Drosophila learn to prefer immobile spherical objects through repeated physical interaction

  1. Kenichi Iwasaki
  2. Sena Kawano
  3. Ashnu Cassod
  4. Charles Neuhauser
  5. Aleksandr Rayshubskiy

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Animals interact with unfamiliar objects to learn about their properties and guide future behavior, but the underlying neurobiological mechanism is not well understood. Here, we developed a behavioral paradigm in which freely walking Drosophila melanogaster are repeatedly guided to spherical objects using a visual cue. Flies exhibited diverse and structured object interaction motifs, including “ball pulling”, and “ball walking”, that evolved over time. Notably, flies developed a strong preference for immobile over mobile spherical objects, despite their near identical appearance, suggesting they learn about the object’s stability through physical interaction. This preference was impaired by silencing specific hΔ neurons in the fan-shaped body, previously implicated in spatial navigation but not known to contribute to object interactions. Our results show that hΔ neurons also modulate object interaction motifs and fidelity of following visual guidance cues, pointing to a role in balancing goal-directed and exploratory behaviors. These findings establish Drosophila as a model for investigating how internal representations and multimodal feedback contribute to adaptive object interaction.

Article activity feed

  1. PREreview

    This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/17172722.

    Summary:

    In this study, the authors investigate how Drosophila fruit flies interact with spherical objects using a new behavioral paradigm. They provide compelling evidence supporting their main claims and include appropriate controls. The study demonstrates that using a visual stimulus to guide flies to the vicinity of a spherical object (ball) increases both the probability and duration of interaction with the object. The authors further provide solid evidence that flies employ a set of different behavioral motifs to interact with the object, and that the mixture of these motifs changes over time. They also present compelling evidence that flies exhibit different interaction levels with a movable versus immovable ball. Finally, they report on silencing experiments screening for the role of different hΔ central complex neurons in the behavioral motifs exhibited during interaction with objects. This very important study lays the groundwork for a mechanistic understanding of the neural substrate controlling object interaction behaviors in Drosophila.

    Strengths: The manuscript is very well written, with clear structure and language. The claims are substantiated and supported by appropriate controls, minimizing potential confounds.

     Weaknesses: Some minor adjustments in figure labeling and captions, along with clarifications in methodology and the main text, would help the readers better appreciate the reported results. Specific suggestions are provided below.

     Major points:

    • The authors quantified the flies' preference for the mobile and the immobile ball by counting the number of contacts with the ball per visit (lines 186–191). A contact event was registered when the fly-to-ball distance reached zero. However, it is unclear whether the tracking software was tracking the fly's centroid or also the position of the legs. If only the centroid was tracked (as is typical for most tracking software), the flies could have been touching the ball with their legs and getting feedback about its mechanical properties before a contact event was registered. The interpretation of the data shown in Figures 4D-G depends on this detail, so clarifying exactly how a zero distance was defined in the methods section would help the readers better appreciate the figures.

    • The finding that flies interacted little with the ball during the first few hours of the experiments is intriguing. It would be valuable to include a discussion of the possible reasons. Does this initial period reflect a familiarization and learning phase, or a motivational state that changes after some time in the arena? An earlier version of the preprint included an additional plot (Figure S3) showing that the animals' locomotion speed followed a similar temporal trajectory, increasing after a few hours. This pattern would suggest a change in motivation state, with the flies ignoring the visual guidance stimulus for the initial period. A full analysis of this would be beyond the scope of this study, but a discussion of authors' observations about this salient feature would be valuable.

