The peripheral adaptation and dynamics has been studied in several types of sensor cells translating a physical stimulus like photon influx or pressure fluctuations into electrical signals. The signals, which receptor cells transmit to their postsynaptic target neurons are in general non-linear functions of the stimulus, with amplitudes and dynamics being shaped by mechanisms like transduction, calcium dynamics and local feedback loops.

For the auditory system, the first important processing stage is the mechano-electrical transduction on the level of the cochlea. We have investigated the nonlinear behavior of the vertebrate inner ear using otoacoustic emissions in frog (Meenderink & van Dijk, 2004) and humans and its relation to model predictions (e.g., Mauermann et al., 2001) using a hydromechanical model of the cochlea (van Hengel et al., 1996). An astonishing property of the cochlea is its ability to detect vibration amplitudes in the order of nanometers. We have identified the transducer channel's molecular gating force as the physical factor controlling this accuracy in proportion to the inverse of its magnitude (Van Netten et al., 2003; Marcotti et al., 2005). The open probability of mechano-electrical transducer channels and thus the micro-mechanical properties of the hair cell bundles depend critically on the extracellular calcium concentration, playing a major role in adaptation (Wiersinga-Post & van Netten, 2002; Dinklo et al., 2003). Further, we have shown that the match of stochastic channel noise to gating spring noise implies that the gating apparatus operates at the threshold of negative stiffness (Van Netten et al., 2003).

Like in sensory hair cells, Ca2+-concentrations play a prominent role also in important photoreceptor functions like stimulus transduction, synaptic transfer and adaptation (Review: Krizaj & Copenhagen, 2002). Recently, we contributed to the identification of a mechanism modulating the Ca2+-current in vertebrate photoreceptors. Ephaptic interactions with horizontal cells cause a local feedback loop, changing the intracellular concentration of Ca2+ in the photoreceptor, which in turn influences chemical synaptic transmission onto bipolar cells (Kamermans et al., 2001; Janssen-Bienhold et al. 2001; Pottek et al., 2003b; Review: Kamermans & Fahrenfort, 2004).

With studying the magnetosensory system of migratory birds, we have the rare opportunity to find out fundamental principles of an almost unknown sensory modality on all levels of investigation. Recently we have gathered evidence that the retina of migratory birds is not only responsible for vision but also for magnetic compass detection. Magneto perception is light-dependent, meaning that the geomagnetic field is sensed indirectly as a modulation of the light-sensitivity of specialized photoreceptor molecules in the birds' retina (Ritz et al. 2000; Mouritsen & Ritz in press). Recently, we have located putative magnetosensory molecules, the cryptochromes, in the retina and showed that the cryptochrome-containing retinal cells are highly active when migratory birds perform magnetic orientation, whereas they are comparatively inactive in non-migratory birds (Mouritsen et al. 2004a).

Since research about the magnetosensory system of migratory birds is still in its infancy, not much is known about the underlying sensory coding and network dynamics. We only recently have located a brain region that is specialized on processing light-mediated signals during the night in migrants, but not in non-migrants (Mouritsen et al. 2005). We suspect that this brain region is responsible for processing the magnetically modulated visual signals forming the basis for the avian magnetic compass.

For hearing, vision and touch, however, we can base our investigations on a much broader knowledge, focusing on sensory coding and network structures based on synaptic interactions. In recent years, substantial evidence was collected showing that the temporal structure of spike trains carries more information about sensory stimuli of several modalities than firing rates (Reviews e.g.: Victor, 1999; Grothe & Klump, 2000, Romo & Salinas, 2003). It is a generally accepted that the temporal structure of time-varying stimuli can be reflected in temporal response patterns. However, the role of spike patterns in encoding static stimuli is still under debate (Reviews e.g.: Lestienne, 2001; Mauk & Buonomano, 2004).

Since the analysis of neural coding requires a good apprehension of the system studied, the relatively simple sensory systems of invertebrates are a good starting point to look for general coding principles. For the visual system of the fly we have shown with combined experimental and modeling studies that timing of spikes with millisecond precision depends crucially on the dynamical properties of the underlying membrane potential. Hence, the filter characteristics between stimulus and neuronal response lead either to time-locking of individual spikes or to a representation of spike rates, depending on the stimulus dynamics (Kretzberg et al., 2001a, Warzecha et al., 2000). The dynamical stimulus properties in combination with the filter characteristics of the sensory system also determine whether two stimuli can be discriminated better based on graded membrane potentials or based on spike responses (Kretzberg et al., 2000b). Both of these coding strategies are used in invertebrate nervous systems like the fly visual system and the leech tactile system.

