Skip to main content
. 2020 Oct 12;9:e51927. doi: 10.7554/eLife.51927

Figure 4. Classic models of neural properties underlying ITD discriminability and their potential for explaining encoding of ITD statistics.

Figure 4.

(A) Distribution of internal delays replotted from Stern and Colburn, 1978, Figure 2b, which proposes a higher density of pairs of fibers encoding frontal ITDs. (B) The density of pairs of fibers proposed by Stern and Colburn, 1978 as the mechanism underlying ITD discriminability could effectively achieve the representation of ITD statistics: the density was adjusted to match the pattern of √FIITD. Note that the adjusted distribution largely preserves the shape of the distribution of the original model. (C) Distribution of IPD-tuning maximizing coding across the physiological range of ITD, as proposed by Harper and McAlpine, 2004. Top, single-neuron Fisher information as a function of IPD. Bottom-left, distribution of best IPDs brain across frequency expected for humans under the framework proposed by the authors; white straight lines indicate physiological range determined by the distance between ears. Bottom-right, reconstructed neuron population Fisher information, converted from IPD to ITD for each frequency for obtaining the predicted ITD discriminability; black curved lines indicate the π-limit, beyond which ITD cues become ambiguous within narrow frequency bands. Spearman correlation coefficients for the relationship between population Fisher information and ITDrc and √FIITD outlined above. (D) The IPD-tuning distribution proposed by Harper and McAlpine, 2004 as a mechanism underlying ITD discriminability was adjusted for matching the neuron population Fisher information to ITDfi. Top, the neuron distributions matching ITD statistics depict best IPDs away from midline across frequency, consistent with a coding strategy based on two clustered subpopulations tuned to IPDs away from the front (McAlpine et al., 2001; Harper and McAlpine, 2004; Hancock and Delgutte, 2004; Pecka et al., 2008). Bottom, the neuron population Fisher information highly correlated with the ITD statistics.