1. Neuroscience
  2. Physics of Living Systems
Download icon

Olfactory Navigation: Tempo is the key

  1. Antonio Celani  Is a corresponding author
  1. Quantitative Life Sciences, International Centre for Theoretical Physics, Italy
  • Cited 0
  • Views 674
  • Annotations
Cite this article as: eLife 2020;9:e63385 doi: 10.7554/eLife.63385


Walking flies find the source of attractive odors by changing how frequently they stop and turn in response to the smell.

Main text

Tracking odors is a matter of life or death for most organisms, and insects are no exception (Baker et al., 2018). Male moths can track the pheromones of their potential mating partners from hundreds of meters away, and even fruit flies successfully engage in long-range searches for food using their sense of smell. However, this task is far from easy. This is because the molecules that carry the odorous message are transported by turbulent wind, where they get mixed with other molecules and form complex structures (Murlis et al., 1992; Crimaldi and Koseff, 2001). As a result, the information about the origin of a scent is seemingly lost, hidden in the intricacies of the intermittent odor signal. Insects appear to be able to overcome this problem, but it is unclear how they extract information about the origin of the odor and translate it into behaviors that allow them to find the source.

A major hurdle in the way of understanding how animals track smells is the need to visualize odors and behaviors simultaneously. While tracking insect behaviors in the wild remains a daunting task, wind-tunnel experiments provide a way to collect high-throughput data in controlled situations (Álvarez-Salvado et al., 2018). However, the artificial environment still poses several challenges: for example, it is unclear whether naturally occurring stimuli can be reproduced, or if it is possible to visualize odor concentrations with sufficient resolution.

Now, in eLife, Thierry Emonet and co-workers from Yale University – including Mahmut Demir and Nirag Kadakia as first authors – report how the fruit fly Drosophila melanogaster behaves in response to an attractive smell while walking (Demir et al., 2020). Quite serendipitously, they discovered that starved flies are attracted to smoke, which can be easily visualized. By manipulating the airflow in a wind tunnel with lateral jets, they generated a stream of smoke with properties similar to the odor signals that flies encounter in the wild (Celani et al., 2014). In agreement with theoretical expectations, they found that, within the smoke, flies have brief, frequent and unpredictable encounters with the odor. But how do these encounters modulate flies’ behavior?

Demir et al. observed that the rich variety of movements exhibited by walking Drosophila could be summarized into just a few behavioral states relevant to olfactory navigation, echoing previous findings (Tao et al., 2019). They found the search process was inherently stochastic: periods of walking in a straight line would randomly be interrupted by rapid turning events or by stopping for longer extents of time.

By comparing the trajectories recorded with or without smoke, but always in presence of turbulent airflow, Demir et al. were able to identify which features of the flies’ movements were affected by encountering the smell. They found that the walking pace, the frequency and speed at which the flies made a turn, and the sharpness of the turns were not affected by the smoke. Conversely, when the smoke was present walks were on average longer and stops shorter. Most importantly, when the flies were turning, they were more likely to reorient upwind against the direction of the wind if they had already encountered the odor. But what are the specific characteristics of the odor that the fly perceives and responds to?

Demir et al. found that neither the concentration of the odor nor the amount of times flies were exposed to it played a significant role in the search process. Instead, the tempo of the flies’ encounters with the odor appeared to be the key determinant of the decision-making process. In a stopped fly, a single encounter was sufficient to initiate a walk, and several encounters close together shortened the duration of the stops. Walking times increased after an encounter, but further exposure to the odor shortly after did not lead to a cumulative effect. This modulation of walks and stops produces a bias that results in the fly visiting regions of space where it is more likely to encounter the smell. Further experiments showed that above a certain frequency of encounters flies were more likely to re-orientate themselves upwind. In fact, the combined effect of the odor on the frequency of walks and stops, as well as the direction of the turns, proved to be fundamental for the flies to find the origin of a smell (Figure 1).

How a walking fly navigates towards the source of an odor.

Flies that have been starved are attracted to smoke. In this schematic diagram of a wind tunnel experiment, the fly starts on the right, the source of the smoke is on the left, and the wind blows from left to right (black arrow). When far from the source (right box), the fly is less likely to encounter the odor, leading to frequent stops (red dots) and a similar number of upwind and downwind (green and blue dots) turns. Once the fly gets closer to the source (left box), it is more likely to encounter the odor, and the number of stops decrease, while the frequency of upwind turns increases.

There are tantalizing similarities between the olfactory-search strategies of walking flies and other insects, but also significant differences (Mafra-Neto and Cardé, 1994; Budick and Dickinson, 2006). In any event, irrespective of their sizes and behaviors, different species must overcome the same challenges posed by the transport of odor molecules by the turbulent airflow, pointing to the existence of general underlying principles. Perhaps, in the future, more studies like this one will pave the way towards a more fundamental understanding of long-range olfactory navigation.


  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9

Article and author information

Author details

  1. Antonio Celani

    Antonio Celani is in the Department of Quantitative Life Sciences, International Centre for Theoretical Physics, Trieste, Italy

    For correspondence
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0339-1073

Publication history

  1. Version of Record published: November 3, 2020 (version 1)


© 2020, Celani

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.


  • 674
    Page views
  • 58
  • 0

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Neuroscience
    Andreas Strube et al.
    Research Advance Updated

    In the context of a generative model, such as predictive coding, pain and heat perception can be construed as the integration of expectation and input with their difference denoted as a prediction error. In a previous neuroimaging study (Geuter et al., 2017) we observed an important role of the insula in such a model but could not establish its temporal aspects. Here, we employed electroencephalography to investigate neural representations of predictions and prediction errors in heat and pain processing. Our data show that alpha-to-beta activity was associated with stimulus intensity expectation, followed by a negative modulation of gamma band activity by absolute prediction errors. This is in contrast to prediction errors in visual and auditory perception, which are associated with increased gamma band activity, but is in agreement with observations in working memory and word matching, which show gamma band activity for correct, rather than violated, predictions.

    1. Neuroscience
    Javier Alegre-Cortés et al.
    Research Article Updated

    Behavioral studies differentiate the rodent dorsal striatum (DS) into lateral and medial regions; however, anatomical evidence suggests that it is a unified structure. To understand striatal dynamics and basal ganglia functions, it is essential to clarify the circuitry that supports this behavioral-based segregation. Here, we show that the mouse DS is made of two non-overlapping functional circuits divided by a boundary. Combining in vivo optopatch-clamp and extracellular recordings of spontaneous and evoked sensory activity, we demonstrate different coupling of lateral and medial striatum to the cortex together with an independent integration of the spontaneous activity, due to particular corticostriatal connectivity and local attributes of each region. Additionally, we show differences in slow and fast oscillations and in the electrophysiological properties between striatonigral and striatopallidal neurons. In summary, these results demonstrate that the rodent DS is segregated in two neuronal circuits, in homology with the caudate and putamen nuclei of primates.