Setting complex tasks to single units in the avian auditory forebrain. II. Do we really need natural stimuli to describe neuronal response characteristics?

The response characteristic of auditory forebrain neurons in the European starling was established both with artificial stimuli (AS) and a conspecific territorial song as a natural stimulus (NS1). Applying experimenter-centred statistical methods for response detection and for scaling response strength, and spike-triggered analyses for the delimitation of the key sound parameters (spectrotemporal receptive field STRF, Aertsen et al. 1980) the study aimed at disclosing differences in the processing of the two stimulus classes, AS and NS. With the STRF as reference, we find congruence (1) in the best frequency with those determined under sweep and bandpass noise stimulation, (2) in response latency, and (3) in response-intensity dependence, further similarity in the overall frequency characteristic. Partitioning the song into 42 acoustically defined segments allowed to further delimit the response criteria under natural stimulation. They are easily understood from the AS response characteristics: (1) In the neuronal sample as a whole, long segments are more effective than short and, among the short, loud segments are more effective than faint; (2) Units showing their best excitatory response to AS in a certain frequency band are most probably excited by segments with a high proportion of their power concentrated upon or near this band; (3) Units with a slow (build-up) AS response react to a lower number of song segments than those dynamically following AS transients. Our data give no hint towards adaptive, feature detection properties of single neurons in field L. Instead, these neurons appear to base their response solely on the short-time spectrotemporal structure of the stimulus, irrespective of its natural or artificial origin.