Depth-dependent temporal response properties in core auditory cortex

Abstract

The computational role of cortical layers within auditory cortex has proven difficult to establish. One hypothesis is that interlaminar cortical processing might be dedicated to analyzing temporal properties of sounds; if so, then there should be systematic depth-dependent changes in cortical sensitivity to the temporal context in which a stimulus occurs. We recorded neural responses simultaneously across cortical depth in primary auditory cortex and anterior auditory field of CBA/Ca mice, and found systematic depth dependencies in responses to second-and-later noise bursts in slow (1-10 bursts/s) trains of noise bursts. At all depths, responses to noise bursts within a train usually decreased with increasing train rate; however, the rolloff with increasing train rate occurred at faster rates in more superficial layers. Moreover, in some recordings from mid-to-superficial layers, responses to noise bursts within a 3-4 bursts/s train were stronger than responses to noise bursts in slower trains. This non-monotonicity with train rate was especially pronounced in more superficial layers of the anterior auditory field, where responses to noise bursts within the context of a slow train were sometimes even stronger than responses to the noise burst at train onset. These findings may reflect depth dependence in suppression and recovery of cortical activity following a stimulus, which we suggest could arise from laminar differences in synaptic depression at feedforward and recurrent synapses.

Figure 1.

Histological confirmation of depth of LFP reversal point. A, A section stained for Nissl substance, used to identify laminar boundaries. B, The adjacent section in the series from this animal, stained for cytochrome oxidase, reveals a lesion near the layer I/II boundary (indicated by the arrowhead). C, LFP traces demonstrating the reversal point (black) where the lesion was placed. Channels both above and below the reversal (gray) are shown for illustration. D, E, Also plotted is the distribution of multiunit clusters by depth for both AI (D) and AAF (E), in percentage of clusters (left axis) and percentage of penetrations which included a cluster at that depth (right axis; note that only one cluster can exist per depth per penetration, so the two axes are linearly related).

Figure 2.

Example data from one multielectrode penetration. A–C, Spike rasters show simultaneous multiunit cluster recordings from 200, 400, and 600 μm depths in the same AAF penetration, for representative noise burst train rates of 1 (A), 3 (B), and 6.25 (C) bursts/s. Rasters have been collapsed into 5 ms bins and plotted as peristimulus time histograms in D–F, using the same gray-scale scheme as in A–C to indicate recording depth. In G, the following index FI is shown for the same three recording depths (gray-scale scheme as in A–F), and plotted for the full range of stimulus train rates.

Figure 3.

Following index magnitude varies with cortical depth. A, Data from two representative penetrations from AI (left) and AAF (right), illustrating FI-rate functions. Each curve represents a different recording depth, with lighter curves indicating more superficial depths. The AAF example is the same as in Figure 2, with more depths included. B, Population averages of FI (mean rate of response to second-and-later bursts normalized by mean rate of response to first bursts) as a function of train rate for AI (left) and AAF (right). Error bars are omitted for visual clarity. C, The peak following decreases with depth in both AI and AAF (Kruskal–Wallis test, p < 10⁻⁶; black lines give mean ± SEM; gray lines in box give the median; box extends from the 25th to 75th percentile; whiskers show the extent of the data out to at most half again the interquantile distance; outliers are marked with x).

Figure 4.

Corner points for FI-rate functions in AI and AAF. A, Schematic illustration of rolloff (r_ro) and floor (r_fl) corner points for an idealized Fl-rate function. B, r_ro was depth dependent in both AI and AAF (Kruskal–Wallis test, p < 0.005 for both); the rolloff in FI-rate functions began at slower rates in deeper clusters. Black lines give mean ± SEM; gray lines in box give the median; box extends from the 25th to 75th percentile; whiskers show the extent of the data out to at most half again the interquantile distance; outliers are marked with x. C, D, Area differences in r_ro and r_fl distributions. Rolloff (C) began at slower rates in AI (left; 421 clusters) than in AAF (right; 413 clusters). Similarly, FI reached its floor value (D) at slower rates in AI (417 clusters with defined r_fl) compared with AAF (418 clusters). However, this effect was not depth dependent (Kruskal–Wallis test, n.s.; data not shown). N/A, Corner point did not occur within the range of train rates presented in this study.

Figure 5.

Non-monotonicity in Fl-rate functions. A, Idealized non-monotonic Fl-rate function; non-monotonicity is circled and identified as r_n. B, In both AI (left) and AAF (right), non-monotonic FI-rate functions occurred almost exclusively at train rates of 3 bursts/s. C, The distribution of clusters with significantly non-monotonic FI-rate functions was depth dependent in AAF (with a peak near 400 μm) but not in AI. D, However, in both AI (circles) and AAF (diamonds), the magnitude of non-monotonicity was depth dependent (Pearson's correlation test, p < 0.01), with the largest non-monotonic increase in FI between adjacent train rates occurring at more superficial depths. Magnitude of non-monotonicity (on y-axis) is defined as the difference in FI between the rate identified as non-monotonic and the immediately slower rate (i.e., FI(r_n) − FI(r_{n − 1})).

