| Abstract No.: | C-D3130 |
| Country: | Canada |
| | |
| Title: | MECHANISMS OF TEMPORAL FILTERING IN A SIMPLE AUDITORY SYSTEM. |
| | |
| Authors/Affiliations: | 1 Patrick Sabourin*; 1 Gerald Pollack;
1 McGill University, Montreal, QC, Canada
|
| | |
| Content: | Objectives. The information content of sensory signals is often coded into their temporal structure, and an important issue in neuroscience is understanding how this information is processed in the nervous system. The auditory system of crickets is specialized for two classes of acoustic signals: cricket songs and bat (predator) echolocation sounds. These differ in sound frequency (cricket songs: ~5 kHz; bat sounds: > 20 kHz) and in temporal structure (cricket songs: stereotypic, with amplitude-modulation (AM) rates < 40 Hz; bat sounds: variable, with AM rates up to 100 Hz). We showed previously that an identified auditory neuron, ON1, accurately encodes the temporal pattern of 5 kHz stimuli only for AM rates up to 40 Hz, matching the structure of cricket songs; but the same neuron accurately represents stimulus timing of ultrasonic (30 kHz) stimuli for AM rates up to 100Hz. Here we address the mechanisms underlying the dual temporal-filtering characteristics of this neuron.
Materials and methods. We used standard neurophysiological recording techniques, and information theory.
Results. In response to intracellular current injection, variations in membrane potential can follow stimulus rates up to 100 Hz, suggesting that intrinsic membrane properties play little if any role in the neuron's filtering of acoustic stimuli. Variations in membrane potential that underlie spiking show the same, sound-frequency-specific, filtering properties as the neuron's spiking responses. Together, these findings suggest that the latter properties arise at the level of ON1's inputs.
ON1 receives direct input from ca. 25 low-frequency-tuned receptors, and ca. 10 high-frequency-tuned receptors. We recorded spike trains from receptor neurons, and characterized their temporal characteristics using information theory. High-frequency-tuned receptors more accurately represent variations in stimulus amplitude than low-frequency receptors, despite the fact that, for AM rates occurring in cricket songs, ON1 more accurately represents the timing of low- than of high-sound-frequency stimuli. This discrepancy, together with the relatively large population size of low-frequency receptors, suggests that ON1 integrates information across a population of low-frequency-tuned afferents. This could occur only if the information carried by individual receptor neurons were non-redundant. We measured mutual information between pairs of receptor spike trains, and found that on average this is low for low-frequency receptors, and high for high-frequency receptors.
We modeled the effects of integration across afferent populations by superimposing the spike trains of varying numbers of receptors and calculating the overall information that they encode. For low rates of AM, the increase in information content resulting from combining receptor spike trains was large for low-frequency receptor neurons, and was roughly proportional to the number of superimposed spike trains. For high frequency receptors, the increase in information was distributed uniformly across AM rates, and saturated once population size reached ca. 5 receptor neurons.
Conclusion. These results suggest that the ability of ON1 to represent, in a sound-frequency-specific manner, behaviorally relevant patterns of AM has its basis in the temporal organization of the spike trains of |
| | |
| Back |
|
