Limits...
The opponent channel population code of sound location is an efficient representation of natural binaural sounds.

Młynarski W - PLoS Comput. Biol. (2015)

Bottom Line: Obtained tuning curves match well tuning characteristics of neurons in the mammalian auditory cortex.This study connects neuronal coding of the auditory space with natural stimulus statistics and generates new experimental predictions.Moreover, results presented here suggest that cortical regions with seemingly different functions may implement the same computational strategy-efficient coding.

View Article: PubMed Central - PubMed

Affiliation: Max-Planck Institute for Mathematics in the Sciences, Leipzig, Germany. wiktor.mlynarski@gmail.com

ABSTRACT
In mammalian auditory cortex, sound source position is represented by a population of broadly tuned neurons whose firing is modulated by sounds located at all positions surrounding the animal. Peaks of their tuning curves are concentrated at lateral position, while their slopes are steepest at the interaural midline, allowing for the maximum localization accuracy in that area. These experimental observations contradict initial assumptions that the auditory space is represented as a topographic cortical map. It has been suggested that a "panoramic" code has evolved to match specific demands of the sound localization task. This work provides evidence suggesting that properties of spatial auditory neurons identified experimentally follow from a general design principle- learning a sparse, efficient representation of natural stimuli. Natural binaural sounds were recorded and served as input to a hierarchical sparse-coding model. In the first layer, left and right ear sounds were separately encoded by a population of complex-valued basis functions which separated phase and amplitude. Both parameters are known to carry information relevant for spatial hearing. Monaural input converged in the second layer, which learned a joint representation of amplitude and interaural phase difference. Spatial selectivity of each second-layer unit was measured by exposing the model to natural sound sources recorded at different positions. Obtained tuning curves match well tuning characteristics of neurons in the mammalian auditory cortex. This study connects neuronal coding of the auditory space with natural stimulus statistics and generates new experimental predictions. Moreover, results presented here suggest that cortical regions with seemingly different functions may implement the same computational strategy-efficient coding.

No MeSH data available.


Related in: MedlinePlus

First layer basis.A) Representative real (black) and imaginary (gray) vectors. B) Isoprobability contours of Wigner-Ville distributions associated with each real vector. Time—frequency plane is tiled uniformly with a weak overlap. Gray-filled oval corresponds to the framed basis function on panel A.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4440638&req=5

pcbi.1004294.g002: First layer basis.A) Representative real (black) and imaginary (gray) vectors. B) Isoprobability contours of Wigner-Ville distributions associated with each real vector. Time—frequency plane is tiled uniformly with a weak overlap. Gray-filled oval corresponds to the framed basis function on panel A.

Mentions: To overcome these problems, a different approach was taken here. The first-layer representation was created in two steps. Firstly a real-valued sparse code was trained (see Methods). Learned basis functions were well localized in time or frequency and tiled the time-frequency plane in a uniform and non-overlapping manner (Fig 2B). They were taken as real vectors Aℜ of complex basis functions A. In the second step, imaginary parts were created by performing the Hilbert transform of real vectors. The Hilbert transform of a time varying signal y(t) is defined as follows:H(y(t))=1πp.v.∫-∞∞y(τ)t-τdτ(6)Where p.v. stands for Cauchy principal value. In such a way every real vector was paired with its Hilbert transform i.e. a vector which complex Fourier’s coefficients are all shifted by in phase. The obtained dictionary is adapted to the stimulus ensemble, hence providing a non-redundant data representation, yet makes phase clearly interpretable as a temporal displacement.


The opponent channel population code of sound location is an efficient representation of natural binaural sounds.

Młynarski W - PLoS Comput. Biol. (2015)

First layer basis.A) Representative real (black) and imaginary (gray) vectors. B) Isoprobability contours of Wigner-Ville distributions associated with each real vector. Time—frequency plane is tiled uniformly with a weak overlap. Gray-filled oval corresponds to the framed basis function on panel A.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4440638&req=5

pcbi.1004294.g002: First layer basis.A) Representative real (black) and imaginary (gray) vectors. B) Isoprobability contours of Wigner-Ville distributions associated with each real vector. Time—frequency plane is tiled uniformly with a weak overlap. Gray-filled oval corresponds to the framed basis function on panel A.
Mentions: To overcome these problems, a different approach was taken here. The first-layer representation was created in two steps. Firstly a real-valued sparse code was trained (see Methods). Learned basis functions were well localized in time or frequency and tiled the time-frequency plane in a uniform and non-overlapping manner (Fig 2B). They were taken as real vectors Aℜ of complex basis functions A. In the second step, imaginary parts were created by performing the Hilbert transform of real vectors. The Hilbert transform of a time varying signal y(t) is defined as follows:H(y(t))=1πp.v.∫-∞∞y(τ)t-τdτ(6)Where p.v. stands for Cauchy principal value. In such a way every real vector was paired with its Hilbert transform i.e. a vector which complex Fourier’s coefficients are all shifted by in phase. The obtained dictionary is adapted to the stimulus ensemble, hence providing a non-redundant data representation, yet makes phase clearly interpretable as a temporal displacement.

Bottom Line: Obtained tuning curves match well tuning characteristics of neurons in the mammalian auditory cortex.This study connects neuronal coding of the auditory space with natural stimulus statistics and generates new experimental predictions.Moreover, results presented here suggest that cortical regions with seemingly different functions may implement the same computational strategy-efficient coding.

View Article: PubMed Central - PubMed

Affiliation: Max-Planck Institute for Mathematics in the Sciences, Leipzig, Germany. wiktor.mlynarski@gmail.com

ABSTRACT
In mammalian auditory cortex, sound source position is represented by a population of broadly tuned neurons whose firing is modulated by sounds located at all positions surrounding the animal. Peaks of their tuning curves are concentrated at lateral position, while their slopes are steepest at the interaural midline, allowing for the maximum localization accuracy in that area. These experimental observations contradict initial assumptions that the auditory space is represented as a topographic cortical map. It has been suggested that a "panoramic" code has evolved to match specific demands of the sound localization task. This work provides evidence suggesting that properties of spatial auditory neurons identified experimentally follow from a general design principle- learning a sparse, efficient representation of natural stimuli. Natural binaural sounds were recorded and served as input to a hierarchical sparse-coding model. In the first layer, left and right ear sounds were separately encoded by a population of complex-valued basis functions which separated phase and amplitude. Both parameters are known to carry information relevant for spatial hearing. Monaural input converged in the second layer, which learned a joint representation of amplitude and interaural phase difference. Spatial selectivity of each second-layer unit was measured by exposing the model to natural sound sources recorded at different positions. Obtained tuning curves match well tuning characteristics of neurons in the mammalian auditory cortex. This study connects neuronal coding of the auditory space with natural stimulus statistics and generates new experimental predictions. Moreover, results presented here suggest that cortical regions with seemingly different functions may implement the same computational strategy-efficient coding.

No MeSH data available.


Related in: MedlinePlus