A Plausible Mechanism for Drosophila Larva Intermittent Behavior
297
feedback [7]. A ‘visceral pistoning’ mechanism involving head and tail-segment
synchronous contraction underlies stride initiation [4]. Speed is mainly controlled
via stride frequency [4]. Crawling is intermittently stopped during both stimulusfree exploratory behavior and chemotaxis, giving rise to non-stereotypical stationary bouts during which reorientation might occur. During the former they
are intrinsically generated without need for sensory feedback or brain input
[9], while during the latter an olfactory-driven sensorimotor pathway facilitates
cessation of runs when navigating down-gradient. Specifically, inhibition of a
posterior-segment premotor network by a sub-esophageal zone descending neuron deterministically terminates runs allowing easier reorientation [11].
It is reasonable to assume that this intermittent crawling inhibition is underlying both free exploration and chemotaxis, potentially in the form of transient
inhibitory bursts. A neural network controlling the CPG through generation of
self-limiting inhibitory waves is well suited for such a role. In the simplest case,
during stimulus-free exploration, the durations of the generated inhibitory waves
should follow a power-law distribution, behaviorally observed as rest bouts. In
contrast, non-power-law distributed quiescent periods of the network would disinhibit locomotion allowing the CPG to generate repetitive peristaltic strides
resulting in behaviorally observed runs.
The model we presented (cf. Sect. 3.1) alternates autonomously between
avalanches of power-law distributed durations and quiescence intervals of exponentially distributed durations without need for external input. Therefore it can
serve as a theoretical basis for the development of both generative models that
reproduce the intermittent behavior of individual larvae and of the above mechanistic hypothesis for the initiation and cessation of peristaltic locomotion in
the larva through disinhibition and inhibition of the crawling CPG respectively.
To uncover the underlying neural mechanism and confirm/reject our hypothesis,
inhibitory input to the crawling CPG should be sought, measured neurophysiologically and correlated to behaviorally observed stride and stride-free bouts
during stimulus-free exploration.
Intermittent behavior in the Drosophila adult is subject to two modes of
modulation, neither of which affects the distribution of the activity bouts.
Firstly, high ambient temperature and daylight raise the activity ratio over
long timescales by raising the number of activity bouts [12]. This is achieved
by lowering the probability of the extremely long rest bouts, without affecting
the power-law exponent of the distribution, which coincides with fewer sleep
events (>5 min) observed during the day. This modulation is long-lasting and
could result from a different constant tonic activation of the system. Secondly,
dopamine neuron activation raises the activity ratio acutely by modulation of
the power-law exponent upwards [12] skewing locomotion towards the brownian limit. This modulation could be transient in the context of salient phasic
stimulation by the environment.
As mentioned above, during chemotaxis larvae perform more and sharpest
reorientations, terminating runs when navigating down-gradient. In case the
above hold for the larval nervous system as well, a hypothesis integrating both
experimental findings could be that this behavior stems from transient olfactorydriven dopaminergically-modulated inhibition of the crawling CPG. Our concep-
297
feedback [7]. A ‘visceral pistoning’ mechanism involving head and tail-segment
synchronous contraction underlies stride initiation [4]. Speed is mainly controlled
via stride frequency [4]. Crawling is intermittently stopped during both stimulusfree exploratory behavior and chemotaxis, giving rise to non-stereotypical stationary bouts during which reorientation might occur. During the former they
are intrinsically generated without need for sensory feedback or brain input
[9], while during the latter an olfactory-driven sensorimotor pathway facilitates
cessation of runs when navigating down-gradient. Specifically, inhibition of a
posterior-segment premotor network by a sub-esophageal zone descending neuron deterministically terminates runs allowing easier reorientation [11].
It is reasonable to assume that this intermittent crawling inhibition is underlying both free exploration and chemotaxis, potentially in the form of transient
inhibitory bursts. A neural network controlling the CPG through generation of
self-limiting inhibitory waves is well suited for such a role. In the simplest case,
during stimulus-free exploration, the durations of the generated inhibitory waves
should follow a power-law distribution, behaviorally observed as rest bouts. In
contrast, non-power-law distributed quiescent periods of the network would disinhibit locomotion allowing the CPG to generate repetitive peristaltic strides
resulting in behaviorally observed runs.
The model we presented (cf. Sect. 3.1) alternates autonomously between
avalanches of power-law distributed durations and quiescence intervals of exponentially distributed durations without need for external input. Therefore it can
serve as a theoretical basis for the development of both generative models that
reproduce the intermittent behavior of individual larvae and of the above mechanistic hypothesis for the initiation and cessation of peristaltic locomotion in
the larva through disinhibition and inhibition of the crawling CPG respectively.
To uncover the underlying neural mechanism and confirm/reject our hypothesis,
inhibitory input to the crawling CPG should be sought, measured neurophysiologically and correlated to behaviorally observed stride and stride-free bouts
during stimulus-free exploration.
Intermittent behavior in the Drosophila adult is subject to two modes of
modulation, neither of which affects the distribution of the activity bouts.
Firstly, high ambient temperature and daylight raise the activity ratio over
long timescales by raising the number of activity bouts [12]. This is achieved
by lowering the probability of the extremely long rest bouts, without affecting
the power-law exponent of the distribution, which coincides with fewer sleep
events (>5 min) observed during the day. This modulation is long-lasting and
could result from a different constant tonic activation of the system. Secondly,
dopamine neuron activation raises the activity ratio acutely by modulation of
the power-law exponent upwards [12] skewing locomotion towards the brownian limit. This modulation could be transient in the context of salient phasic
stimulation by the environment.
As mentioned above, during chemotaxis larvae perform more and sharpest
reorientations, terminating runs when navigating down-gradient. In case the
above hold for the larval nervous system as well, a hypothesis integrating both
experimental findings could be that this behavior stems from transient olfactorydriven dopaminergically-modulated inhibition of the crawling CPG. Our concep-