The modulating role of calcium on the accuracy of mechano-electrical transduction of sensory hair cells will be investigated. This will be accomplished by simultaneous recordings of transducer current and hair bundle displacement of cochlear outer hair cells. The stiffness of the gating spring is a pivotal factor in current (gating spring) models of mechano-electrical transduction of vertebrate hair cells. If the gating spring stiffness is affected by calcium, as some experimental reports suggest, this may have important consequences for the accuracy with which basilar membrane vibration is detected by sensory hair cells. The project is expected to yield quantitative measures on the controlling action of calcium on the accuracy of mechano-electrical transduction in hair cells.

The frog inner ear includes two auditory papillae, the amphibian papilla and the basilar papilla. These papillae are unique among vertebrate hearing organs in that the hair cells are not on a basilar membrane, but imbedded in a relatively stiff support structure. The frog inner ear generates otoacoustic emissions with properties that are very similar to those of emissions in other vertebrate species (including humans). Recent data shows that the amphibian papilla generates both spontaneous and distortion product emissions. In contrast, the basilar papilla generates no spontaneous emissions, but only distortion product emissions. In addition, the distortion product emissions from the basilar papilla are much less sensitive to temperature changes as compared to emissions from the amphibian papilla. This suggests that the basilar papilla may be a passive hearing organ, incapable of generating spontaneous emission, and with distortion product otoacoustic emissions arising from passive distortion alone. In contrast, the amphibian papilla appears to be active, in the sense similar to active hearing in other vertebrates.
The difference between these papillae offers the opportunity to study the effect of active processes on auditory mechanics within a single hearing organ. Thus, we aim to measure the mechanical response of the tectorial membranes of the amphibian and basilar papillae, respectively. Currently, an experimental setup is developed to record tectorial membrane motion using an optical sectioning technique. The aim of this project is to describe the mechanical responses in the amphibian papilla and the basilar papilla, in order to better understand the role of the active mechanism.

Photoreceptors of all vertebrate retinas are electrically coupled. In the light of high spatial resolution such coupling is at first glance counterproductive. Several recent reports have addressed this aspect and could show that the coupling is not general but task-specific and subserves a variety of functions. During the past granting period, we were able to demonstrate that cone photoreceptors express connexin36, and that this connexin is most likely involved in the formation of gap junctions. Connexins expressed by rods are not yet elucidated although it is now accepted that rods not only make gap junctions with cones but form also gap junctions among each other. The molecular analysis of rods is very difficult due to their small size but we have developed a method to collect cytoplasma from several rods using a patch pipette. We will use this method to search for connexins expressed by rods. Antibodies will be generated against potential candidates and employed in immunocytochemical studies at the confocal and electron microscopic level.

The synaptic mechanisms underlying the negative feedback between horizontal cells and photoreceptors are still a matter of debate (see Special Interest Symposium, Annual Meeting of the Association for Research in Vision and Ophthalmology, USA, 2005). We have recently proposed an ephaptic mechanism based on hemichannels (Kamermans et al., 2001) and were able to demonstrate its presence in the non-mammalian retina. Physiological evidence from the primate retina suggests that such an ephaptic mechanism is also present in the mammalian retina but its molecular verification is still lacking. We will address this issue in a project analysing the feedback in the mouse retina exploiting its transgenic potential. First we will record intracellularly from horizontal cells and establish the functional profile of the feedback. Second we will search for hemichannels expressed by mouse horizontal cells using single cell RT-PCR techniques. Third we will generate antibodies against candidates and localize its presence at the ultrastructural level. Fourth we will generate corresponding knockout animals and analyse their retinae electrophysiologically.

