Tactile sensors such as vertebrate whiskers or arthropod antennae are important for near-range orientation. As many insects, the stick insect Carausius morosus samples the surrounding area by continuously moving its antennae during locomotion (Dürr et al. 2001, J Comp Physiol A 187). Antennal contact can trigger fast reaching movements of a front leg. In the present study, we identify properties of antennal tactile information input to the prothoracic ganglion and, potentially, to motor networks of the front leg. In a first step we probed for descending interneurons (DINs) which responded to touch, bending or vibration stimuli applied to the flagellum. During natural contact with obstacles, bending of the antennal flagellum can be very strong, and release of obstacle contact causes a fast, over-damped return to resting posture.
Activity of neck connectives was recorded extracellularly. Intracellular recordings from DINs were performed in the anterior medial part of the prothoracic ganglion near the entry of the neck connectives. DINs were characterized by their responses to touch at three flagellar sites, horizontal displacement with ramp-and-hold stimuli of different velocity and displacement, and vibration within a frequency range from 0.5 to 300 Hz. Both antennal joints were fixed (passive sensing).
We recorded from 37 DINs that responded to stimulation of the flagellum by at least one of the three mentioned submodalities. 18 DINs were spontaneously active with spike rates of 0.2 to 20.5 Hz. Analysis of response characteristics revealed several groups of DINs: the majority (N=14, group 1) coded for velocity as well as frequency of antennal deflections. Touch responses of group 1 were often strongest at proximal stimulus sites (N=12), never at distal sites. 12 DINs showed velocity coding but no frequency dependence (group 2). 9 of these had a monotonous velocity dependence. Neither velocity nor frequency coding was obvious in 4 DINs (group 3), however touch responses were significant in 3 of them. Spike timing variability was less than 1 ms in 17 DINs. The shortest response latency observed was 6.7±0.2 ms.
Several DINs had their soma in the subesophageal ganglion (SEG). One, a typical group 1 cell, responded best to lateral deflection of the ipsilateral antenna (ramp velocity: 2.3 mm/s). Mean spike rate was up to 20 Hz for vibration stimuli. 3 of 4 DINs in group 2 had their soma in the SEG. In a typical example, velocities of up to 60 mm/s elicited mean spike rates of 20 Hz.
The response characteristics and short latencies of some of the described DINs make them likely candidates for the neural substrate mediating tactually elicited reaching behavior.

In contrast to whiskers of mammals, insect antennae not only transmit but also sense mechanosensory cues of various sub-modalities, including contact location, bending and vibration. Many insects actively move their antennae to sample the space ahead. In the walking stick insect Carausius morosus, the antennal movement pattern is rhythmical, coordinated with the stepping patterns of the legs, and oriented towards the walking direction. Antennal joints of stick insects are non-orthogonal and slanted revolute joints, an arrangement known to improve tactual acuity.
Here we report results from motion capture experiments on unrestrained walking C. morosus with and without the presence of obstacles (a vertical rod of 6 mm diameter or steps in the height range 19 to 66 mm). Joint angles were measured for both the head scape joint (HSj) and scape pedicel joint (SPj).
During straight walking, the two antennal joints were strongly coupled to each other with a mean phase lag of 30 deg., resulting in elliptic movements relative to the head. The azimuth of head and prothorax were strongly coupled to the front leg step cycle, adding approx. 30 deg. of azimuth to antennal orientation. Thus, rhythmic head and prothorax movements may induce coupling of antenna and front leg.
When touching the rod with an antenna, animals responded with an aimed reaching-movement. During reaching, up to three further antennal contacts occurred (med. duration: 90 ms), each one above the previous. The location of the last antennal contact significantly predicted leg contact location. When the first antennal contact occurred during a swing movement, the tarsus was redirected towards the contact location with a mean delay of 40 ms.
Upon the first contact with a rod, cycle frequencies in both antennal joints increased from approx. 2 to 6 Hz and more. The prevalent response to antennal contact was HSj levation followed by SPj abduction (med. delays: 22.5 and 50 ms, resp.), resulting in an upward movement. After the first antennal contact, inter-joint coupling of HSj and SPj switched from coupling of downward to upward movements. Owing to the slanted joint axes, this change in coupling is an efficient strategy to sample a vertical edge.
When climbing a step, bilaterally symmetrical movements of both antennae, caused by contralateral coupling of functionally analogous joints (e.g. left HSj with right SPj), was common. Also, cycle frequency of both joints increased after antennal contact while joint amplitudes were reduced.
We conclude that stick insects alter their tactile sampling behavior after the first obstacle contact by an increase of cycle frequency and context-dependent changes in inter-joint coupling.

Autonomous behaviour such as locomotion in natural terrain requires continuous integration of sensory information from different modalities. In terrestrial locomotion, tactile and visual cues are of particular relevance for the control of heading, speed and foot placement. As yet, the working ranges of tactile and visual sensing differ strongly: Tactile sensing yields near-range cues about local obstacles that may affect the execution of the next step. Vision yields additional far-range cues about distant features that are important for choice of heading, course control and navigation during many steps. An intriguing feature of these two sensory systems is that both of them affect motor behaviour and are affected, themselves, by motor behaviour such as active head movements.
Here we study the effect of active head movements on tactile and visual sampling in walking stick insects (Carausius morosus). Unrestrained walking insects were tracked either on a straight walkway (120x4 cm) or in a round arena (r=60 cm) that held three visual landmarks (either black stripes on a white wall, subtending 5° as seen from the centre, or black pillars of 3 cm diameter, placed at 27-51 cm radial distance on the corners of an arbitrary triangle). In the arena, animals were released near the centre and approached one of the three targets. Kinematic analysis was based on tracked retro-reflective markers on various body segments (0.25mm resolution at 50-100 fps). Trial durations were 11-24s on the walkway (n=10; N=5) and 5-30 s in the arena (n=120; N=6). The dependence of choice likelihood for one of three targets on visual eccentricity of targets was tested in 6 animals (n=240; three 2.5° black stripes). Finally, we reconstructed a compound eye of one stick insect from a stack of histological sections. 3D orientation of all 467 ommatidia yielded both the visual field and spatial resolution.
On the walkway, horizontal head movements covered a range of ca. 12° and were strongly coupled to front leg stepping. In the arena, the frequency of rhythmic head movements increased linearly with walking speed, also indicating strong coupling with stepping. The turning points of the head nearly coincided with the end of stance of the ipsilateral front leg. This improved the coupling between antenna and front leg. Head and prothorax movements increased the horizontal working range of the antennae by ca. 30%.
Since insects lack eye movements, only head movements control their gaze. Head movements never stabilized the gaze against thorax rotation. As maximum spatial resolution of the eye is 4° (at 45° azimuth) and declines to >10° toward both ends of the visual field (0-165°), head movements cause an image slip over up to three ommatidia. This may decrease the effect of local visual adaptation. As the eyes are located 3mm distal to the neck, head movements induce lateral translation of the eyes by 0.6mm (equiv. to ca. 20 ommatidial diameters) and associated motion parallax. Prothorax movements further increased these effects. However, neither head nor thorax fixation affected the choice likelihood distribution for alternative targets in the arena, suggesting that head movements do not affect visual scene analysis. Also, active horizontal gaze changes showed no relation to target distance or size, suggesting that the head is not involved in active fixation.
We conclude that active head movements affect tactile near-range sampling more than far-range vision.