TrackFly: Virtual reality for a behavioral system analysis in free-flying fruit flies

https://doi.org/10.1016/j.jneumeth.2008.02.016Get rights and content

Abstract

Modern neuroscience and the interest in biomimetic control design demand increasingly sophisticated experimental techniques that can be applied in freely moving animals under realistic behavioral conditions. To explore sensorimotor flight control mechanisms in free-flying fruit flies (Drosophila melanogaster), we equipped a wind tunnel with a Virtual Reality (VR) display system based on standard digital hardware and a 3D path tracking system. We demonstrate the experimental power of this approach by example of a ‘one-parameter open loop’ testing paradigm. It provided (1) a straightforward measure of transient responses in presence of open loop visual stimulation; (2) high data throughput and standardized measurement conditions from process automation; and (3) simplified data analysis due to well-defined testing conditions.

Being based on standard hardware and software techniques, our methods provide an affordable, easy to replicate and general solution for a broad range of behavioral applications in freely moving animals. Particular relevance for advanced behavioral research tools originates from the need to perform detailed behavioral analyses in genetically modified organisms and animal models for disease research.

Introduction

A detailed understanding of how animals control their movements in a complex natural environment is likewise relevant to neuroscientists exploring neuromotor control mechanisms and engineers attempting to implement biological control principles in microrobots, such as micro air vehicles (MAVs). The reflexive flight control mechanisms of insects are experimentally highly amenable and therefore serve as powerful model systems to explore neuromotor control mechanisms (e.g. Frye and Dickinson, 2001).

Here we describe methods developed for a detailed behavioral system analysis in freely flying fruit flies (Drosophila melanogaster Meigen). Based on Virtual Reality (VR) display techniques implemented in standard digital hardware, we designed an automated ‘one-parameter open loop’ testing paradigm that allowed us to quantify the open loop transfer properties of the fly's visual ground speed response. The ability to perform meaningful behavioral analyses with a high throughput meets the demands for an interdisciplinary research effort on neuromotor control strategies based on advanced genetic tools and concepts derived from control systems engineering.

The mechanisms underlying visuomotor flight control, like most other behaviors, are highly complex. This complexity can be approached with a reductionist approach, in which the animal is considered as a system of interconnected control loops, which can be analyzed using standard control system analysis techniques. In this approach, various sensory modalities are considered as inputs to the system, which after a sensorimotor transduction process lead to the motor output. As a result of the interaction with the physical environment, this leads to appropriate behavior and consequently generates sensory feedback, closing the feedback control loop (Fig. 1).

In the past, visuomotor flight control mechanisms of flies and other insects have been explored extensively under restricted experimental conditions. As a classic approach, input–output relationships of identified neuromotor control loops have been measured from rigidly tethered insects, in which sensory stimuli can be delivered precisely and the resulting motor output measured with comparatively simple tools. As a result, it has been possible to characterize sensorimotor systems from their transfer properties, allowing structure–function relationships to be inferred. In the biological literature the stimulus condition is referred to as open loop to reflect the experimental de-coupling of the sensory stimulus from the motor behavior.

Countless combinations of sensory stimulation and recording of motor actions have been employed alone in flies, including for example: leg extension elicited from optic flow (Borst and Bahde, 1986), head movements from visual or mechanical stimulation (Hengstenberg et al., 1986), walking behavior in presence of visual cues (Götz and Wenking, 1973), flight behavior in presence of varying visual (Götz, 1964, Götz, 1965, Götz, 1968, Blondeau, 1981; reviews: Buchner, 1984, Collett et al., 1993), olfactory (Frye and Dickinson, 2004) or wind (Gewecke, 1967) stimuli.

Examples of open loop experiments in other species include acoustic target tracking in crickets (e.g. Hoy and Paul, 1973, Hedwig and Poulet, 2004; for recent methods see Lott et al., 2007), optomotor (Baader, 1991) and object avoidance responses (Robertson and Johnson, 1993) in tethered flying locusts.

