Limits...
A neural mechanism for time-window separation resolves ambiguity of adaptive coding.

Hildebrandt KJ, Ronacher B, Hennig RM, Benda J - PLoS Biol. (2015)

Bottom Line: Experimental data on the trade-off between benefits and loss through adaptation is scarce and very few mechanisms have been proposed to resolve it.We demonstrate how focusing the response of localization neurons to the onset of relevant signals separates processing of localization and pattern information temporally.In this way, the ambiguity of adaptive coding can be circumvented and both absolute and relative levels can be processed using the same set of peripheral neurons.

View Article: PubMed Central - PubMed

Affiliation: Cluster of Excellence "Hearing4all", Department for Neuroscience, University of Oldenburg, Oldenburg, Germany.

ABSTRACT
The senses of animals are confronted with changing environments and different contexts. Neural adaptation is one important tool to adjust sensitivity to varying intensity ranges. For instance, in a quiet night outdoors, our hearing is more sensitive than when we are confronted with the plurality of sounds in a large city during the day. However, adaptation also removes available information on absolute sound levels and may thus cause ambiguity. Experimental data on the trade-off between benefits and loss through adaptation is scarce and very few mechanisms have been proposed to resolve it. We present an example where adaptation is beneficial for one task--namely, the reliable encoding of the pattern of an acoustic signal-but detrimental for another--the localization of the same acoustic stimulus. With a combination of neurophysiological data, modeling, and behavioral tests, we show that adaptation in the periphery of the auditory pathway of grasshoppers enables intensity-invariant coding of amplitude modulations, but at the same time, degrades information available for sound localization. We demonstrate how focusing the response of localization neurons to the onset of relevant signals separates processing of localization and pattern information temporally. In this way, the ambiguity of adaptive coding can be circumvented and both absolute and relative levels can be processed using the same set of peripheral neurons.

Show MeSH

Related in: MedlinePlus

Emergence of level invariance along the pathway.A: In grasshoppers, the ears are located on each side of the first abdominal segment. The first step of neural processing takes place in the metathoracic ganglion. B: Metathoracic auditory pathway of grasshoppers: sound is transduced in receptor neurons in the ear and information is carried to the metathoracic ganglion, where receptors synapse on local neurons (both gray). These connect to ascending neurons, which in turn transfer auditory information about sound pattern (red) and direction (blue) separately up to the brain. C: Mean response of six local neurons (TN1) to an amplitude-modulated (AM) sound (bottom panel). The upper panel shows the spike frequency elicited by the sound played back at two different levels (black and red). The middle panel shows the first 130 ms of the response in more detail; the shaded areas depict standard deviation. The lower panel is the average difference between the responses in the middle panel ([response to louder sound minus response to softer sound]/response to louder sound). D: Example of a measurement of a level-response curve adapted to a given background level in a TN1 neuron. Gray boxes indicate the examples in panels E and F. E: Onset responses to level steps relative to a background level (numbers at the bottom) are independent of different background levels (indicated by line color); the neurons were adapted to a mean level as shown in D. F: Adaptation of level-response curves of a TN1 neuron to different background levels of sound. The black line is the response curve when tested in silence (error bars: standard error of the mean [SEM]); the colored curves represent adapted response curves of the same cell after adaptation to different background levels (dotted vertical lines). G: Shift of the parameterized level-response curves as a function of the background sound for all recorded TN1 (left panel) and receptor cells (right panel). The dashed black line is the prediction of complete compensation of the background level by adaptation, the red line a fit of a straight line to the data. See S1 Data for data underlying panels C–F.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4356587&req=5

