proportion of responses in each group during each test with the ‘barnard’ package. One-tailed or two-tailed Barnard’s tests were applied depending on the specific test (see supplementary materials).  To determine whether ants responding during testing were also those that learnt, we used a classification of ‘learners’ or ‘non-learners’ established in Fernandes et al. (2018); those individuals that responded on 4 or more of the 10 training trials were ‘learners’. Bonferroni corrections (Bonferroni, 1936) were applied when appropriate (see supplementary materials). 


University of Sussex