While tethered paradigms offer a broad range of powerful experimental possibilities, their significance for realistic behavior is limited due to the significant inconsistencies between natural and experimental conditions (discussed, e.g. in Buchner, 1984, Gray et al., 2002, Taylor and Zbikowski, 2005). Flies, for example, rely on mechanosensory feedback from specialized balance organs, the halteres (Pringle, 1984, Nalbach and Hengstenberg, 1994) for flight stabilization and bilateral haltere ablation indeed renders the fly incapable of stable flight (Derham, 1714, cited by Dickinson, 2005). Tethering disrupts the reafferent mechanical input to the halteres and tethered flies accordingly show strong behavioral artifacts, including distorted wing stroke kinematics (Fry et al., 2005) and a several-fold prolonged time course of turning maneuvers in visual flight simulators (Heisenberg and Wolf, 1979). Though realistic flight dynamics can be experimentally implemented at least partially with loose tethers (e.g. Baker, 1979, Heisenberg and Wolf, 1979, Mayer et al., 1988, Bender and Dickinson, 2006), a meaningful analysis of neuromotor flight control mechanisms ultimately requires behavioral data to be measured in flies flying freely under realistic flight conditions (e.g. Drosophila: Fry et al., 2003; Fannia: Land and Collett, 1974).

Numerous behavioral studies performed in freely flying insects have addressed sensorimotor control mechanisms (reviews: Collett et al., 1993, Srinivasan and Zhang, 2004), and visual flight speed responses of various insect species in particular (e.g. mosquito: Kennedy, 1939; moth: Willis and Arbas, 1991; bee: Srinivasan et al., 1996; fly: David, 1979). Previous studies of visual speed control in free-flying insects showed that insects maintain a ‘preferred’ flight speed relative to the visual surround, irrespective of the pattern's spatial frequency (David, 1982, Srinivasan et al., 1996).

To explore the control properties of the flight speed response, it was necessary to stimulate flies with arbitrary visual patterns over a broad parameter range and measure the resulting corrective responses. Furthermore, the measurement of transient responses is essential to a formulation of time-continuous models, as presented elsewhere (Rohrseitz and Fry, in preparation).

To this end, we implemented a free flight, ‘one-parameter open loop’ paradigm using a wind tunnel equipped with Virtual Reality display technology. Our methods allowed fully automated mass testing of individual flies over a broad range of visual stimuli, combining the advantages of open loop visual stimulation with the advantages of performing behavioral experiments under realistic free flight conditions (see Schuster et al., 2002, for a comparable approach in walking fruit flies). Furthermore, our methods allowed a straightforward implementation of various other behavioral paradigms, including the presentation of virtual objects, Gabor patches and naturalistic images.

Section snippets

Wind tunnel

The behavioral tests were performed in a commercial open circuit, closed throat wind tunnel (Engineering Laboratory Design, Inc., Lake City, MN, USA), equipped with a real time 3D tracking system and virtual reality display technology (Fig. 2 and see below). The wind tunnel provided a laminar airflow in a working section made of clear acrylic, 1.55 m in length and 0.305 m in width and height. Standard tests were performed using a wind speed of 0.37 m s−1. An attractant odor (‘Kressi’ herb vinegar,

Overview

As an application example, we describe the implementation of an automated, ‘one-parameter open loop’ testing paradigm, which was used to explore the dynamics of visual flight speed control in the fruit fly. An automated largely unsupervised testing scheme was required to provide standardized measurement conditions and explore a large parameter space (Section 3.2). We measured transient visual responses under open loop conditions by controlling the phase of sine grating stimuli in real time

Discussion

We described methods and concepts that allowed freely flying fruit flies to be stimulated with arbitrary visual stimuli in an automated experimental paradigm. Specifically, the implementation of an automated ‘one-parameter open loop’ paradigm allowed testing a freely flying fly under realistic free flight conditions, while controlling a single parameter, the horizontal optic flow, in open loop for a detailed characterization of transient speed responses. The described procedures were also

Acknowledgements

The authors thank Jérôme Frei, Marie-Christine Fluet and Martin Bichsel for their contributions to this work. Financial support was provided by the Human Frontiers Science Program, the University of Zürich (to S.N.F.), the Swiss Federal Institute of Technology (ETH-0-20338-06 to N.R.), the National Science Foundation (FIBR 0623527) and the Air Force Office of Scientific Research (FA9550-06-1-0079 to M.H.D).