Horizontal cells are engaged in a negative feedback loop with the photoreceptors, which is the essential circuit for neuronal adaptation within the outer plexiform layer of the retina and supposed to be crucial for a constant perception. A direct demonstration of such an involvement has not been possible so far since interference with the synaptic interactions always affects multiple sites and prevents an unequivocal dissection of the horizontal cell contribution. Horizontal cells are also electrically coupled, and we could recently demonstrate that this coupling involves connexin57 (Cx57; Hombach et al., 2004), and that horizontal cells are the only neurons expressing Cx57. This exclusivity opens the door to genetical ablation of horizontal cells at different points of development and eventually the generation of mice without functional horizontal cells. Since horizontal cells are among the last generated neurons within the retina, the impact on the overall structure of the retina is expected to be rather limited. Once generated, such mice would be of great value for an in-depth study of the functional contribution of horizontal cells to retinal processing. The genetical part of the project will be done in very close collaboration with the Institute for Molecular Genetic of the University of Bonn with which we have already a long standing and successful collaboration. The physiological part including behavioural studies would be done in Oldenburg.

Temporal correlation of neural activity in the central nervous system, e.g. as found in the form of gamma-band activity, is critically affected by a network of inhibitory interneurons that are electrically coupled by gap junctions. These gap junctions are formed by Connexin 36 (Cx36). A study by Buhl et al. (2003) has demonstrated reduced gamma-band oscillations in the hippocampus of the Cx36-knock-out mouse indicating a reduced temporally correlated activity. In the planned project it will be tested whether gamma-band oscillations occur in the mouse auditory cortex and are reduced in the Cx36-knock-out mouse. Stimulus induced gamma-band oscillations have been shown to exist in the monkey auditory cortex (Brosch et al., 2002). These oscillations that result from the inhibitory network dynamics may play an important role in across-frequency processing in the context of auditory grouping. It will be tested whether the neural across-frequency processing of acoustic stimuli is changed in the primary auditory cortex of the Cx36-knock-out mouse compared to the wild-type mouse when presented with stimulus paradigms that will lead to auditory grouping of the frequency components in the presented sounds.

Detection and classification of sensory stimuli is an important function of the nervous system. In the visual system, all information that is transmitted from the retina to the brain must be encoded by the responses of the retinal ganglion cell (RGC) ensemble. It is still under debate which features (e.g. spike rate, latency, synchronous activity, spike patterns) of the RGC responses are relevant for the central nervous system to decode visual information. Since most RGCs fire sparsely, we expect them to use firing patterns and relative spike timing between members of the ensemble to encode visual information. Based on multielectrode recordings from turtle retinal ganglion cells we will use two complementary approaches for data analysis to test these hypotheses about coding principles.
On the one hand, Baysian reconstruction and a physiological model of retinal information processing will be used (e.g. Wilke et al., 2001; Thiel et al. 2005). Based on these methods we will test our hypothesis that the relative timing of spikes in the RGC ensemble transmits information about dynamically changing visual stimuli. In particular, we will investigate the effect of light adaptation on the RGC population responses to find out if the coding principle of relative spike timing could be used by the nervous system for all levels of background illumination.
On the other hand, we will use symbolic dynamics for pattern analysis. This method has proven useful in the analysis of spatiotemporal dynamics, and is suited for an information theoretic analysis of the correlation between dynamic stimuli (input) and firing patterns (output). The finite sample size that is constituted by the data necessitates the introduction of aggregated response variables to comply with the requirements of reliable statistics. The choice of appropriate aggregated variables will be guided by cross-correlation and/or cluster analyses of the recorded RGC responses. Accordingly, the emphasis of this approach will be on the detection of significant correlations between the input and such aggregated output variables. In the same spirit we plan to establish and analyse conceptual models of aggregated population response variables.

The leech is a good model organism to study basic mechanisms of sensory coding because of the simplicity of its nervous system that allows the identification of individual cells between body segments and between animals. With only 14 mechanosensory neurons per segment the leech is able to perceive and encode all behaviorally relevant information about mechanical stimuli of different strengths, positions, sizes, surface structures and temporal dynamics. Our hypothesis is, that the coding of tactile information is based on the relative spike timing between members of the population of mechanosensory cells and that different parameters of tactile information are coded by different aspects of the neuronal responses (e.g. spike rate and latency). We want to test this hypothesis with a combination of electrophysiological experiments and model simulations. Using a preparation with a piece of the body wall attached to a neuronal ganglion enables us not only to electrophysiologically record from sensory neurons during dynamical mechanical stimulation of the skin, but also to stimulate individual neurons electrically while video taping the resulting motor responses (Baca et al., 2005). Intermediate steps of information processing like synaptic interactions can also be analyzed in detail in this preparation (Kretzberg et al., 2005). Hence, the entire process of stimulus encoding, neuronal signal processing and decoding the signal to a motor response can be controlled and appreciated in this simple nervous system.

