Limits...
Neural circuits underlying adaptation and learning in the perception of auditory space.

King AJ, Dahmen JC, Keating P, Leach ND, Nodal FR, Bajo VM - Neurosci Biobehav Rev (2011)

Bottom Line: Sound localization mechanisms are particularly plastic during development, when the monaural and binaural acoustic cues that form the basis for spatial hearing change in value as the body grows.Recent studies have shown that the mature brain retains a surprising capacity to relearn to localize sound in the presence of substantially altered auditory spatial cues.Through a combination of recording studies and methods for selectively manipulating the activity of specific neuronal populations, progress is now being made in identifying the cortical and subcortical circuits in the brain that are responsible for the dynamic coding of auditory spatial information.

View Article: PubMed Central - PubMed

Affiliation: Department of Physiology, Anatomy and Genetics, Sherrington Building, University of Oxford, Parks Road, Oxford, UK. andrew.king@dpag.ox.ac.uk

Show MeSH

Related in: MedlinePlus

Auditory spatial processing adapts to stimulus statistics. (A) Human listeners and anesthetized ferrets were presented with noise sequences in which interaural level differences (ILD) rapidly fluctuated according to a Gaussian distribution. Negative values indicate that the sound level was higher in the contralateral ear (left ear for psychophysics). (B) Changing the mean of the distribution biased the perceived laterality of a subsequent stimulus, resulting in shifts in the listeners’ psychometric functions, which plot the percentage of trials that a subject perceived the sound presented over headphones to come from the left as a function of the ILD. (C) Changing the distribution's variance altered the listeners’ spatial sensitivity, as shown by increases (low variance) or decreases (high variance) in the slopes of the psychometric functions. (D–K) The responses of neurons in the inferior colliculus changed in line with these perceptual phenomena. For each neuron and each stimulus condition, spatial response properties were characterized both in terms of the components of a linear–nonlinear model (D–G) and a more conventional rate–ILD function (H, I). Slope and response variability of all rate-ILD functions were also analyzed further to obtain a population measure of neural sensitivity, the standard separation D (J, K). The linear–nonlinear model analysis describes neural coding using a two-stage process, consisting of linear filtering (D, E) of the stimulus by the neuron, which provides an estimate of the stimulus feature that best drives it, followed by spike generation according to a nonlinear function (F, G) of the similarity of the stimulus to that feature. This analysis revealed that neurons match their stimulus preference to the stimulus distribution's mean (inset in panel D shows filters before mean-subtraction), but retain similar gain (F) across different means. This results in large shifts in rate-ILD functions (H) and allows the population to maintain the highest sensitivity near the mean of the distribution (J). Across distributions with different variances, neurons largely retained their filter shape (E), but increased their gain as the variance was reduced (G). This resulted in steeper rate-ILD functions (I) and higher neural sensitivity in a low variance context (K).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3198863&req=5

fig0005: Auditory spatial processing adapts to stimulus statistics. (A) Human listeners and anesthetized ferrets were presented with noise sequences in which interaural level differences (ILD) rapidly fluctuated according to a Gaussian distribution. Negative values indicate that the sound level was higher in the contralateral ear (left ear for psychophysics). (B) Changing the mean of the distribution biased the perceived laterality of a subsequent stimulus, resulting in shifts in the listeners’ psychometric functions, which plot the percentage of trials that a subject perceived the sound presented over headphones to come from the left as a function of the ILD. (C) Changing the distribution's variance altered the listeners’ spatial sensitivity, as shown by increases (low variance) or decreases (high variance) in the slopes of the psychometric functions. (D–K) The responses of neurons in the inferior colliculus changed in line with these perceptual phenomena. For each neuron and each stimulus condition, spatial response properties were characterized both in terms of the components of a linear–nonlinear model (D–G) and a more conventional rate–ILD function (H, I). Slope and response variability of all rate-ILD functions were also analyzed further to obtain a population measure of neural sensitivity, the standard separation D (J, K). The linear–nonlinear model analysis describes neural coding using a two-stage process, consisting of linear filtering (D, E) of the stimulus by the neuron, which provides an estimate of the stimulus feature that best drives it, followed by spike generation according to a nonlinear function (F, G) of the similarity of the stimulus to that feature. This analysis revealed that neurons match their stimulus preference to the stimulus distribution's mean (inset in panel D shows filters before mean-subtraction), but retain similar gain (F) across different means. This results in large shifts in rate-ILD functions (H) and allows the population to maintain the highest sensitivity near the mean of the distribution (J). Across distributions with different variances, neurons largely retained their filter shape (E), but increased their gain as the variance was reduced (G). This resulted in steeper rate-ILD functions (I) and higher neural sensitivity in a low variance context (K).

Mentions: Recent evidence provides some insight into how the processing of auditory space is affected by spatial input statistics (Fig. 1). By presenting human listeners over headphones with broadband noise sequences whose ILDs fluctuated rapidly according to a Gaussian distribution, and altering the mean or variance of that distribution (Fig. 1A), Dahmen et al. (2010) showed that the perception of auditory space strongly depends on the statistics of the sensory context. When the mean of the ILD distribution was changed, the perceived laterality of a subsequent stimulus was shifted away from the mean (Fig. 1B). Manipulating the variance of the stimulus distribution also affected perception, such that spatial sensitivity improved as the variance was decreased and declined when the variance was increased (Fig. 1C).


