Skip to Main content Skip to Navigation

Visual learning in Apis mellifera under virtual reality conditions

Abstract : Equipped with a brain smaller than one cubic millimeter and containing ~950,000 neurons, honeybees display a rich behavioral repertoire, among which appetitive learning and memory play a fundamental role in the context of foraging activities. Besides elemental forms of learning, where bees learn specific association between environmental features, bees also master different forms of non-elemental learning, including categorization, contextual learning and rule abstraction. These characteristics make them an ideal model for the study of visual learning and its underlying neural mechanisms. In order to access the working brain of a bee during visual learning the insect needs to be immobilized. To do so, virtual reality (VR) setups have been developed to allow bees to behave within a virtual world, while remaining stationary within the real world. During my PhD, I developed a flexible and open source 3D VR software to study visual learning, and used it to improve existing conditioning protocols and to investigate the neural mechanism of visual learning. By developing a true 3D environment, we opened the possibility to add frontal background cues, which were also subjected to 3D updating based on the bee movements. We thus studied if and how the presence of such motion cues affected visual discrimination in our VR landscape. Our results showed that the presence of frontal background motion cues impaired the bees' performance. Whenever these cues were suppressed, color discrimination learning became possible. Our results point towards deficits in attentional processes underlying color discrimination whenever motion cues from the background were frontally available in our VR setup. VR allows to present insects with a tightly controlled visual experience during visual learning. We took advantage of this feature to perform ex-vivo analysis of immediate early gene (IEG) expression in specific brain area, comparing learner and non-learner bees. Using both 3D VR and a lore restrictive 2D version of the same task we tackled two questions, first what are the brain region involved in visual learning? And second, is the pattern of activation of the brain dependent on the modality of learning? Learner bees that solved the task in 3D showed an increased activity of the Mushroom Bodies (MB), which is coherent with the role of the MB in sensory integration and learning. Surprisingly we also found a completely different pattern of IEGs expression in the bees that solved the task in 2D conditions. We observed a neural signature that spanned the optic lobes and MB calyces and was characterized by IEG downregulation, consistent with an inhibitory trace. The study of visual learning's neural mechanisms requires invasive approach to access the brain of the insects, which induces stress in the animals and can thus impair behaviors in itself. To potentially mitigate this effect, bumble bees Bombus terrestris could constitute a good alternative to Apis mellifera as bumble bees are more robust. That's why in the last part of this work we explored the performances of bumblebees in a differential learning task in VR and compared them to those of honey bees. We found that, not only bumble bees are able to solve the task as well as honey bees, but they also engage more with the virtual environment, leading to a lower ratio of discarded individuals. We also found no correlation between the size of bumble bees and their learning performances. This is surprising as larger bumble bees, that assume the role of foragers in the colony, have been shown to be better at learning visual tasks in the literature.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, July 5, 2022 - 5:50:13 PM
Last modification on : Wednesday, July 6, 2022 - 11:42:10 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03714699, version 1


Gregory Lafon. Visual learning in Apis mellifera under virtual reality conditions. Neuroscience. Université Paul Sabatier - Toulouse III, 2022. English. ⟨NNT : 2022TOU30016⟩. ⟨tel-03714699⟩



Record views


Files downloads