Neural correlates of contrast normalization in the Drosophila visual system

The fruit fly Drosophila melanogaster has long become a paramount model organism for research in life sciences. As a result of the fly’s high temporal resolution and its reliable optomotor response - a reflex that helps it compensate for movements of the environment, - Drosophila lends itself exceptionally well to the study of vision and, in particular, the mechanism of motion detection. In the wild, Drosophila is active throughout the day, with especially high levels of activity at dawn and dusk, the periods of time when the visuals of the environment are changing rapidly. The fruit flies can also be found in a variety of habitats, from the expanse of an open field to the inside of a cluttered kitchen. Altogether, Drosophila encounters a variety of visual statistics it must employ to robustly respond to the outside world and succeed in finding food, escaping predators, and carrying out courtship behavior. In my thesis, I focused on the effects of visual contrast, i.e., differences in brightness in the environment, on the fly’s motion vision. I studied the impact of the surround contrast on the filtering properties of the visual interneurons within the motion detection circuit, including the first direction-selective T4 and T5 cells and their main inputs, and how the fly compensates for the changes in contrast to faithfully match the direction and speed of its movement to the external motion under various contrast conditions. Firstly, in Manuscript 1, we established the existence of contrast normalization in the early visual system of Drosophila and demonstrated its suppressive effect on the response amplitude at higher contrasts. We determined where contrast normalization first arises in the optic lobe and identified the main inputs into the T4 and T5 cells that exhibit contrast normalizing properties. We comprehensively characterized the normalization process: namely, it is fast, not dependent on the direction of motion, its effect comes from outside the receptive field of a cell and increases in strength with the size of the visual surround. Additionally, we demonstrated that the normalization relies on neuronal feedback and showed that adding a contrast normalization stage to the existing models of motion detection improves their robustness, matching their performance to the results obtained in behavioral experiments. In Manuscript 2, we further investigated the effects of contrast normalization on the main inputs to T4 and T5 cells, now focusing on its effect on the filtering properties of the cells. We demonstrated that spatially or temporally dynamic surrounds elicit contrast normalization, while static ones do not. We further showed that, in addition to the suppressive effect on the amplitude, contrast normalization speeds up the kinetics of the response and confirmed that this effect is not due to signal saturation and involves a change in the filtering properties of the cell. In summary, we elucidated the role of contrast normalization in the motion detection circuit in the early visual system of Drosophila, comprehensively described the characteristics of the normalization process, and outlined its effects on the filtering properties of the cells. We also emphasized the potential role of shunting inhibition and narrowed down the search for the main candidates in the contrast normalization mechanism, paving the way for future studies to further delve into the contrast normalization circuit and implementation mechanism