In the auditory as well as the visual system, the first steps of sensory information processing are based on analog, graded signals. Subsequent to this peripheral preprocessing, digital signals in form of spikes are used to transmit the information over a long distance to the brain. In well-analyzed nervous systems of invertebrates it was shown that sensory information of all modalities is also processed with analog signals (without spikes) in their central nervous systems. Based on a phenomenological model for spike responses in the visual system of the fly (Kretzberg et al., 2001a) we could show that the use of spikes can sharpen the temporal structure of neuronal responses to high-frequency sensory input, while in the low-frequency range graded membrane potentials lead to a better discriminability of neuronal responses (Kretzberg et al., 2001b). Interestingly, in the visual system of the fly the advantages of both modes of coding seem to be combined by superimposing graded membrane potential fluctuations with spike-like processes.
In this project, the advantages and disadvantages of analog versus digital signals for biological neural coding of sensory information will be analyzed for a wide range of stimulus dynamics. We will use an information theoretical approach and model simulations based on electrophysiological data to compare different sensory systems and animal species. In particular, the ability to encode dynamically varying stimuli and the impact of different noise sources on neural coding with analog and digital signals will be investigated.

In previous studies, we have shown that night-migratory birds have a brain cluster (Cluster N) processing specialized visual information at night, but not during the day (Mouritsen et al., 2005). Since Cluster N is the only strongly active, movement independent part of the forebrain in birds that perform magnetic compass orientation at night, we suspect that the visual information processed is magnetically modulated visual signals that provide the birds with compass information. What are the dynamic light-intensity and magnetic field strength ranges under which Cluster N is active and provides stable information in day- and night migrants? Is Cluster N active during the day in day-migrants? And, what kind of information, night-vision and/or magnetic compass orientation, is being processed? Several different physiological, molecular, behavioural approaches can all be used to answer these questions. Behavioral molecular mapping based on mRNA in-situs can be used to compare the brain activity of animals performing "Zugunruhe" with birds not performing "Zugunruhe", and this can be done under various light- and magnetic field conditions. Electrophysiological recordings from the retina and/or from Cluster N can elucidate what type of information is being encoded by the night-vision/magnetic compass neurons. Alternatively, we can use the fact that migratory birds seem to perform head scans to detect the Earth's magnetic field (Mouritsen et al., 2004) as a long sought after psychophysical measure indicating what magnetic fields a bird can sense and which it cannot. In a normal magnetic field, migratory garden warblers make 1 head scan per minute on average and immediately after they performed a head scan they move in the correct migratory direction. In contrast, birds perform about 3 head scans per minute when no magnetic field is available, and a head scan does not lead to movement in a particular direction. In other words, when birds cannot sense the magnetic field they "tell" the experimenter by shaking their head, making it finally possible to determine the dynamic range and sensory limits of the magnetic compass sense of migratory birds. Within this project complex several doctoral theses could be made, since one doctoral student is not likely to be able to use all these approaches simultaneously.

When considering models of the "effective" information processing in the eye and the ear, several similarities become obvious that are probably due to similar biological solutions to the same physical problems, such as, e.g.