pbio.1002096.g001: Emergence of level invariance along the pathway.A: In grasshoppers, the ears are located on each side of the first abdominal segment. The first step of neural processing takes place in the metathoracic ganglion. B: Metathoracic auditory pathway of grasshoppers: sound is transduced in receptor neurons in the ear and information is carried to the metathoracic ganglion, where receptors synapse on local neurons (both gray). These connect to ascending neurons, which in turn transfer auditory information about sound pattern (red) and direction (blue) separately up to the brain. C: Mean response of six local neurons (TN1) to an amplitude-modulated (AM) sound (bottom panel). The upper panel shows the spike frequency elicited by the sound played back at two different levels (black and red). The middle panel shows the first 130 ms of the response in more detail; the shaded areas depict standard deviation. The lower panel is the average difference between the responses in the middle panel ([response to louder sound minus response to softer sound]/response to louder sound). D: Example of a measurement of a level-response curve adapted to a given background level in a TN1 neuron. Gray boxes indicate the examples in panels E and F. E: Onset responses to level steps relative to a background level (numbers at the bottom) are independent of different background levels (indicated by line color); the neurons were adapted to a mean level as shown in D. F: Adaptation of level-response curves of a TN1 neuron to different background levels of sound. The black line is the response curve when tested in silence (error bars: standard error of the mean [SEM]); the colored curves represent adapted response curves of the same cell after adaptation to different background levels (dotted vertical lines). G: Shift of the parameterized level-response curves as a function of the background sound for all recorded TN1 (left panel) and receptor cells (right panel). The dashed black line is the prediction of complete compensation of the background level by adaptation, the red line a fit of a straight line to the data. See S1 Data for data underlying panels C–F.

Mentions: The ears of grasshoppers are located laterally in the first abdominal segment (Fig. 1A). Receptor neurons transduce sound and encode information about the stimulus pattern in action potential frequency [23,24]. The receptor axons enter the metathoracic ganglion, where they synapse on a population of local neurons (Fig. 1B, [25]). In this ganglion, information about the pattern and the directionality of the sound is separated into two channels, represented by different ascending neurons [21,26]. Both channels make use of the same peripheral input from both ears combined in central, ascending neurons. Receptors do not synapse directly onto ascending neurons, but on local neurons only (Fig. 1B). In the case of pattern coding, summation over the peripheral responses from both sides increases the signal-to-noise ratio but leads to a loss of directional information. For directionality, the system evaluates ILD, whereas inter-aural time differences are much too small to be evaluated in grasshoppers (differences of 5–6 μs at most; [27]). In order to evaluate ILDs, the grasshopper ear works as a pressure gradient receiver [28]. In addition, the differences between the peripheral inputs from the two ears are enhanced by contralateral inhibition, emphasizing the directional tuning. The ascending neuron AN2 in locusts is thought to code for the direction of the stimulus [25].


A neural mechanism for time-window separation resolves ambiguity of adaptive coding.

Hildebrandt KJ, Ronacher B, Hennig RM, Benda J - PLoS Biol. (2015)

Emergence of level invariance along the pathway.A: In grasshoppers, the ears are located on each side of the first abdominal segment. The first step of neural processing takes place in the metathoracic ganglion. B: Metathoracic auditory pathway of grasshoppers: sound is transduced in receptor neurons in the ear and information is carried to the metathoracic ganglion, where receptors synapse on local neurons (both gray). These connect to ascending neurons, which in turn transfer auditory information about sound pattern (red) and direction (blue) separately up to the brain. C: Mean response of six local neurons (TN1) to an amplitude-modulated (AM) sound (bottom panel). The upper panel shows the spike frequency elicited by the sound played back at two different levels (black and red). The middle panel shows the first 130 ms of the response in more detail; the shaded areas depict standard deviation. The lower panel is the average difference between the responses in the middle panel ([response to louder sound minus response to softer sound]/response to louder sound). D: Example of a measurement of a level-response curve adapted to a given background level in a TN1 neuron. Gray boxes indicate the examples in panels E and F. E: Onset responses to level steps relative to a background level (numbers at the bottom) are independent of different background levels (indicated by line color); the neurons were adapted to a mean level as shown in D. F: Adaptation of level-response curves of a TN1 neuron to different background levels of sound. The black line is the response curve when tested in silence (error bars: standard error of the mean [SEM]); the colored curves represent adapted response curves of the same cell after adaptation to different background levels (dotted vertical lines). G: Shift of the parameterized level-response curves as a function of the background sound for all recorded TN1 (left panel) and receptor cells (right panel). The dashed black line is the prediction of complete compensation of the background level by adaptation, the red line a fit of a straight line to the data. See S1 Data for data underlying panels C–F.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4356587&req=5

