Convergence of submodality-specific input onto neurons in primary somatosensory cortex

Abstract

At the somatosensory periphery, slowly adapting type 1 (SA1) and rapidly adapting (RA) afferents respond very differently to step indentations: SA1 afferents respond throughout the entire stimulus interval (sustained response), whereas RA afferents respond only at stimulus onset (on response) and offset (off response). We recorded the responses of cortical neurons to step indentations and found many neurons in areas 3b and 1 to exhibit properties that are intermediate between these two extremes: These neurons responded during the sustained portion of the stimulus and also at the offset of the stimulus. Several lines of evidence indicate that these neurons, which exist in large proportions even at these early stages of somatosensory cortical processing, receive input from both populations of afferents. First, we show that many cortical neurons have both a significant sustained response and a significant off response. Second, the strength of the off response is uncorrelated with that of the sustained response, which is to be expected if sustained and off responses stem from different populations of afferent fibers. Third, the bulk of the variance in cortical responses to step indentations can be accounted for using a linear combination of both SA1 and RA responses. Finally, we show that the off response in cortical neurons does not reflect rebound from inhibition. We conclude that the convergence of modality specific input onto individual neurons is common in primary somatosensory cortex and discuss how this conclusion might be reconciled with previous findings.

FIG. 1.

Top left inset: 892 accepted spikes (black) and 1 rejected spike (gray) of a representative isolation in area 1 (3-ms trace). Responses of a typical slowly adapting type 1 (SA1) (A) and rapidly adapting (RA) (B) afferent to 60 repeated presentations of a step indentation (top right inset); the gray trace in the inset shows the ideal stimulus trajectory. C: responses of an SA1-like neuron in area 3b. D: response of an RA-like neuron in area 1. E: response of a neuron that exhibits both SA1-like and RA-like responses in area 3b. The adaptation index, AI, is a function of the ratio of the off response to the sustained response, each normalized by their population means. As reflected in the on responses, there was a slight ringing in the stimulus after its onset. Notch-filters were used to dampen it, but these did not eliminate it completely. In all analyses, we ensured that the sustained response was measured after the ringing had subsided. The yellow bar shows the response latency. The RF of each neuron, measured using from data obtained in the RF-mapping protocol, is shown to the left of the corresponding raster plot, along with the locations of the 9 probes (centered on the hot spot) indented in the adaptation and variable-durations indentations protocols.

FIG. 2.

Responses to step indentations of 8 SI neurons that vary in their adaptation properties, ranging from almost purely SA1-like neurons (bottom) to almost purely RA-like neurons. The anatomical location and the adaptation index, AI, of the neuron are shown to the right and to the left, respectively, of the corresponding raster. Responses are shifted left (by their latency) so that on and off responses are aligned in the figure.

FIG. 3.

Distribution of the adaptation index, AI, computed from the responses of neurons to step indentations (N_SA1 = 19, N_RA = 13, N_{area 3b} = 83, N_{area 1} = 132). A: the index perfectly dichotomizes SA1 and RA afferents such that SA1 afferents yield AIs near 0 and RA afferents yield AIs near 1. B: the distribution of AI for cortical neurons is bimodal with peaks near 0 and 1 (with a strong preponderance at the high end). However, most neurons yield intermediate values of AI, suggesting that submodality convergence is the rule rather than the exception in SI. Indeed, 59% of neurons in area 3b and 58% of neurons in area 1 yield values of AI that fall between 0.1 and 0.9.

FIG. 4.

Proportion of neurons that exhibited a significant sustained response with no off response (SA1-like), neurons that did not exhibit a significant sustained response but produced a significant off response (RA-like), and neurons that exhibited both a significant sustained and a significant off response. Analyses were performed on the responses of cortical neurons to the adaptation protocol, which consisted of 60 presentations of a step indentation lasting 500ms, interleaved with 500-ms silent intervals (see methods). This analysis was performed on 83 neurons in area 3b and 132 neurons in area 1.

FIG. 5.

Responses of a neuron in area 1 to bars at 8 orientations indented into the skin for 100 ms. The rasters of alternating colors denote responses to bars at different orientations (shown to the left of the corresponding rasters). The sustained response was much stronger when the neuron was stimulated at its preferred orientation than when it was stimulated at the orthogonal orientation. In contrast, the on and off responses were relatively insensitive to orientation. The on and off responses occurred at different latencies because bars at each orientation were indented at different locations within the neuron's RF (see Bensmaia et al. 2008 for details).

FIG. 6.

Interrelationships between the responses evoked during each stimulus interval (on, sustained, and off; N_SA1 = 19, N_RA = 13, N_{area 3b} = 83, N_{area 1} = 132). A: at the somatosensory periphery, the strength of the off response of RA afferents is correlated with the strength of their on response (ρ denotes the partial correlation between the off and on responses, controlling for the sustained response); B and C: the magnitude of the off response of neurons in areas 3b and 1 is correlated with the magnitude of their on response. D: the strength of the sustained response of SA1 afferents is correlated with the strength of their on response (ρ denotes the partial correlation between the sustained and on responses, controlling for the off response); E and F: the magnitude of the sustained and off responses of cortical neurons are correlated, particularly in area 3b. G: RA afferents do not produce sustained responses and SA1 afferents do not produce off responses. H and I: as predicted from the convergence hypothesis, the magnitude of the sustained response of cortical neurons is uncorrelated with that of their off response (r denotes the correlation between the sustained and off responses). Asterisks denote significant correlations (P < 0.01). At the periphery, the mean on responses were 129 and 138ips for SA1 and RA afferents, respectively. The mean sustained response of SA1 afferents was 33ips; the mean off response of RA afferents was 36ips. In cortex, the mean on responses were 62 and 48 ips in areas 3b and 1, respectively. The mean sustained responses across SA1-like and mixed neurons were 8 and 6ips in areas 3b and 1, respectively. The mean off responses of RA-like and mixed neurons were 43 and 32ips in areas 3b and 1, respectively.

FIG. 7.

A: distribution of coefficient of determination (R²) for the fits obtained using the linear model. B: fitted regression parameters for SA1 and RA input.

FIG. 8.

A: responses of an RA afferent to step indentations varying in duration. Responses are aligned on the onset of the off response. The inset shows the magnitude of the off response as a function of the indentation duration. The response is stable for indentations that last longer than 62 ms. B: responses of a neuron in area 3b (AI = 0.52) to indentations varying in duration. The magnitude of the off response was strongly dependent on the indentation duration.

FIG. 9.

Magnitude of the off response as a function of indentation duration for RA afferents (square), neurons in area 3b (circle), and neurons in area 1 (triangle) (N_RA = 7, N_{area 3b} = 28, N_{area 1} = 46). The off response obtained from each neuron for each stimulus duration was normalized by dividing it by the mean off response obtained for that neuron at the longest duration tested (250 ms for RA afferents, 2 s for cortical neurons). For all individual neurons in both anatomical areas, the off response increased as the indentation duration increased. For RA afferents, indentation duration had little effect on the off response for durations >100 ms.