    • In lines 107–108, the text states: "Flies were not required to engage with the object and could leave without penalty." However, one possible interpretation of the data is that the flies avoided the rotating pinwheel stimulus. The only location where the pinwheel was turned off was at the ball and a 5-mm zone around it. Upon leaving this area, the rotating pinwheel was projected again. If the flies perceive the rotating pinwheel as aversive, then it would constitute a penalty for leaving the ball vicinity. In this case, it would be preferable to remove the statement that the flies could leave "without penalty". Alternatively, the authors could design a control experiment to test whether the flies avoided or followed the rotating pinwheel stimulus. For example, one or more rotating pinwheels could be projected at random locations in an empty arena (remaining in place instead of tracking the fly), and the occupancy of the arena surface by the fly over time could be analyzed. If the regions with projected rotating pinwheels are less occupied than other areas, it would suggest aversion to the stimulus. If the rotating pinwheels are not less occupied the statement that the flies were not penalized for leaving the ball vicinity could be justified. Another possible control would be to delay turning on the rotating pinwheel stimulus until the fly moves further away from the ball than the 5-mm zone used in the current experiments. If the visual stimulus is not aversive, the number of contacts per visit should remain similar to the currently reported values.

    Minor points:

    • In a few figures, the labeling could be made more descriptive to help readers interpret the presented data. Specifically, consider replacing "Interaction score" with "Ball displacements per visit" in Figure 1F, and with "Contacts per visit" in Figures 4F–G, 5E–G.

    • In Figures 3B and 3C, there appears to be a discrepancy between the y-axis label, "Event #," and the figure caption. Please clarify whether these panels report counts or frequencies.

    • Although not necessary, it would be interesting for the authors to comment on potential reasons for the sharp dip in the score early in the experiment (around the 8th–10th visit) shown in Figure 5G.

    • In Figure S1A, please define in the figure caption what the "active time" percentage displayed on the x-axis represents.

    • In Figure S5H, the plot does not show a pronounced dip like the one in Figure 5H. The caption mentions a dip, which could be revised for consistency with the plot.

    • In line 140: Looking at Figure 2B, the wording could perhaps be clarified as "with varying degrees of fly displacement".

    Finally, a more detailed discussion of the study's limitations, especially regarding the silencing of the hΔ neurons, would further strengthen the already well-presented manuscript.

    Competing interests

    The authors declare that they have no competing interests.

  2. PREreview

    This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/17149340.

    Review for "Drosophila learn to prefer immobile spherical objects through repeated physical interaction"

    He Yang and Yoshinori Aso

    Preprint doi: https://doi.org/10.1101/2025.06.09.658381

    In this preprint, the authors used Drosophila as a model organism to study how animals physically interact with objects, taking advantage of its extensive genetic toolkit and accessibility. With a novel technique, they visually guided freely walking fruit flies to approach small balls and analyzed the subsequent behaviors for over 1000 times per fly. Flies displayed complex and structured patterns of interaction, which could be categorized into five distinct behavioral motifs. The ratio of these behavioral motifs shifted over time. Flies also developed a preference for immobile over mobile objects. They then examined the behaviors of flies expressing Kir2.1 with driver lines for different hD neuron types and observed driver-specific loss or enhancement of selective behavioral motifs and the preference to immobile balls. The evidence presented convincingly supports the conclusion that flies are capable of learning about the physical properties of objects and adjusting future decisions based on prior experience, thereby addressing the gap in the field to understand how flies interact with their environment.

    Strengths:

    The authors developed an innovative way to analyze fly-object interactions by employing the "pinwheel" cue to guide flies toward inanimate objects. This new approach enabled the authors to observe hundreds of object interactions for individual flies within a reasonable timeframe and how they developed over time.  The authors collected movies of nearly 30,000 interaction events and defined five behavioral motifs. This highly efficient way of data collection provided a novel framework for quantifying and comparing complex behavioral patterns across time and examining the effect of cell type-specific perturbation.

    Weakness:

    The main conclusion of this study is that flies develop object interaction motifs and learn to prefer immobile balls over time. The data presented in Figure 3 and Figure 5 convincingly show that flies change these behaviors over time. However, these changes could partly arise from changes in the physiological status of flies rather than learning physical properties of objects. As detailed below, providing more details on methods, experimental design and data interpretation would enhance the manuscript's clarity, reproducibility, and overall impact.

