Limits...
Forward masking estimated by signal detection theory analysis of neuronal responses in primary auditory cortex.

Alves-Pinto A, Baudoux S, Palmer AR, Sumner CJ - J. Assoc. Res. Otolaryngol. (2010)

Bottom Line: This is reminiscent of the reduction in neuronal responses to a sound following prior stimulation.However, although methodological differences make comparisons difficult, the threshold shifts in cortical neurons were, in contrast to subcortical nuclei, actually larger than those observed psychophysically.Masking was largely attributable to a reduction in the responses to the probe, rather than either a persistence of the masker responses or an increase in the variability of probe responses.

View Article: PubMed Central - PubMed

Affiliation: MRC Institute of Hearing Research, Science Road, University Park, Nottingham, Nottinghamshire, UK. ana@ihr.mrc.ac.uk

ABSTRACT
Psychophysical forward masking is an increase in threshold of detection of a sound (probe) when it is preceded by another sound (masker). This is reminiscent of the reduction in neuronal responses to a sound following prior stimulation. Studies in the auditory nerve and cochlear nucleus using signal detection theory techniques to derive neuronal thresholds showed that in centrally projecting neurons, increases in masked thresholds were significantly smaller than the changes measured psychophysically. Larger threshold shifts have been reported in the inferior colliculus of awake marmoset. The present study investigated the magnitude of forward masking in primary auditory cortical neurons of anaesthetised guinea-pigs. Responses of cortical neurons to unmasked and forward masked tones were measured and probe detection thresholds estimated using signal detection theory methods. Threshold shifts were larger than in the auditory nerve, cochlear nucleus and inferior colliculus. The larger threshold shifts suggest that central, and probably cortical, processes contribute to forward masking. However, although methodological differences make comparisons difficult, the threshold shifts in cortical neurons were, in contrast to subcortical nuclei, actually larger than those observed psychophysically. Masking was largely attributable to a reduction in the responses to the probe, rather than either a persistence of the masker responses or an increase in the variability of probe responses.

Show MeSH
The SDT based method for analysing neural responses. A Deriving the percentage of correct detections from trial-by-trial spike count comparisons. Left panel shows a set of trials in which the probe was presented. Right panel shows a set of trials in which no probe was presented. Greater than symbol indicates a trial in which there were more spikes in the probe condition (correct detection); less than symbol indicates a trial in which there were more spikes in the no-probe condition (so an incorrect decision is made); question mark indicates a trial in which spike counts were equal, so a guess was made. B A ‘neurometric function’ describes the percentage of correct responses as a function of probe level, from which a threshold (here we use a 60% criterion—see ‘Results’) is derived. A shift in the neurometric function on addition of a masker produces a shift in the threshold (masking). C and D The SDT method described in terms of spike count distributions. CLeft panel shows a set of spike count distributions from within the indicated analysis window when there is a probe (coloured lines) and when there is not (dashed line). Right panels show example PSTHs that might be associated with these distributions: different colours indicate either a change in the level of the probe or the masker. D Spike count distributions in the no-probe condition (coloured lines) can also be dependent on the masker condition (here, the dashed line shows a potential spike count distribution for the probe condition). An increase in the response to the masker within the analysis window will reduce the percentage correct.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2914239&req=5

Fig1: The SDT based method for analysing neural responses. A Deriving the percentage of correct detections from trial-by-trial spike count comparisons. Left panel shows a set of trials in which the probe was presented. Right panel shows a set of trials in which no probe was presented. Greater than symbol indicates a trial in which there were more spikes in the probe condition (correct detection); less than symbol indicates a trial in which there were more spikes in the no-probe condition (so an incorrect decision is made); question mark indicates a trial in which spike counts were equal, so a guess was made. B A ‘neurometric function’ describes the percentage of correct responses as a function of probe level, from which a threshold (here we use a 60% criterion—see ‘Results’) is derived. A shift in the neurometric function on addition of a masker produces a shift in the threshold (masking). C and D The SDT method described in terms of spike count distributions. CLeft panel shows a set of spike count distributions from within the indicated analysis window when there is a probe (coloured lines) and when there is not (dashed line). Right panels show example PSTHs that might be associated with these distributions: different colours indicate either a change in the level of the probe or the masker. D Spike count distributions in the no-probe condition (coloured lines) can also be dependent on the masker condition (here, the dashed line shows a potential spike count distribution for the probe condition). An increase in the response to the masker within the analysis window will reduce the percentage correct.