Neural circuits underlying adaptation and learning in the perception of auditory space.

King AJ, Dahmen JC, Keating P, Leach ND, Nodal FR, Bajo VM - Neurosci Biobehav Rev (2011)

Auditory spatial processing adapts to stimulus statistics. (A) Human listeners and anesthetized ferrets were presented with noise sequences in which interaural level differences (ILD) rapidly fluctuated according to a Gaussian distribution. Negative values indicate that the sound level was higher in the contralateral ear (left ear for psychophysics). (B) Changing the mean of the distribution biased the perceived laterality of a subsequent stimulus, resulting in shifts in the listeners’ psychometric functions, which plot the percentage of trials that a subject perceived the sound presented over headphones to come from the left as a function of the ILD. (C) Changing the distribution's variance altered the listeners’ spatial sensitivity, as shown by increases (low variance) or decreases (high variance) in the slopes of the psychometric functions. (D–K) The responses of neurons in the inferior colliculus changed in line with these perceptual phenomena. For each neuron and each stimulus condition, spatial response properties were characterized both in terms of the components of a linear–nonlinear model (D–G) and a more conventional rate–ILD function (H, I). Slope and response variability of all rate-ILD functions were also analyzed further to obtain a population measure of neural sensitivity, the standard separation D (J, K). The linear–nonlinear model analysis describes neural coding using a two-stage process, consisting of linear filtering (D, E) of the stimulus by the neuron, which provides an estimate of the stimulus feature that best drives it, followed by spike generation according to a nonlinear function (F, G) of the similarity of the stimulus to that feature. This analysis revealed that neurons match their stimulus preference to the stimulus distribution's mean (inset in panel D shows filters before mean-subtraction), but retain similar gain (F) across different means. This results in large shifts in rate-ILD functions (H) and allows the population to maintain the highest sensitivity near the mean of the distribution (J). Across distributions with different variances, neurons largely retained their filter shape (E), but increased their gain as the variance was reduced (G). This resulted in steeper rate-ILD functions (I) and higher neural sensitivity in a low variance context (K).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3198863&req=5

fig0005: Auditory spatial processing adapts to stimulus statistics. (A) Human listeners and anesthetized ferrets were presented with noise sequences in which interaural level differences (ILD) rapidly fluctuated according to a Gaussian distribution. Negative values indicate that the sound level was higher in the contralateral ear (left ear for psychophysics). (B) Changing the mean of the distribution biased the perceived laterality of a subsequent stimulus, resulting in shifts in the listeners’ psychometric functions, which plot the percentage of trials that a subject perceived the sound presented over headphones to come from the left as a function of the ILD. (C) Changing the distribution's variance altered the listeners’ spatial sensitivity, as shown by increases (low variance) or decreases (high variance) in the slopes of the psychometric functions. (D–K) The responses of neurons in the inferior colliculus changed in line with these perceptual phenomena. For each neuron and each stimulus condition, spatial response properties were characterized both in terms of the components of a linear–nonlinear model (D–G) and a more conventional rate–ILD function (H, I). Slope and response variability of all rate-ILD functions were also analyzed further to obtain a population measure of neural sensitivity, the standard separation D (J, K). The linear–nonlinear model analysis describes neural coding using a two-stage process, consisting of linear filtering (D, E) of the stimulus by the neuron, which provides an estimate of the stimulus feature that best drives it, followed by spike generation according to a nonlinear function (F, G) of the similarity of the stimulus to that feature. This analysis revealed that neurons match their stimulus preference to the stimulus distribution's mean (inset in panel D shows filters before mean-subtraction), but retain similar gain (F) across different means. This results in large shifts in rate-ILD functions (H) and allows the population to maintain the highest sensitivity near the mean of the distribution (J). Across distributions with different variances, neurons largely retained their filter shape (E), but increased their gain as the variance was reduced (G). This resulted in steeper rate-ILD functions (I) and higher neural sensitivity in a low variance context (K).
Mentions: Recent evidence provides some insight into how the processing of auditory space is affected by spatial input statistics (Fig. 1). By presenting human listeners over headphones with broadband noise sequences whose ILDs fluctuated rapidly according to a Gaussian distribution, and altering the mean or variance of that distribution (Fig. 1A), Dahmen et al. (2010) showed that the perception of auditory space strongly depends on the statistics of the sensory context. When the mean of the ILD distribution was changed, the perceived laterality of a subsequent stimulus was shifted away from the mean (Fig. 1B). Manipulating the variance of the stimulus distribution also affected perception, such that spatial sensitivity improved as the variance was decreased and declined when the variance was increased (Fig. 1C).

Bottom Line: Sound localization mechanisms are particularly plastic during development, when the monaural and binaural acoustic cues that form the basis for spatial hearing change in value as the body grows.Recent studies have shown that the mature brain retains a surprising capacity to relearn to localize sound in the presence of substantially altered auditory spatial cues.Through a combination of recording studies and methods for selectively manipulating the activity of specific neuronal populations, progress is now being made in identifying the cortical and subcortical circuits in the brain that are responsible for the dynamic coding of auditory spatial information.

View Article: PubMed Central - PubMed

Affiliation: Department of Physiology, Anatomy and Genetics, Sherrington Building, University of Oxford, Parks Road, Oxford, UK. andrew.king@dpag.ox.ac.uk

Show MeSH
Related in: MedlinePlus