pbio.1002096.g001: Emergence of level invariance along the pathway.A: In grasshoppers, the ears are located on each side of the first abdominal segment. The first step of neural processing takes place in the metathoracic ganglion. B: Metathoracic auditory pathway of grasshoppers: sound is transduced in receptor neurons in the ear and information is carried to the metathoracic ganglion, where receptors synapse on local neurons (both gray). These connect to ascending neurons, which in turn transfer auditory information about sound pattern (red) and direction (blue) separately up to the brain. C: Mean response of six local neurons (TN1) to an amplitude-modulated (AM) sound (bottom panel). The upper panel shows the spike frequency elicited by the sound played back at two different levels (black and red). The middle panel shows the first 130 ms of the response in more detail; the shaded areas depict standard deviation. The lower panel is the average difference between the responses in the middle panel ([response to louder sound minus response to softer sound]/response to louder sound). D: Example of a measurement of a level-response curve adapted to a given background level in a TN1 neuron. Gray boxes indicate the examples in panels E and F. E: Onset responses to level steps relative to a background level (numbers at the bottom) are independent of different background levels (indicated by line color); the neurons were adapted to a mean level as shown in D. F: Adaptation of level-response curves of a TN1 neuron to different background levels of sound. The black line is the response curve when tested in silence (error bars: standard error of the mean [SEM]); the colored curves represent adapted response curves of the same cell after adaptation to different background levels (dotted vertical lines). G: Shift of the parameterized level-response curves as a function of the background sound for all recorded TN1 (left panel) and receptor cells (right panel). The dashed black line is the prediction of complete compensation of the background level by adaptation, the red line a fit of a straight line to the data. See S1 Data for data underlying panels C–F.
Mentions: The ears of grasshoppers are located laterally in the first abdominal segment (Fig. 1A). Receptor neurons transduce sound and encode information about the stimulus pattern in action potential frequency [23,24]. The receptor axons enter the metathoracic ganglion, where they synapse on a population of local neurons (Fig. 1B, [25]). In this ganglion, information about the pattern and the directionality of the sound is separated into two channels, represented by different ascending neurons [21,26]. Both channels make use of the same peripheral input from both ears combined in central, ascending neurons. Receptors do not synapse directly onto ascending neurons, but on local neurons only (Fig. 1B). In the case of pattern coding, summation over the peripheral responses from both sides increases the signal-to-noise ratio but leads to a loss of directional information. For directionality, the system evaluates ILD, whereas inter-aural time differences are much too small to be evaluated in grasshoppers (differences of 5–6 μs at most; [27]). In order to evaluate ILDs, the grasshopper ear works as a pressure gradient receiver [28]. In addition, the differences between the peripheral inputs from the two ears are enhanced by contralateral inhibition, emphasizing the directional tuning. The ascending neuron AN2 in locusts is thought to code for the direction of the stimulus [25].

Bottom Line: Experimental data on the trade-off between benefits and loss through adaptation is scarce and very few mechanisms have been proposed to resolve it.We demonstrate how focusing the response of localization neurons to the onset of relevant signals separates processing of localization and pattern information temporally.In this way, the ambiguity of adaptive coding can be circumvented and both absolute and relative levels can be processed using the same set of peripheral neurons.

View Article: PubMed Central - PubMed

Affiliation: Cluster of Excellence "Hearing4all", Department for Neuroscience, University of Oldenburg, Oldenburg, Germany.

ABSTRACT
The senses of animals are confronted with changing environments and different contexts. Neural adaptation is one important tool to adjust sensitivity to varying intensity ranges. For instance, in a quiet night outdoors, our hearing is more sensitive than when we are confronted with the plurality of sounds in a large city during the day. However, adaptation also removes available information on absolute sound levels and may thus cause ambiguity. Experimental data on the trade-off between benefits and loss through adaptation is scarce and very few mechanisms have been proposed to resolve it. We present an example where adaptation is beneficial for one task--namely, the reliable encoding of the pattern of an acoustic signal-but detrimental for another--the localization of the same acoustic stimulus. With a combination of neurophysiological data, modeling, and behavioral tests, we show that adaptation in the periphery of the auditory pathway of grasshoppers enables intensity-invariant coding of amplitude modulations, but at the same time, degrades information available for sound localization. We demonstrate how focusing the response of localization neurons to the onset of relevant signals separates processing of localization and pattern information temporally. In this way, the ambiguity of adaptive coding can be circumvented and both absolute and relative levels can be processed using the same set of peripheral neurons.

Show MeSH
Related in: MedlinePlus