Figure 6.

Augmenting responses in AAF. A, Augmenting responses to second-and-later bursts in a noise burst train occurred primarily at slower train rates near 3–4 bursts/s. B, Augmenting responses were also restricted primarily to mid-to-superficial layers, particularly 300–400 μm below the cortical surface (distribution significantly different from distribution of all clusters across depth, χ² goodness-of-fit, p < 10⁻⁴). C, The magnitude of the augmenting response was negatively correlated with depth (Pearson's correlation test, p < 0.05).

Figure 7.

Time course of augmenting responses. A, B, Example augmenting responses at 4 bursts/s (A; cluster m00167p32d300) and 3 bursts/s (B; cluster m00203p2d400). PSTHs (black) are superimposed over spike rasters (gray dots). C, Average normalized PSTH for all AAF clusters with (gray) or without (black) significant augmenting responses. Each cluster PSTH contributing to these averages was computed for responses to all noise bursts separated by at least 500 ms, and normalized by subtracting spontaneous rate and scaling the maximum value to one. Width of traces gives mean ± SEM; dotted line at zero represents spontaneous activity level. The clusters with augmenting responses show more post-activation suppression, and greater rebound activation ∼300–350 ms after stimulus onset.

Figure 8.

Magnitude of post-activation suppression is depth dependent. A, Population PSTHs broken into coarse depth categories reveal that the post-activation suppression of activity is greater in superficial layers in AI (left), and to a lesser extent in AAF (right). Each cluster PSTH contributing to these population PSTHs was shifted to have a spontaneous rate of zero and then normalized to a maximum of one before averaging across clusters. Inset shows an expanded view of activity over the 100–250 ms poststimulus interval. Width of population PSTHs gives mean ± SEM of all cluster PSTHs within the relevant depth range; lighter shades correspond to more superficial depths (color scheme as in B). B, Mean values of the normalized individual cluster PSTHs over the time range of 100–250 ms, grouped into coarse depth categories as in A. More negative values indicate deeper postactivation suppression. In AI, the suppression is greater in mid-to-superficial layers (Kruskal–Wallis test, p < 10⁻⁵; black lines give mean ± SEM; colored lines in box give the median; colored box extends from the 25th to 75th percentile; whiskers show the extent of the data out to at most half again the interquantile distance; outliers are marked with x). In AAF, the effect is weaker, but the suppression is significantly less in the deepest layer than in the most superficial layers (Kruskal–Wallis test, p < 0.01).

Figure 9.

Precision of spike timing in responses to noise bursts is depth dependent in AI and AAF. A, Population averages of within-train spike-time SD as a function of train rate for AI (left) and AAF (right); each curve represents a different depth, with lighter gray indicating more superficial depths. Dotted line indicates the 28.8 ms SD expected if spikes were uniformly distributed over the analysis interval. Error bars are omitted for visual clarity. B, The minimum within-train spike-time SD over train rates increases with depth in both AI and AAF (Kruskal–Wallis test, p < 10⁻⁵; black lines give mean ± SEM; gray lines in box indicate the median; box extends from the 25th to 75th percentile; whiskers show the extent of the data out to at most half again the interquantile distance; and outliers are marked with x). C, The difference between first-burst spike-time SD and minimum within-train spike-time SD, normalized by the sum of the two SD, is near zero at all depths, and shows no significant depth dependence in either AI or AAF (Kruskal–Wallis test, p > 0.1).

Figure 10.

Spontaneous and driven firing rates are depth dependent. A, Both the spontaneous firing rate (solid line) and the mean firing rate in response to the first noise burst in all trains (dashed line) vary with depth in both AI (left) and AAF (right). Data are mean firing rates ± SEM over all multiunit dusters recorded at each depth. Box plots are omitted for visual clarity since two datasets are plotted in the same axes. For both areas, the dependence on depth is significant (Kruskal–Wallis test, p < 10⁻⁶) with a peak at ∼600 μm. B, The difference between first-burst response rate and spontaneous rate, normalized by the sum of the two rates, is constant relative to depth in AAF (right; Kruskal–Wallis test, p > 0.5), but varies with depth in AI (left; Kruskal–Wallis test, p < 10⁻³). Black lines give mean ± SEM; gray lines in box indicate the median; box extends from the 25th to 75th percentile; whiskers show the extent of the data out to at most half again the interquantile distance; and outliers are marked with x.