Dynamical feature extraction at the sensory periphery guides chemotaxis

Abstract

Behavioral strategies employed for chemotaxis have been described across phyla, but the sensorimotor basis of this phenomenon has seldom been studied in naturalistic contexts. Here, we examine how signals experienced during free olfactory behaviors are processed by first-order olfactory sensory neurons (OSNs) of the Drosophila larva. We find that OSNs can act as differentiators that transiently normalize stimulus intensity—a property potentially derived from a combination of integral feedback and feed-forward regulation of olfactory transduction. In olfactory virtual reality experiments, we report that high activity levels of the OSN suppress turning, whereas low activity levels facilitate turning. Using a generalized linear model, we explain how peripheral encoding of olfactory stimuli modulates the probability of switching from a run to a turn. Our work clarifies the link between computations carried out at the sensory periphery and action selection underlying navigation in odor gradients.DOI: http://dx.doi.org/10.7554/eLife.06694.001

The use of computational modeling to link sensory processing with behavior in Drosophila

“…Neural encoding models can be derived from recordings in fixed, non-behaving animals -these models can then be used for predicting neural responses to the sensory stimuli from behavioral data sets. Thus, computational models can serve as stand-ins for recording neural activity during behavior and thus facilitate overcoming experimental hurdles when linking neural codes and natural behaviors (Parnas et al 2013;Schulze et al 2015;Clemens et al 2015a;Badel et al 2016). Here, we detail this approach using data from Drosophila (both adults and larvae), and we discuss selected studies that highlight both the challenges and advantages associated with computational modeling in this model system.…”

“…(Schwartz et al 2006;Pillow et al 2008;Sharpee 2013;Aljadeff et al 2016)) but has also recently found applications in predicting behavior from both sensory stimuli and neuronal responses (e.g. (Kato et al 2014;Coen et al 2014;Schulze et al 2015;Clemens et al 2015a)). LN models treat the brain as a black box, i.e.…”

“…A number of studies using Drosophila have examined the impact of visual, chemosensory, or thermal cues on navigational decisions (Gomez-Marin et al 2011;Clark et al 2011;Censi et al 2013;Klein et al 2014;Gepner et al 2015;Schulze et al 2015). In these studies, the detailed recording of fly movement parameters with a quantitative, model-supported analysis of the behavior revealed the navigational strategies with which Drosophila negotiates its environment to find food and avoid noxious chemical, heat, or light stimuli.…”

“…Drosophila larval behaviors can easily be classified into three relatively stereotyped motor programs during chemotaxis: phases of straight locomotion (termed "runs"), intense side-to-side movement of the head ("head casting"), and fast reorientation events ("turns"). Models that explain the relation between chemosensory cues and navigation behaviors usually use the rate or probability of performing one of these behaviors as a function of the stimulus (Gomez-Marin et al 2011;Gepner et al 2015;Schulze et al 2015;Hernandez-Nunez et al 2015). Based on an accurate description of the larva's sensory experience as well as of its complete body posture during navigation, the authors determined the features of the odor concentration gradient that triggered turning and head casting using STA-like analyses.…”

“…Bout starts are encoded in positive Vm transients due to adaptation, negative Vm transients encode bout ends due to the negative filter lobe, and the amount of song is represented in sustained responses during bouts due to the positive filter lobe. (Schulze et al 2015) used models to show how responses of larval olfactory sensory neurons (OSN) encode the temporal pattern of odorant concentration as encountered during chemotaxis. Instead of using a linearnonlinear model to represent the dynamical codes underlying concentration coding in these OSN, they employed a model of the signal transduction cascade involved in the transformation of odorant concentration to OSN firing rate.…”