Award details

Revealing the principles of auditory pattern recognition and steering: a combined neurobiological and engineering study

Reference	BB/D004241/1
Principal Investigator / Supervisor	Professor Berthold Hedwig
Co-Investigators / Co-Supervisors	Professor Barbara Webb
Institution	University of Cambridge
Department	Zoology
Funding type	Research
Value (£)	266,284
Status	Completed
Type	Research Grant
Start date	21/11/2005
End date	20/02/2009
Duration	39 months

Abstract

Our aim is to obtain a comprehensive understanding of the neural circuits underlying the complex auditory pattern recognition and localisation behaviour of crickets. Our previous experiments, using a highly sensitive trackball system we developed to obtain precise measurements of the cricket's response, demonstrated that: 1. Crickets rapidly steer towards individual sound pulses with a latency of just 55-60 ms. Proposed pattern recognition mechanisms based on temporal filtering are too slow to be involved in producing these steering responses. 2. The animals always generate small steering responses towards sound pulses, independent of pattern recognition. However, when the correct pattern is recognised, the gain of auditory steering is substantially increased. Crickets will thus steer towards non-attractive test pulses if they are inserted into an attractive song. These results indicate that steering is modulated by pattern recognition. Based on these findings, we have proposed a neural network for the behaviour, which has been demonstrated to produce similar results when implemented on a robotic cricket. To test the hypotheses embodied in this network, we will now identify and analyse the cricket's neural circuitry by intracellular recording, staining and stimulation of single auditory neurons in the brain of walking crickets. Central topics are: 1. How are steering signals generated? What is the nature of the descending control signal / does it closely specify the motor response or provide only high-level modulation of thoracic motor processes? We will record the activity of descending brain neurons and identify cells with activity closely related to the steering response. We will present split-song patterns with each successive pulse being presented from alternate directions to reveal the nature of directional processing in the brain. Furthermore we will use intracellular current injection to manipulate the activity of descending neurons, to identify those interneurons that elicit significant changes. 2. How is pattern recognition organised? Is the sound information from both sides combined or kept separate in the brain? Pattern recognition is performed either by one central network or by two separate networks, one at each side of the brain. We will perform behavioural tests with low amplitude split-song paradigms. If one recognition network sums information from both sides the threshold for phonotaxis should the same as if the sound pulses were presented from one side only. The outcome of the behavioural tests will be used to identify local brain neurons of the recognition network If one network integrates the information then recognition neurons should respond to pulses presented to each ear. 3. What is the neural mechanism of the modulation process that increases the gain of auditory steering? We will analyse which neurons change their response properties and start to respond to non-attractive test pulses during phonotaxis. We consider that the modulation could either take place in the brain or by local processing in the thoracic ganglia. If the modulation is localised in the brain, then we will inject neuroactive substance into the neuropil to alter the process and we will identify efficient neurochemicals to interfere with it. If we do not find any evidence in the brain then we will analyse the thoracic auditory-to-motor interface as a site of modulation. At each stage of the analysis the neurobiological results will be implemented into an artificial neural network controlling the auditory responses of a robot cricket. The modelling studies will test and identify if the neurobiological data are sufficient to explain the cricket's behaviour, and make predictions about necessary elements, to guide the neurophysiological experiments. This integrated approach will lead to a fundamental understanding of auditory processing and the sensory organisation of a complex behaviour.

Summary

Sensory processing is fundamental animal behaviour: it leads to the recognition of complex stimuli and causes adaptive behavioural responses. We aim to reveal a complete sensory-to-motor pathway from the perception of sensory signals, to the central processing of these signals and the generation of the appropriate motor commands. We will use neurobiological methods to analyse the neural principles underlying sensory processing and the collaborating engineers will then use these data for the design of artificial neural networks in biologically inspired robots. These modelling studies will test hypotheses derived from the biological analysis and in turn will feed back workable designs to formulate further analysis. We will focus on sound communication in crickets. Male crickets produce a simple species-specific pattern of sound pulses. Females recognise and localise the song and walk toward the singing male. This phonotactic behaviour can be studied with animals walking in the same position on a trackball that enables the movements to be recorded accurately while the animals are exposed to natural experimentally generated sound patterns How does the nervous system deal with the auditory patterns and extract the necessary information? Pattern recognition and localisation are performed by a small number of neurons that can be identified as individuals. Sensory neurons from the ears on the front legs project to a thoracic ganglion from which projection neurons forward the auditory information to the brain. Current concepts propose that pattern recognition occurs in the brain by a series of temporal filters and that it is essential for initiating and directing the steering movements. Our recent experiments in which we measured with high time resolution the behavioural responses of females to experimental sound patterns demonstrated rapid steering movements to individual sound pulses that cannot involve the proposed recognition process. Moreover, once the animals recognisea song the amplitude and tuning of the rapid auditory responses is modulated. For example, if 100 ms sound pulses are presented on their own, they elicit only a small steering response but when inserted into a sequence of species-specific song, the animal's response is much greater. These changes cannot be explained by previous models of auditory pattern recognition and steering. They indicate a modulation of sensory processing that so far has not been described in any other auditory pathway. How can we analyse the underlying neural processing? We will record the activity of auditory neurons in the brain of phonotactic walking crickets allowing us to analyse the ascending thoracic interneurons forwarding information to the brain, local interneurons of the recognition network and descending interneurons acting on thoracic motor networks while at the same time relating their actions to the resulting behaviour. To identify neurons involved in recognition, steering and modulation of the responses we will generate artificial song patterns. For example, we will insert test pulses into the species-specific song and analyse if brain neurons will start to respond to these non-attractive patterns? If they do, we then will analyse the nature of these changes. At all stages of our research the neurobiological findings will be implemented into a neural network controlling the behaviour of a robotic cricket and their functional significance will be tested. Insight and feedback from the modelling studies will be used to identify crucial steps in sensory processing and additionally guide and specify the neurophysiological experiments.

Committee	Closed Committee - Animal Sciences (AS)
Research Topics	Neuroscience and Behaviour, Systems Biology
Research Priority	X – Research Priority information not available
Research Initiative	X - not in an Initiative
Funding Scheme	X – not Funded via a specific Funding Scheme

I accept the terms and conditions of use (opens in new window)

export PDF file

back to list new search