Visual learning and processing in the honeybee, Apis mellifera

Abstract

The honeybee, Apis mellifera, is well-known for her crucial role in crops pollination and for producing an highly-appreciated food, the honey. Indeed, the first historical record of bee-keeping dates from the Mesolithic at least 9000 years ago. The honeybee possesses a miniature brain (approximately one million of neurons) but exhibits incredibly sophisticated behavior. For instance, a forager honeybee can learn abstract concepts such as sameness/difference or above/below, to recognize specific human faces and even count. The neural basis of those behaviors have yet to be elucidated due to the difficulty in linking free-flying behaviors with tethered preparations typically used in 'in vivo' neurophysiological recordings. In this dissertation, my goal is to develop a protocol to induce visual learning in a tethered walking honeybee and record from an optic lobe during visually guided behaviors. In the first chapter, I present an introduction on honeybee learning and memory, how learning protocols were developed, and what we know about the neural bases of these behaviors. I briefly summarize honeybee vision and what is known about internal state and behavioral state modulation of visual processing in insects. In the second chapter, I present our study on visual learning in tethered walking honeybees. We developed a Virtual Reality Environment (VR) composed of a walking treadmill connected to a computer and surrounded by a screen. Visual stimuli were projected on the screen based on the honeybee motion (closed-loop) or based on predefined inputs (open-loop). We showed that tethered honeybees in our VR could learn to discriminate between visual stimuli, although not all combinations of shapes and colors were learned equally. We hypothesized that optic flow, the motion of the visual panorama on the retina, was critical for learning. In the third chapter, I present our study on sensory feedback and internal state modulation in an optic lobe of the honeybee brain. In this study, we combined behavioral and neural recordings to explore how internal state and sensorimotor feedback impacted neural activity in the medulla, the second optic lobe of the honeybee. We presented the honeybee with visual stimuli in closed-loop, where the animal had control over the motion of the visual scene (i.e., self-generated optic flow) and subsequently replayed the visual scene motion in open-loop (i.e., externally generated optic flow). We found that neural activity in the medulla is modulated by locomotory state (i.e., walking versus non-active locomotion bouts) almost exclusively in the presence of closed-loop behavioral control. Overall, around a third of the neural population recorded was influenced by behavioral control. Honeybees exhibited a surprisingly high ability to adapt to multiple levels of motor gain (i.e., the relationship between her motion and the motion of the visual scene) and this capacity is likely to rely on the release of octopamine, an invertebrate neuromodulator, in the medulla. In summary, our work provides support to the growing idea that internal and behavioral state are essential for studying how the brain produces behavior. The link between brain and behavior will require an interdisciplinary approach and the development of novel methodologies that are spread across disciplines (neurophysiology, behavioral, neurogenetic, psychology, engineering, mathematics, etc.) and species.