Walking Drosophila navigate complex plumes using stochastic decisions biased by the timing of odor encounters

Abstract

How insects navigate complex odor plumes, where the location and timing of odor packets are uncertain, remains unclear. Here we imaged complex odor plumes simultaneously with freely-walking flies, quantifying how behavior is shaped by encounters with individual odor packets. We found that navigation was stochastic and did not rely on the continuous modulation of speed or orientation. Instead, flies turned stochastically with stereotyped saccades, whose direction was biased upwind by the timing of prior odor encounters, while the magnitude and rate of saccades remained constant. Further, flies used the timing of odor encounters to modulate the transition rates between walks and stops. In more regular environments, flies continuously modulate speed and orientation, even though encounters can still occur randomly due to animal motion. We find that in less predictable environments, where encounters are random in both space and time, walking flies navigate with random walks biased by encounter timing.

Adaptive temporal processing of odor stimuli

“…Away from the source and from surfaces, odor stimuli can be modelled as scalar quantities (concentrations) superimposed on a vector field (air speed). Therefore, a possible approach is to neglect odor-specific effects (by using mono-molecular high volatility inert compounds) and to use optical methods for quantifying the spatiotemporal distribution of tracer molecules, such as acetone (Connor et al 2018) or smoke (Demir et al 2020). These approaches allow direct spatiotemporal quantification of the stimulus used for behavioral analysis.…”

“…Surely, there is no odor source localization without the olfactory system, but it remains unclear whether and in which range of conditions wind detection is necessary in addition to being useful. Turning upwind at an odor encounter is not an optimal strategy if the turbulence is too high and wind direction uninformative (Demir et al 2020). Similarly, it is not optimal to try to estimate the mean stimulus intensity when the rate of plume encounter is low (Boie et al 2018;Victor et al 2019).…”

“…Odors rarely exist in isolation and the ability to localize an odor source requires the ability to recognize it across a wide range of stimulus conditions. Recent work in model systems like Drosophila and mice has begun to quantify in detail the behavioral responses to odors in different conditions (Álvarez-Salvado et al 2018;Demir et al 2020; Radvansky and Dombeck 2018; Tadres and Louis2020). Understanding which decision the animal takes in response to a specific stimulus and context is fundamental to infer the underlying neural computations.…”

A neural circuit for wind-guided olfactory navigation

“…Wind-guided olfactory navigation presents an ethological model for the integration of value information and spatial or directional information. Attractive food and mate odors drive robust orientation upwind (3)(4)(5)(6)(7)(8)(9). Attraction and repulsion can be both innate or learned (10,11,25,46).…”

“…A natural example of this occurs in wind-guided olfactory navigation (3,4), in which the scent of food, or a potential mate, gates orientation relative to wind direction. This basic algorithm has been observed in diverse species (5)(6)(7), and allows animals to solve the problem of locating an odor source in turbulence (8,9). In insects, numerous theoretical studies have proposed that the output of the mushroom body, a brain structure involved in olfactory and visual memory, might encode stimulus value (10)(11)(12), whereas the integration of value and direction cues might occur within the central complex (13)(14)(15)(16).…”