On the Tonotopy of the Low-Frequency Region of the Cochlea

It is with great interest we read the above paper (Recio-Spinoso, Dong et al. 2023). We accept that our findings (Burwood, Hakizimana et al. 2022) depart from prior understandings of cochlear tonotopy and therefore merit close scrutiny, but there are several aspects of Recio-Spinoso and colleagues’ data that merit a response.
The first issue regards the measurement location along the cochlear duct. Examination of the results from the guinea pig leads us to conclude that the two publications are not comparable, because Recio-Spinoso et al. are not examining the same cochlear region as we did. Recio-Spinoso et al. base their conclusions on data from the 95% and 75% locations. According to their 20 mm value for the length of the guinea pig cochlea, their most apical measurement location is 1 mm from the helicotrema, which would be more akin to the “middle location” data that we presented. This location shows low-pass tuning both in our data and in Recio-Spinoso’s.
Our most apical measurement is at an estimated distance of 360 microns from the helicotrema. This location showed a higher best frequency than the middle and basal locations, a finding clearly at odds with traditional models of tonotopicity. Recio-Spinoso’s data do not address this issue, because they only probed the response of locations closer to the cochlear base in their guinea pig data. They did, however, find that their velocity plots indicated a CF of 300 Hz for their 95% location. This is higher than the 200 Hz predicted by standard cochlear maps (e.g.(Viberg and Canlon 2004)).
A similar problem affects their most basal measurement site, reported as being at 75% of the cochlear length. This corresponds to a distance of 5 mm from the helicotrema, whereas our most basal site was at the 4.3 mm location. The 700-micron difference in measurement location can explain the apparent differences in tuning.
We estimate our places of measurement to be at 98%, 88% and 78% of the distance from the base. Hence, our most basal location is found at a more apical cochlear place – and our measurement locations are closer together than the ones shown by (Recio-Spinoso, Dong et al. 2023). Our conclusions were that the apical 20% of the cochlea shows physiologically vulnerable, coherent vibration during near-threshold stimulation, and this led us to speculate that a transition to tonotopy must exist. It seems likely that the 75% location is located basal to this transition.
Other aspects of Recio-Spinoso’s data are consistent with our findings, despite their conclusion being different. For instance, the authors recognize that there were “little differences between the 95% and 92% sites, as well as between the 80% and 75% sites”. This is not consistent with traditional models of tonotopicity. Furthermore, the data in the paper consist of examples from individual animals, which does not show the repeatability of the data. Descriptive statistics such as means, standard deviations, and ranges would have been informative, as would a systematic plot of the estimated best frequencies with position, like the one in our last figure.
Similar problems with measurement locations affect the gerbil data. The measurement positions are reported to be 89% and 64% of the cochlear length. This distance places the latter measurement in the middle turn, not in the apical turn – and notably the best frequency is likely to be above any phase locking frequency.
Finally, the recording locations from the chinchilla are separated by 4.8 mm, the largest difference in the dataset. A Greenwood function would predict that the response at the 99.6% location of the Chinchilla cochlea should be lower than the reported 100 Hz. The 100 Hz place should be close to 95% of the cochlear length if tonotopy is maintained throughout the cochlea (Qiu, Salvi et al. 2000). This finding supports our observation that the very apical locations of the cochlea appear to respond best to frequencies higher than predicted. The center of the response for the 73% location seems to respond at 350 Hz, which is in line with Greenwood’s prediction, but hints that perhaps these two measurement locations are again too far apart to compare to our measurements.
A second point that Recio-Spinoso et al do not fully address is the influence of sound pressure level on the observed phenomena. This point is fully addressed in a prior paper from the authors (Recio-Spinoso and Oghalai 2018). Their figure 3D on group delay differences between recording locations show a minimization of this difference at the lowest sound levels tested, and a divergence with increasing sound pressure level. We therefore feel that the use of a click that is “20 dB below the maximum response of the sound system” may not induce displacements that reflect domination of the active component. As is clearly seen in our data, the minimization of delays is a near-threshold phenomenon that does not occur at higher stimulus levels or after cochlear amplification has been abolished. This would also explain the difference in phase slope between 95% and 75% locations – as these are highlighted from a 70 dB SPL recording (See figure 3E in (Recio-Spinoso, Dong et al. 2023)).
This point also contends with the authors’ observation about (Warren, Ramamoorthy et al. 2016). The tuning curves in figure 3 were garnered from a single location and using a stimulus level of 90 dB SPL, far in excess of those shown by either (Burwood, Hakizimana et al. 2022) or (Recio-Spinoso, Dong et al. 2023). The matter of tuning is also interesting – we find in our data that the lower frequencies below the CF are more non-linear in the middle and basal regions. Higher SPLs could therefore sharpen the response of the reticular lamina. Certainly this phenomenon is magnified by the use of a multi-tone stimulus (see (Burwood, Hakizimana et al. 2022) fig 2D, where the response at the Middle location is sharper at a higher SPL, but is low pass at a lower SPL).
We also note an error that the authors should correct. On line 190, it is stated that the separation between the 95% and 75% sites is 400 µm. This should be 4000 µm based upon the percentage difference in locations relative to a 20 mm cochlea.
In summary, the conclusion of Recio-Spinoso et al is not adequately supported by their data. Due to their choice of measurement locations and sound parameters, their data does not address the findings by Burwood et al. Further concerns on the use of ROI averaging will be addressed in a forthcoming publication.

References:
Burwood, G., P. Hakizimana, A. L. Nuttall and A. Fridberger (2022). "Best frequencies and temporal delays are similar across the low-frequency regions of the guinea pig cochlea." Science Advances 8(38): eabq2773.
Qiu, C., R. Salvi, D. Ding and R. Burkard (2000). "Inner hair cell loss leads to enhanced response amplitudes in auditory cortex of unanesthetized chinchillas: evidence for increased system gain." Hearing research 139(1-2): 153-171.
Recio-Spinoso, A., W. Dong and J. S. Oghalai (2023). "On the tonotopy of the low-frequency region of the cochlea." Journal of Neuroscience.
Recio-Spinoso, A. and J. S. Oghalai (2018). "Unusual mechanical processing of sounds at the apex of the guinea pig cochlea." Hearing research.
Viberg, A. and B. Canlon (2004). "The guide to plotting a cochleogram." Hearing research 197(1-2): 1-10.
Warren, R. L., S. Ramamoorthy, N. Ciganovic, Y. Zhang, T. M. Wilson, T. Petrie, R. K. Wang, S. L. Jacques, T. Reichenbach, A. L. Nuttall and A. Fridberger (2016). "Minimal basilar membrane motion in low-frequency hearing." Proc Natl Acad Sci U S A 113(30): E4304-4310.