Mentions: In a psychophysical forward-masking experiment each trial typically consists of two intervals: one in which only the masker is presented and another in which the masker is followed by the probe. In each trial, the participant has to indicate which of the two intervals contained the probe. The probe level is varied between trials and the level at which the probe is correctly detected on a certain proportion of the trials is determined. Forward masked thresholds, in our study, were derived using a method similar to that described by Britten et al. (1992), and is illustrated in Figure 1. This method is analogous to the two-alternative forced choice task used in psychophysics. For each cortical neuron the response to the probe was determined as the number of spikes that occurred within the time window of presentation of the probe (Fig. 1A). This time window is of the same duration as the probe stimulus and was set to start according to the latency of the unit, assessed by a post-stimulus time histogram of the response to both masker and probe. The spike count evoked by the probe was compared, for each stimulus presentation, with that measured within the equivalent temporal window when no probe was presented. Detection was considered to occur when the probe condition elicited more spikes than the no-probe condition. If responses were equal then a guess was made. Responses to the probe and no-probe conditions were paired and compared across all the 50 repetitions of each condition, and the number of ‘correct responses’ counted. The probability that a neuron responded in such a way as to allow correct detection of the probe, was calculated as the proportion of pairs for which the response to the probe was higher than the response to the no probe condition. This procedure was repeated for the different probe levels to derive the neurometric function (Fig. 1B; probability of correct responses as a function of probe level) for each unit, for each masker level and masker–probe time interval. The resulting neurometric function is equivalent to a psychometric function generated using a two-alternative forced choice procedure, and constitutes a prediction of performance based on the spike counts from a single unit. The threshold of detection of the probe was finally estimated from the neurometric function as the probe level for which a correct response occurred on at least 60% of the comparisons (horizontal dashed line in Fig. 1B).FIG. 1.


Forward masking estimated by signal detection theory analysis of neuronal responses in primary auditory cortex.

Alves-Pinto A, Baudoux S, Palmer AR, Sumner CJ - J. Assoc. Res. Otolaryngol. (2010)

The SDT based method for analysing neural responses. A Deriving the percentage of correct detections from trial-by-trial spike count comparisons. Left panel shows a set of trials in which the probe was presented. Right panel shows a set of trials in which no probe was presented. Greater than symbol indicates a trial in which there were more spikes in the probe condition (correct detection); less than symbol indicates a trial in which there were more spikes in the no-probe condition (so an incorrect decision is made); question mark indicates a trial in which spike counts were equal, so a guess was made. B A ‘neurometric function’ describes the percentage of correct responses as a function of probe level, from which a threshold (here we use a 60% criterion—see ‘Results’) is derived. A shift in the neurometric function on addition of a masker produces a shift in the threshold (masking). C and D The SDT method described in terms of spike count distributions. CLeft panel shows a set of spike count distributions from within the indicated analysis window when there is a probe (coloured lines) and when there is not (dashed line). Right panels show example PSTHs that might be associated with these distributions: different colours indicate either a change in the level of the probe or the masker. D Spike count distributions in the no-probe condition (coloured lines) can also be dependent on the masker condition (here, the dashed line shows a potential spike count distribution for the probe condition). An increase in the response to the masker within the analysis window will reduce the percentage correct.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2914239&req=5

