Learning from Crickets

The wind whistles through the grass and the cricket sits calmly on the ground. All at once a wolf spider appears, ready to attack. Its approach is almost soundless amongst the general rustling of leaves and stalks. But the cricket senses the characteristic vibrations made by the predator, and as fast as lightning it gets out of harm’s way. “How the cricket manages to filter out such quiet signals in a noisy environment is what we would like to be able to understand better, partly so that we can use it for technical applications,” says Professor Andreas Offenhäusser of the Institute of Biosystems and Nanosystems at Forschungszentrum Jülich. The physicist coordinates the European project CILIA (Customised Intelligent Life-inspired Arrays), in which experts in biology, physics and engineering sciences from the universities in Bonn and Munich (in Germany), Antwerp (in Belgium), and Twente in the Netherlands, from Odense in Denmark, Reading in the UK, Tours in France and Shandong in China are participating.

Crickets, fish, bats and many other living creatures often use several hundred sensitive sensors in order to gain their bearings or to recognise dangers quickly. Here both the geometrical arrangement of these sensors and also the subsequent processing of the incoming signals in the nerve cells play a crucial role in determining how efficiently the sensory network filters out the important signal from the rustling sound. Compared with what nature can do, technology is still very primitive and achieves a much poorer result even with more components and using more energy. Anyone who has a hearing aid experiences this when trying to concentrate on a conversation at a noisy party. The “danger detectors” in the case of the cricket are apendages on the rear body (cerci) which are covered with several hundred very thin hairs. At the hair roots there are nerve cells, which measure the air stream via the deflection of the hairs. These signals are processed in a nerve node in the rear body of the cricket, which consists of only a few hundred nerve cells. Scientists suspect that unimportant information is already filtered out at this stage, so that the succeeding nerve centres are not overburdened with data and can quickly triggerthe life-saving reflexes. “We are above all interested in the principles which the sensor systems of different species have in common, as we can learn most from these,” says Offenhäusser.

Researchers at Tours in France simulate the air streams of various ecological situations in the laboratory and test how the cricket reacts to them. “These experiments are very important for us, since now we know what the incoming signals which are processed in the nerve node look like,” explains Offenhäusser. The engineering scientists at the University of Twente have in fact already constructed artificial sensors modelled on the cricket hairs, which they also arranged as they appear on the cricket itself. In this way they can examine how the sensors work together. Offenhäusser and his team are working on how to decipher the way in which the signals are processed in the nerve node of the cricket’s rear body. They are looking at both living crickets and single nerve cells in order to understand how signals are passed on and how the nerve cells link up to become a neural network, which can calculate the correct reaction from the incoming “noise”. For this purpose they place the nerve cells on a substrate, where they can link up with one another. Using a procedure which is reminiscent of potato printing, they first impress thin nanometre patterns consisting of specific proteins on the substrate.

The nerve cells, which they have previously extracted from the nerve node on the rear body of the cricket, adhere to the proteins and interconnect along the predefined structures. In this way Offenhäusser tests how certain patterns affect the performance of the neural network. Placed in a line, each nerve cell has only two neighbours, so that the complexity is small, whereas in right-angled networks each cell can communicate with four neighbours, and in honeycomb-like structures with six neighbours. “However, it is extremely difficult to place large functioning networks consisting of live nerve cells on this substrate; we are working here at the limits of what is possible,” Offenhäusser explains. The Jülich group is also working on the development of interfaces between biological and electronic systems by linking nerve cells with components such as transistors. The intelligent networking which crickets, fish or bats have optimised in their sensor fields in the course of evolution could be transferred to the design of circuitry for technology, the CILIA partners hope. And thus in the long term hearing aids could be improved, so that even conversations in the middle of a lively party will once again become a pleasure.