Team : BeeAntCE
Supervisors : Antoine WYSTRACH & Jacques GAUTRAIS
Committee members :
- M. Andrew PHILIPPIDES, University of Sussex, Rapporteur
- Mme Emily BAIRD, Stockholm University, Examinatrice
- M. Antoine WYSTRACH, CNRS, Co-directeur de thèse
- M. Franck RUFFIER, CNRS – Institut des Sciences du Mouvement Marseille, Examinateur
- Mme AUDREY DUSSUTOUR, CNRS, Examinatrice
- M. Simon BENHAMOU, CNRS / CEFE, Rapporteur
Abstract :
Navigation in natural environments is an essential ability for animals, but it can be highly challenging to achieve due to the complex, ever-changing, nature of the environment they navigate through. Solitary foraging hymenopterans, such as ants, have evolved remarkable abilities to navigate in these complex environments despite a nervous system much simpler than those of vertebrates. What’s more, ants’ displacements can be easily tracked, thus providing a powerful system to investigate the mechanisms underlying navigation in the wild.
Thanks to the development and application of neurobiological tools in insects, we now have an increasingly detailed description of the circuits in these « mini-brains ». These networks, composed of a large number of interacting units (neurons), are the site of very dynamic internal activity that allows for an effective coupling of the organism with its environment. The field of insect navigation has integrated this neurobiological knowledge with a long-standing tradition of behavioural experiments, as well as in-silico modelling approaches, to gain a deeper understanding of the mechanisms and emergent properties in these systems. This multi-level approach has provided valuable insights into how complex behaviour emerges.
This thesis presents an integrative approach combining behavioural experiments and computational modelling with the aim to gain further understanding of the mechanisms of ant navigation.
First, I investigate the dynamics of visual learning when ants navigate in natural conditions. There are already good insights into how the insect brain can memorize and recognize views, and how these views might be used for navigation. However, practically nothing is known about how learning is orchestrated in the first place. Learning a route cannot be governed only by rewards and punishments but may happen ‘continuously’; a vague concept so-called ‘latent learning’ in the vertebrate literature. By combining field experiments and modelling, I show that ants learn the route they travel in a continuous fashion. I also explain how such a continuous mechanism of learning and recalling can be implemented in the light of the known insect brain circuitry. Our work shows that such a continuous spatial learning is supported by egocentric route memories rather than a map-like reconstruction of space.
Secondly, I shed light on how higher-level visual recognition and lower-level motor control interact. Most -if not all-studies in animal navigation have focused on higher level strategies such as the use of path integration, the recognition of learnt visual cues and the reconstruction of so-called cognitive maps. However, how these higher-level strategies are supported by lower level (often more ancestral) motor behaviours remains largely unexplored. We used a multidirectional treadmill set up -a trackball- to record with high precision the motor behaviour of ants directly in their natural environment. Results explain how higher-level navigation strategies are supported by lower-level motor control and demonstrate the existence – and importance – of an intrinsic oscillator at the core of navigational behaviours.
Finally, I investigate the specific cues encoded by ants in visual scenes. Insects’ visual system performs various feature extractions. I used virtual reality (VR) to explore which visual cues are encoded by ants to consider the presence of a natural panorama and trigger exploratory behaviours. Through this approach, I shed light on the perceptual encoding of naturalistic environments and provide a promising avenue for further investigation using virtual reality setups.
Taken together, this thesis allowed me to understand that behaviour, rather than being a set of discrete actions, is a continuous process where intelligence can be seen as an emergent property. I employed various approaches, some successful, some failing, but all improved my vision of how I should ask and answer a scientific question.