Art - Rose (D)

519 Biologically relevant information in acoustic signals often is present in its temporal structure. For example, the temporal relationships of particular acoustic elements are important for identifying human speech signals1, the proximity of a target in bat sonar2, songs in birds3, and calls in anuran amphibians4 (frogs and toads) and crickets5. In the sensory periphery, these signals are represented by unique spatiotemporal patterns of activity. It is still unclear to what extent these spatiotemporal patterns are read by higher-order ‘detectors’ of temporal features6. Several decades of work have provided substantial evidence for temporal feature detectors. Neurons have been recorded that respond best when tonal or frequency-modulated stimuli are of a particular duration7–9. ‘Combination-sensitive’ neurons have been found that respond best when at least two acoustic elements are presented in the appropriate temporal order and spacing10–17. In these cases, the response magnitude to the temporally appropriate combination of acoustic elements is much greater than the sum of their responses to each element presented alone. Ultimately, however, recognition and discrimination of biologically important sounds might entail integration over many acoustic elements, and over hundreds of milliseconds or seconds. In this scenario, each pair of elements in the appropriate temporal relationship would activate combination-sensitive neurons, which then provide inputs to a long-term integrator. This stage might represent the top of a neuronal hierarchy, where the number of consecutive correct ‘bits’ of information would be evaluated. Such integration processes would complete the formation of selective filters for biologically relevant signals. To investigate such long-term integration processes, we have conducted neurophysiological studies of the anuran auditory system. Because anurans are able to discriminate between calls that differ primarily in temporal pulse density18,19, it is likely that they use temporal integration. We tested this hypothesis by making recordings from single units in the torus semicircularis (auditory midbrain) of two species of frogs, R. pipiens and H. regilla. Rana pipiens has been used in many of the previous studies of temporal processing in the auditory system; H. regilla was studied because these frogs are able to discriminate behaviorally between calls that differ almost exclusively in pulse repetition rate. The torus is a logical place to look for neurons that integrate information over multiple pulses because neurons in this area respond selectively to particular rates of amplitude modulation (AM)20,21, and the distribution of ‘best rates of AM’ is species specific22,23. Here we show that long-term integration occurs in the auditory system of anurans. As a result, neurons show strong selectivity for temporal patterns of acoustic signals that represent particular call types.