References (53)

  • S. Schuster et al.

    Virtual-reality techniques resolve the visual cues used by fruit flies to evaluate object distances

    Curr Biol

    (2002)
  • P. Stockinger et al.

    Neural circuitry that governs Drosophila male courtship behavior

    Cell

    (2005)
  • B. Webb et al.

    Sensorimotor control of navigation in arthropod and artificial systems

    Arthropod Struct Dev

    (2004)
  • P.S. Baker

    Flying locust visual responses in a radial wind tunnel

    J Comp Physiol A Neuroethol Sens Neural Behav Physiol

    (1979)
  • J.A. Bender et al.

    Visual stimulation of saccades in magnetically tethered Drosophila

    J Exp Biol

    (2006)
  • J. Blondeau

    Aerodynamic capabilities of flies, as revealed by a new technique

    J Exp Biol

    (1981)
  • A. Borst et al.

    What kind of movement detector is triggering the landing response of the housefly?

    Biol Cybern

    (1986)
  • E. Buchner

    Behavioral analysis of spatial vision in insects

  • T. Collett et al.

    Visual stabilization in arthropods

    Rev Oculomot Res

    (1993)
  • C.T. David

    Compensation for height in the control of groundspeed by Drosophila in a new, ‘barber's pole’ wind tunnel

    J Comp Physiol A Neuroethol Sens Neural Behav Physiol

    (1982)
  • C.T. David

    Optomotor control of speed and height by free-flying Drosophila

    J Exp Biol

    (1979)
  • M.H. Dickinson

    The initiation and control of rapid flight maneuvers in fruit flies

    Integr Comp Biol

    (2005)
  • J.B. Duffy

    GAL4 system in Drosophila: a fly geneticist's Swiss army knife

    Genesis

    (2002)
  • S.N. Fry et al.

    The aerodynamics of free-flight maneuvers in Drosophila

    Science

    (2003)
  • S.N. Fry et al.

    The aerodynamics of hovering flight in Drosophila

    J Exp Biol

    (2005)
  • M.A. Frye et al.

    Motor output reflects the linear superposition of visual and olfactory inputs in Drosophila

    J Exp Biol

    (2004)
  • Cited by (89)

    • Recent approaches to study the neural bases of complex insect behavior

      2021, Current Opinion in Insect Science
      Citation Excerpt :

      Finally, imaging calcium signals from ellipsoid body neurons in head-fixed walking flies, the authors dissected how the different structural motifs of this network facilitate its functioning as a ring attractor (a compass-like neural network where neurons tuned to the same head direction excite each other, but neurons encoding different head directions inhibit one another). VR techniques have also been successfully used to study the neural bases for a wide range of behaviors including, looming objects and motion parallax in locusts [53], stereopsis in the praying mantis [54], sensorimotor flight control in Drosophila [55],and selective attention in honeybees [28]. While there are some restrictions imposed on the mobility of the organism, this approach offers a great alternative when outdoor studies can be too expensive or infeasible (e.g. due to weather conditions).

    • Using virtual worlds to understand insect navigation for bio-inspired systems

      2020, Current Opinion in Insect Science
      Citation Excerpt :

      With the advent of newer tether designs that can freely yaw, partial aspects of free flight have been recovered [82]. Recently, using real-time 3D tracking and rapid computation, freely flying insects were provided with closed-loop visual feedback [33••,34,65]. What this VR lost in manipulative abilities of imaging and neural recording, it gained by simulating genuine free flight in VR.

    View all citing articles on Scopus
    View full text