Acoustic pattern recognition in a grasshopper: processing in the time or frequency domain?

摘要: Males of the grasshopper Chorthippus biguttulus produce songs which consist stereotyped and rhythmic iteration a sound unit (termed syllable) separated by distinct syllable pauses. Virgin females respond to this signal, similar artificial signals, with song phrases their own. In behavioural experiments response probability virgin can be measured acoustic stimuli. The stimuli consisted an amplitude modulated noise envelope was altered. We investigated several hypotheses on mechanisms conspecific recognition special emphasis question whether occurs in frequency domain or time domain. (1) Females Ch. required only first five Fourier components function (corresponding 50 Hz for fundamental 10 Hz) detect syllable/pause structure. addition, they detected small gaps within syllables if signal contained at least ca. 15 150 Hz). Further showed that correct phase information is necessary recognition, indicating pattern not achieved merely basis band pass filtering. (2) A cross correlation between assumed internal template yields inconsistent predictions probabilities. (3) recognizer system probably works domain, possibly direct comparison adjacent pause durations. It yet clear duration evaluated respect preceding succeeding pause. emphasize neural does examine its similarity template, but it also takes into account features indicate incorrect signal. This may general feature neuronal systems have been shaped natural selection.