    Major comments:

    1.     In this behavioral assay, flies are forced to keep walking for ~18 hours until death in a chamber without food or water. The observed changes in patterns of interaction events (Figure 3) over time may therefore correlate not only with learning but also with the fly's physiological state - transitioning from an eager search for resources to eventual weakness. For instance, even guided flies take ~6 hours to start showing enhanced ball displacement (Figure 1C and D). This could be a period necessary for learning to interact with balls. Alternatively, flies may need to keep walking ~6 hours in response to pinwheel stimulus to change their status. One possible control experiment to rule out the latter possibility is to guide them to walk to a fixed point with pinwheel for 6 hours without a ball and then introduce a ball. If the latter scenario is true, flies without prior experience with a ball may show enhancement of ball displacement or preference to immobile balls immediately after the introduction of the ball. Another possible control experiment is to capture experienced flies and retesting them after feeding.

    2.     In the methods, the authors described that a "contact" was defined as a period during which the fly maintained any physical contact with the ball, with a fly–ball distance of zero. Clarify how this fly-ball distance was defined in methods. Is it a distance between the surface of fly's body and the surface of the ball? If flies could extend legs and touch a ball and examine the mobility of the ball from non-zero distance, Figure 5 data could be explained without hypothesizing that "flies may associate spatial locations with object properties, and that these associations could guide future interactions." Instead, flies may have developed a quick way to access the mobility of balls with light touch from non-zero distance. To rule out this possibility, the authors may show analysis of leg-ball contact in a greater detail. One idea for an additional control experiment is to change the location of balls during the experiments. The authors could set up two mobile balls and two immobile balls in the arena. Initially flies are guided to one set of mobile and immobile balls until they develop stable preference to the immobile ball, and then they can be guided to another set of balls. If they need to form a new association between the location and the properties of the ball, it should take many trials to re-establish preference to the immobile ball.

    3.     For Figure 6 and Figure 7, it is generally recommended to attribute behavioral phenotypes to a particular cell type by replicating the phenotype with multiple driver lines. Although these driver lines are highly specific to the hΔ neurons in adult brains, the observed phenotypes could be due to off-targeted expression during development or driver specific genetic backgrounds. It would also be a good practice to validate the expression of Kir-eGFP in the targeted cell types by immunolabeling and confocal microscope, because the expression patterns of 10XUAS-IVS-eGFPKir2.1 could be different from that of 20xUAS-CsChrimson in attP18 used by Wolff et al., 2025 (doi: 10.7554/eLife.104764).

    Minor comments:

    1.     The videos referenced in the text are currently missing from the preprint. Including them would greatly assist readers in appreciating the nuanced behaviors described. Overlaying segmentations of fly body parts and balls on movies would be also helpful to understand the definition of "fly-ball contact".

    2.     Various interaction scores (cumulative / mean / global) are used as important parameters to quantify fly-ball interactions. For improved readability, consider including a brief explanation of how these scores are calculated in the main text or figure legends (e.g., in Figure 1F-G and Figure 4G) when they are first introduced, even if the full details are in Methods.

    3.     When introducing the "pinwheel", the authors may include a very brief background introduction to explain that this is a projected visual cue, with which clockwise rotation guides the fly to turn right and counterclockwise rotation guiding left. This way readers can understand the basics of this key experimental component without checking the previous paper.

    4.     The authors summarized a good review of how animals generally explore other objects in the environment in the Introduction. In either the Introduction or Discussion, the authors may consider discussing whether the preference for immobile objects is observed in any other animal to determine if the behavioral principles observed in flies are conserved across animals, thereby broadening the impact of this work within the field of animal behavior.

    5.        Consider an alternative hypothesis about the contents of learning in Figure 5. The authors favored a hypothesis that "flies may associate spatial locations with object properties, and that these associations could guide future interactions". Instead, flies may associate spatial locations with jumping, a potentially aversive action. Amin et al. (doi: https://doi.org/10.1101/2025.07.07.663268) reported that backward walking via moonwalker neurons can activate punishment-representing dopaminergic neurons and thereby induce aversive olfactory memory. Similarly, interaction with mobile balls may result in jump-induced activation of punishment pathway to promote avoidance of the location around the mobile balls. It would improve reader understanding if the authors could comment on the alternative hypotheses in the discussion.

    Competing interests

    The authors declare that they have no competing interests.

  3. Version published to 10.1101/2025.06.09.658381 on bioRxiv