Faux frogs: multimodal signalling and the value of robotics in animal behaviour

A long-standing interest in animal behaviour has been to understand chains of stimuluseresponse. One approach to understanding the links between stimulus and response has been to use models (animal replicas) in which details of a behaviour can be manipulated to determine how such manipulations influence the response of the focal animal. Static models (Tinbergen & Perdeck 1950; Searcy 1998; McLister 2003), manually controlled robots (Brown & Kiely 1974; Taylor et al. 2007) and motorized static models (MacLaren et al. 2004; Gumm et al. 2006) have been effective in eliciting responses from animals in behavioural studies. The current availability, however, of low-cost electric motors, easily designed circuit boards and a wide variety of sculpting materials has created even greater possibilities for developing robots as tools in studies of animal behaviour. In recent years a number of workers have capitalized on these technologies which are now increasingly being employed in controlled experiments (Knight 2005). In some research programmes, robots have been developed as a physical algorithm to test hypotheses about the mechanisms of behaviour (reviewed in: Webb 2000). In these studies, biological systems are modelled with robots and the behaviour of the robots is analysed in response to some stimulus input. Examples of these include studies of navigation (Lambrinos et al. 2000) and chemical trail following (Kuwana & Shimoyama 1998; Grasso et al. 2000). Robotics have also been used as tools in studies where the robot interacts with living animals and it is this arena in which we are