Article Figures & Data
Figures
- Fig. 1.
Simultaneous recordings from geniculate cells and simple cells with overlapping receptive fields. Top, Illustration of the experiment. A single electrode was inserted into layer 4 of visual cortex; up to seven electrodes were inserted into the LGN. Visual responses to a white-noise stimulus (shown) and drifting sine-wave grating were recorded. Bottom, Receptive fields and cross-correlograms for each pair of cells (n = 11) that were monosynaptically connected and studied for interactions between geniculate spikes. Regions of a cell's receptive field excited by the bright phase of the white-noise stimulus (see Materials and Methods) are shown in white; regions excited by the dark phase are shown in black. The two panels for each pair of receptive fields correspond to the same stimulus window.Circles represent a Gaussian fit to the geniculate receptive field (radius: 2.5ς). Stimulus pixel size was 0.4°. The cross-correlograms shown to the right of the receptive fields each have a short-latency peak (above the stimulus-dependent shuffle correlogram, shown in gray), indicating a monosynaptic connection between the geniculate cell and the simple cell.
- Fig. 2.
Comparison of the effectiveness of two geniculate spikes (interspike interval = 5.0 ± 0.2 msec) from the same geniculate cell. A, B, Receptive fields of a geniculateoff-cell and cortical simple cell mapped with white noise (details as in Fig. 1). C, Raster plot of simple-cell spikes relative to 1000 geniculate spikes. The simple cell often fired ∼2 msec after a geniculate spike. D, Cross-correlogram, which is equivalent to the sum of all rows of the raster plot (C); units are simple-cell spikes per second after the average geniculate spike. The narrow, short-latency peak [above the stimulus-dependent shuffle correlogram (gray line);Perkel et al. (1967)] indicates that the geniculate cell provided monosynaptic input to the simple cell (Tanaka, 1983, 1985; Reid and Alonso, 1995). Visual stimulus: drifting sine-wave grating of optimal orientation. Total time: 2062 sec. Geniculate spikes: 64,779; simple-cell spikes: 8635. Efficacy of geniculate input: 2.1% (see Materials and Methods). E, Raster plot of simple-cell firing relative to 1000 pairs of geniculate spikes, each separated by 5.0 ± 0.5 msec and preceded by > 5.0 msec dead time. More simple-cell spikes followed second geniculate spikes than followed first geniculate spikes. F, Cross-correlogram between paired geniculate spikes and simple-cell spikes. Geniculate paired spikes: 5450. Efficacy of first spikes (peak 1): 1.2%. Efficacy of second spikes (peak 2): 5.1%.
- Fig. 3.
Time course and magnitude of homosynaptic paired-spike enhancement. A, Illustration of analysis showing the temporal relationships between dead time and two successive LGN spikes. Pairs of spikes with a given interspike interval (ISI) were included in the analysis if there were no spikes before spike 1 within the dead time (10 or 20 msec). B, C, Efficacy of pairs of geniculate spikes (percentage that evoked a simple-cell spike) that occurred at different ISIs after a dead time > 10 msec and > 20 msec; these dead times were selected because they are shorter (> 10 msec) and longer (> 20 msec) than the duration of homosynaptic reinforcement (∼15 msec; see Results for reasoning). Curves in B and C are from the same geniculate cell and simple cell. Gray lines: efficacy of first geniculate spikes; black lines: efficacy of second geniculate spikes. Efficacy at each geniculate ISI was calculated as in Figure 2, then smoothed with 4 msec boxcar average. D, E, Average efficacy profiles for all 11 pairs of cells after a dead time > 10 msec and > 20 msec.
- Fig. 4.
Scatter plot of efficacies of second versus first geniculate spikes; all but one pair of cells showed homosynaptic enhancement of second spikes. Efficacies were averaged over ISI range 4–10 msec, weighted by number of occurrences of each ISI. The visual stimulus was a drifting (4 Hz) sine-wave grating. Dead time (DT) was either > 10 msec (white circles) or > 20 msec (gray circles). Points on either axis are for efficacies < 0.1%.
- Fig. 5.
Time course and magnitude of heterosynaptic paired-spike enhancement. A, Illustration of analysis showing the temporal relationships between spikes from two LGN cells (LGN A and B). Spike 1 was from cellA; spike 2 was from cell B. For various ISIs between spike 1 and spike 2, it was required that cell A fire no spikes during a lockout period, which included the ISI plus 4 msec. The additional 4 msec insured that a second spike from cell A did not sum with spike 2 (from cell B). In addition, the analysis required that cell B produced no spikes during a dead time of 15 msec (the approximate duration of homosynaptic interactions) before spike 2. B, Time course of heterosynaptic interactions (effect of cell A on cell B) versus homosynaptic interactions (effect of B on B) for a single cell. C, Averaged heterosynaptic interactions for the 4 ‘triplets’ in our data set. For each triplet, the effect was more pronounced when the strong input preceded the weak input. D, Comparison of average heterosynaptic interactions (strong input preceding weak) versus average homosynaptic interactions for same cells (as in B).
