Vibration Sensitivity and a Computational Theory for Prey-Localizing Behavior in Sand Scorpions

As burrowing, nocturnal predators of small arthropods, sandscorpions have evolved exquisite sensitivity to vibrationalinformation that comes to them through the substrate they liveon, dry sand. Over distances of a few decimeters, sand conductslow velocity (50 m/sec) surface (Rayleigh) waves of sufficientamplitude and bandwidth (200<f<500 Hz) to be biologicallydetectable. Eight acceleration-sensitive receptors (slit sensilla)at the tips of the scorpion's circularly arranged legs detectsurface waves generated by prey movements or "juddering" signalsfrom other scorpions. From this input alone, direction of thedisturbance source is calculated up to 20 cm distance. By ablatingslit sensilla in various combinations on the eight legs, thecontribution each makes in computing target location can beassessed. Other behavioral experiments show that differentialtiming of surface wave arrival at each sensor is most likelythe cue that determines target location. Given the simplicityof this sensory system, a computational theory to account forwave source localization has been developed using a populationof second-order neurons, each receiving excitatory input fromone vibration receptor and inhibition from the triad of receptorsopposite to it in the eight-element array. Input from a passingsurface wave opens and closes a time widow, the width of whichdetermines the firing probability of second-order neurons. Targetdirection is encoded as the relative excitation of these neurons,and stochastic optimization tunes the relative strengths ofexcitatory and inhibitory inputs for accuracy of response. Theexcellent agreement between predictions of the model and observedbehavior of sand scorpions confirms a simple theory for computationalmapping of surface vibration space.