A particularly important, though yet unresolved question is, how information about sensory stimuli is distributed across members of neuronal populations and how neuronal ensembles have to act in concert to transmit and process sensory information (Review: Averbeck and Lee, 2004). The retina group in Oldenburg was among the first groups to establish multielectrode recordings of large populations of retinal ganglion cells and combine them with theoretical analysis and model simulations (Fernandez et al., 2000; Ammerm¨ller et al., 2000; Normann et al, 2001; Ferrandez, 2001). Recent findings included an increase in synchronous activity elicited by dynamical stimulus changes during fixational eye movements (Greschner et al., 2002) and distinct patterns of spike events in response to visual stimuli. A computational model reproducing experimental results was developed using physiological properties of the different cell types and interactions between them (Thiel et al., 2002), making predictions about the interconnection of network structure and sensory coding testable.

Another key aspect for understanding perception of stimulus dynamics are network structures based on synaptic connections and their tuning mechanisms. In the InterGK, extensive work was done on the role of electrical synaptic connections for network dynamics. For the vertebrate retina (Review: Wassle, 2004) it is known that electric synapses are mainly involved in lateral connections that change their properties according to stimulus conditions, leading to higher signal to noise levels in dark und higher spatial resolution in bright conditions (Review: Cook & Becker, 1995). We have contributed considerably to the knowledge about electrical connections in the retina, using molecular anatomical and physiological techniques. In collaboration with the group of Prof. Willeke (University of Bonn) transgenic mice were used to show that different retinal cell types express different connexins and thus form different hemichannels. The types of connexins expressed in a neuron determine the physiological properties of the gap junctions the cell is able to establish. Recently, our investigations were focused mainly on the implications of electrical connections in the horizontal cell network and within the rod-cone pathway for adaptation and on the characterization of electrically coupled ganglion cell networks (Bornstein et al, 2002; Pottek et al., 2003a; Feigenspan et al., 2004; Feigenspan & Weiler, 2004; Dirks et al., 2004; Hombach et al. 2004; Schubert et al. 2004, 2005, Maxeiner et al., 2005). In contrast to the retina, the impact of electrical synapses on central neuronal circuits of vertebrates is still underappreciated, even though they are often strong enough to mediate close synchronization of subthreshold and spiking activity among clusters of neurons (Review: Connors & Long 2004).

When studying the physiology it is important to know, if the physiology correlates with perception, i.e. the psychophysics of the external dynamics. In combined behavioral and physiological studies of the the magnetosensory system, we have shown that migrating birds use their eyes to detect the reference direction provided by the magnetic field and have established a psychophysical test of magetoperception (Mouritsen et al. 2004b). We have also compared the physiology and animal psychophysics of the vertebrate auditory system. For example, we have shown similarities in the response of auditory nerve fiber to temporal amplitude modulated noise and modulation thresholds as measured behaviorally in the starling (Klump & Okanoya, 1991, Gleich & Klump, 1995). The birds showed a reduced sensitivity to high modulation frequencies similar to humans. Psychoacoustical data in humans on modulation masking (Houtgast, 1989; Ewert et al., 2002), as well as physiological data in several vertebrates (Review: Langner, 1992) suggest that the auditory system performs a frequency-selective analysis of the amplitude modulation similar to the well know frequency selectivity in the audio-frequency domain.

The effect of superthresholds temporal modulations on loudness perception is investigated in several psychophysical studies. Whereas the loudness of long fluctuating signals is dominated by the loudness of intensity peaks (Zwicker and Fastl, 1999, Chapter 16), for shorter signal durations (in the range of one second) the overall intensity seems to determine loudness (Moore et al., 1999a). In addition, we showed that loudness of modulated signal depends on the spectrum (Grimm et al., 2001) and that spectral loudness is different at onset than for the stationary part of a signal (Verhey & Kollmeier, 2002). Current loudness models (Overview: Appell et al., 2001) are not able to account for these dynamic effects. Loudness was also studied in cochlear implant patients (Zeng & Shannon, 1994, Zeng, 2004) and recently models of loudness perception of complex stimuli in cochlear implant patients have been proposed (Chatterjee, 1999, McKay, 2003).
Compared to audition, little is known about the perception of vibration in humans. Almost all research on vibration perception is closely linked either to a specific application or to basic neurology, in particular focused on tactile perception of roughness. In addition to studying whole-body vibrations (Bellmann et al., 2000, 2004) we have investigated the perception of vibrations at two different parts of the human hand - fingertip and palm - and gained some basic psychophysical insight into a functional relation between physical stimulus and subjective perception (e.g. frequency-dependent thresholds, JNDs, dependency on experimental settings of the tactile boundary conditions). The input-output relations to three receptors of the somatosensosry system were identified and a first approach to a functional model was achieved (Oey & Mellert, 2004).