Figure 1.
Experimental set-up of the cyborg's search task.
(A) Schematic general set-up: the cyborg starts 2 m from the pheromone source in a 2.54 m region. A fan provides a wind blowing from the top (towards the cyborg). (B) Photo of our cyborg: a Khepera III robot with a moth fixed in a styrofoam roll. Zoom-in 1: top of the styrofoam roll with the insect's head and the two antennae on the outside. Zoom-in 2: one antenna enters the tip of a glass electrode. Photographs by H. Raguet — INRIA.
Figure 2.
Reactive search strategies and their biological motivation.
(A) MGC recordings for pheromone stimulation: spike times for seven trials of one neuron and the corresponding average firing rate over time (Peri-Stimulus-Time-Histogram): inhibition separates the On from the Off response which smoothly decreases to baseline firing (Bl). (B) Analysis of MGC recordings of multiphasic neurons: Calculating the regularity () and reliability (
) over time exhibits an Off phase, whereas Off and baseline firing show uniformly low synchrony values (
). Dotted black lines represent single neuron trials, the red line gives the averages. (C) Analysis of MGC recordings of monophasic neurons: neither synchrony nor regularity nor reliability over time exhibit any Off phase. Dotted black lines represent single neuron trials, the blue line gives the averages. (Right side: za, ze, sp) Schematic representation of the corresponding movement sequences: Bl
spiraling, On
upwind surge, and Off
zigzagging (if considered) which are combined into three search strategies, sp, za, and ze.
Table 1.
Experimental robotic trials.
Figure 3.
Success rates, trajectory lengths, and deviation from the optimal path.
(A) Success rates of reactive search strategies, different colors indicate different strategies (legend in C), grouping indicates different stimulation doses (three doses plus no stimulation). (B) Success rates of cognitive searching with infotaxis (three doses plus no stimulation). (C) Trajectory lengths of reactive search strategies, different colors indicate different strategies, grouping indicates different stimulation doses. (D) Trajectory lengths of cognitive searching with infotaxis (three doses plus no stimulation). (E) Schematic drawing to explain the measure: the average of horizontal deviations (Xi) from trajectory to shortest path between start and source. (F) Deviation from the optimal path (
) for reactive searching, different colors indicate different reactive strategies for the three groups of pheromone doses. (G) Deviation from the optimal path (
) for cognitive searching (three doses plus no stimulation). Box plots are explained in the Methods, part 2, the numbers indicate mean
standard deviation.
Table 2.
Statistics of pairwise comparisons.
Figure 4.
(A) Examples of sp search trajectories (spirals only, i.e., no Off), medium dose. For a better visualization single paths are plotted in distinct colors (cyan and light blue on top of mostly blue trajectories). The dots on the trajectories indicate pheromone detections. The black dashed line indicates the plume contour (see Methods). (B) Examples of search trajectories including Off zigzagging, medium dose. Red, yellow and pink trajectories use arithmetic spirals (za), bluish trajectories originate from assuming exponential spirals (ze). Identical conventions as in (A). (C and D) Track-angle histogram of sp and za trajectories, respectively, different colors indicate different pheromone doses. (E) Total number of turns for different stimulations, different colors indicate different reactive strategies for the three groups of pheromone doses, identical conventions as in Fig. 3.
Figure 5.
Cognitive search trajectories obtained using infotaxis.
(A) Example it trajectories for no stimulation (green, left), minimum (dark green, middle) and medium (cyan, right) stimulation doses. The dots on the trajectories indicate pheromone detections. The black dashed line indicates the plume contour (see Methods). (B) Total number of turns in it trajectories for different stimulations. Identical conventions as in Fig. 3. (C) Track-angle histograms of it trajectories, different colors indicate different doses. (D) Total number of detections measured during reactive (sp, za, ze) and cognitive (it) searching using three stimulation doses and no stimulation. Identical conventions as in Fig. 3.
Figure 6.
Contrasting juxtaposition of our trajectories to behavioral data.
(A) Two representative reactive (za) search trajectories: the pink path is characteristic for a maximum stimulation dose, the red path is a typical result of medium or low stimulation. The dots on the trajectories indicate pheromone detections. (B) Two representative infotaxis trajectories: the cyan path is characteristic for using the maximum or medium dose, the green path is a typical result of minimal stimulation. (C) Three exemplary behavioral search trajectories, provided by the Kanzaki-Takahashi Laboratory (University of Tokyo) [36]. A silkmoth walking in a wind channel started at (0,0.6). The source was at (0,0) but was considered to be reached when within 5 cm of the latter. (D) Track-angle histograms (bin size = 10°) of the trajectories shown in (C). (E) Tabular comparison of the resulting trajectory lengths of strategic versus behavioral searching.