Fig1: The SDT based method for analysing neural responses. A Deriving the percentage of correct detections from trial-by-trial spike count comparisons. Left panel shows a set of trials in which the probe was presented. Right panel shows a set of trials in which no probe was presented. Greater than symbol indicates a trial in which there were more spikes in the probe condition (correct detection); less than symbol indicates a trial in which there were more spikes in the no-probe condition (so an incorrect decision is made); question mark indicates a trial in which spike counts were equal, so a guess was made. B A ‘neurometric function’ describes the percentage of correct responses as a function of probe level, from which a threshold (here we use a 60% criterion—see ‘Results’) is derived. A shift in the neurometric function on addition of a masker produces a shift in the threshold (masking). C and D The SDT method described in terms of spike count distributions. CLeft panel shows a set of spike count distributions from within the indicated analysis window when there is a probe (coloured lines) and when there is not (dashed line). Right panels show example PSTHs that might be associated with these distributions: different colours indicate either a change in the level of the probe or the masker. D Spike count distributions in the no-probe condition (coloured lines) can also be dependent on the masker condition (here, the dashed line shows a potential spike count distribution for the probe condition). An increase in the response to the masker within the analysis window will reduce the percentage correct.
Mentions: In a psychophysical forward-masking experiment each trial typically consists of two intervals: one in which only the masker is presented and another in which the masker is followed by the probe. In each trial, the participant has to indicate which of the two intervals contained the probe. The probe level is varied between trials and the level at which the probe is correctly detected on a certain proportion of the trials is determined. Forward masked thresholds, in our study, were derived using a method similar to that described by Britten et al. (1992), and is illustrated in Figure 1. This method is analogous to the two-alternative forced choice task used in psychophysics. For each cortical neuron the response to the probe was determined as the number of spikes that occurred within the time window of presentation of the probe (Fig. 1A). This time window is of the same duration as the probe stimulus and was set to start according to the latency of the unit, assessed by a post-stimulus time histogram of the response to both masker and probe. The spike count evoked by the probe was compared, for each stimulus presentation, with that measured within the equivalent temporal window when no probe was presented. Detection was considered to occur when the probe condition elicited more spikes than the no-probe condition. If responses were equal then a guess was made. Responses to the probe and no-probe conditions were paired and compared across all the 50 repetitions of each condition, and the number of ‘correct responses’ counted. The probability that a neuron responded in such a way as to allow correct detection of the probe, was calculated as the proportion of pairs for which the response to the probe was higher than the response to the no probe condition. This procedure was repeated for the different probe levels to derive the neurometric function (Fig. 1B; probability of correct responses as a function of probe level) for each unit, for each masker level and masker–probe time interval. The resulting neurometric function is equivalent to a psychometric function generated using a two-alternative forced choice procedure, and constitutes a prediction of performance based on the spike counts from a single unit. The threshold of detection of the probe was finally estimated from the neurometric function as the probe level for which a correct response occurred on at least 60% of the comparisons (horizontal dashed line in Fig. 1B).FIG. 1.

Bottom Line: This is reminiscent of the reduction in neuronal responses to a sound following prior stimulation.However, although methodological differences make comparisons difficult, the threshold shifts in cortical neurons were, in contrast to subcortical nuclei, actually larger than those observed psychophysically.Masking was largely attributable to a reduction in the responses to the probe, rather than either a persistence of the masker responses or an increase in the variability of probe responses.

View Article: PubMed Central - PubMed

Affiliation: MRC Institute of Hearing Research, Science Road, University Park, Nottingham, Nottinghamshire, UK. ana@ihr.mrc.ac.uk

ABSTRACT
Psychophysical forward masking is an increase in threshold of detection of a sound (probe) when it is preceded by another sound (masker). This is reminiscent of the reduction in neuronal responses to a sound following prior stimulation. Studies in the auditory nerve and cochlear nucleus using signal detection theory techniques to derive neuronal thresholds showed that in centrally projecting neurons, increases in masked thresholds were significantly smaller than the changes measured psychophysically. Larger threshold shifts have been reported in the inferior colliculus of awake marmoset. The present study investigated the magnitude of forward masking in primary auditory cortical neurons of anaesthetised guinea-pigs. Responses of cortical neurons to unmasked and forward masked tones were measured and probe detection thresholds estimated using signal detection theory methods. Threshold shifts were larger than in the auditory nerve, cochlear nucleus and inferior colliculus. The larger threshold shifts suggest that central, and probably cortical, processes contribute to forward masking. However, although methodological differences make comparisons difficult, the threshold shifts in cortical neurons were, in contrast to subcortical nuclei, actually larger than those observed psychophysically. Masking was largely attributable to a reduction in the responses to the probe, rather than either a persistence of the masker responses or an increase in the variability of probe responses.

Show MeSH