1. the dynamic range problem (implemented by light/dark adaptation mechanisms in the retina, and nonlinear, instantaneous dynamic compression in the cochlear)
2. the detection of contrasts (implemented by spatial lateral inhibition in the retina and by onset/offset enhancement in auditory neurons)
3. basic mechanisms of object recognition (implemented, e.g., by correlated change detectors across different receptive fields in the retina and by correlated amplitude modulation detection across different frequency band in the brainstem)

Hence, "effective" models of contrast perception in the visual system (Snippe et al. 2000, 2004, see publication list D. Stavenga) are very similar to "effective" model of envelope processing in the auditory system (Dau et al., 1997, see publication list B. Kollmeier). In addition, higher-order effects, such as second-order envelope perception or spatio-temporal receptive fields in the auditory system seem to operate quite analogous to processing principles of the visual system. The aim of the current dissertation is therefore to bridge the gap between the appropriate models from vision and audition (that have evolved fairly independently of each other) by defining appropriate analogies in the processed perceptual dimensions that lead to similar model structures. Building on this, a hierarchical model structure should be constructed that treats vision and audition in a similar way and combines the output at the level of the "internal representation" of objects.

Appropriate models that characterize the "effective" processing in hearing-impaired listeners and even cochlear implant (CI) patients are comparatively rare (Chatterjee, 1999, McKay et al., 2003). However, they are necessary to optimize the preprocessing in electrical stimulation of the auditory nerve in order to perceive speech even in background noise. The current dissertation project therefore combines psychophysical experiments with CI patients (both from the Evangelisches Krankenhaus in Oldenburg and the CI program in Groningen) with models of the "effective" processing in "electrical" hearing. Specifically, the spatio-temporal extent of the pulsatile stimulation with the individual electrode is measured by appropriate masking experiments. The effect of an electrode combination is measured by loudness summation experiments. Modified loudness models, physiological models (Dau, 1993) and processing models in hearing-impaired subjects (Kollmeier et al., 2001) shall be used to model the "effective elementary excitation" produced by single electrode stimulation in humans and the resulting psychoacoustical effects for combined electrode stimulation an "eventually" complex input signals.

Several modern hearing aids aim to restore the sensation of loudness in the hearing impaired listeners. Since environmental sounds are usually non-stationary the understanding of loudness of fluctuating sound is crucial. Some aspects of loudness of time varying sounds can be modeled reasonably well by assuming that the loudness is related to the average spectrum of the sounds (e.g., Moore et al., 1999). However, recent studies on spectral loudness summation as a function of duration indicate differences in loudness perception at the beginning and at the end of signals, which is not predicted by current loudness models (Verhey & Kollmeier, 2002). The aim of the project is the development of a dynamic loudness model on the basis of the data in the literature and new data gathered in the project, which e.g. determine the relative importance of different time segments of the signals on the loudness perception. Different model hypothesis will be tested which are either peripheral such as dynamic aspects of the cochlear or central, e.g. a limited spectro-temporal capacity of the loudness calculation mechanism. In a second step the model should be extended towards longer signals. Several studies with long-duration environmental sounds have shown that for very long durations the loudness is dominated by intensity peaks rather than the overall intensity averaged across the whole signal and that more recent intensity peaks may be especially important. The project investigates if there is a transition region between loudness dominated by the peaks and loudness essentially dominated by the average intensity of the sound.

Based on previous investigations in the context of the InterGK concerning human vibration perception (Bellmann et al., 2000, 2004; Oey & Mellert, 2004), psychophysical experiments and modeling will be continued both for the hand and the whole body of humans. For vibration perception of the human hand a comprehensive model is not yet available and numerous additional psychophysical parameters related to tactile and vibration perception are still unknown. Effects of JNDs, the relation to the dynamic range, and the relevance to ecologically relevant stimulus configurations are to be investigated. The transition from perception of (tactile) motion (low-frequency vibration) to acoustic perception (high-frequency vibration) including interfering and modulating perception is not yet clear.
As in our previous projects, basic psychophysical investigation of the perception of vibration will be complimented by applied research related to travel comfort, i.e. by whole-body perception of vibration in seats, in particular for the automotive field and in the aircraft. The relation to comfort levels was derived and principal levels of equal perception with respect to frequency and just-noticeable differences in whole-body vibrations were determined. Numerous basic psychophysical parameters (e.g. sensitivity, equal-level contours, masking) are still unknown in the perception of whole-body vibration, and like for the human hand a comprehensive model of whole-body vibration perception is not available up to-date.