With suitable steps of signal amplification and filter settings we could avoid an overload of the amplifier and thus reduce the duration of the TMS artefact to allow reliable spike detection about 5–10 ms after TMS onset. Figure 1B shows traces of the analog signal including the TMS artefact and extracellularly recorded action potentials. Figure 1A shows the position and orientation of the coil relative to the cat's brain and the recording site. In total, the effects of TMS on single-unit activity were studied in 85 neurons of cat primary visual cortex with one or both of the two stimulus protocols described in Methods. The sample includes cells of both the simple and complex type distributed over layers 2–6 of the visual cortex. All cells were pooled because we found no significant differences in the TMS effect for cell type and recording depth in this sample.
Protocol I: the effect of TMS on spontaneous activity
With stimulus protocol I we systematically tested the effect of single TMS pulses of different strength on spontaneous activity. In a few cases, when spontaneous activity was very low, a flickering bar was presented to the receptive field of the neuron to elevate activity. The ISI was 6 s to enable the analysis of long-lasting effects. Examples for the effect of single TMS pulses of different strength on spontaneous or visually elevated activity are shown in Fig. 2. Common to all examples is a strong facilitation of spike activity during the first 500 ms for stimuli stronger than 40 % and a subsequent long-lasting suppression of activity, the duration of which increases with stimulus strength. Shorter and earlier episodes of facilitation were also evident with lower stimulus strength (see Fig. 2A and B). Often, an early and strong suppression of activity occurred with TMS strength exceeding 40 %. This suppression had a maximal duration of 200 ms from TMS onset (Fig. 2C and D; sometimes between 50 and 100 ms, Fig. 2B) and seemed to cancel out the early part of facilitation already evoked with lower stimulus strengths. The increased activity following this early suppression may be in part related to rebound excitation as the result of release from inhibition. The sharp peaks in Fig. 2D are very likely be produced by such a rebound process and consist of high-frequency bursts of action potentials. In some cases, as shown in Fig. 2B and C, the episode of facilitation showed multiple peaks of activity, indicating an oscillation of neuronal activity. The peaks occurred at intervals of about 100 ms, corresponding to an oscillatory activity within the EEG alpha range (≈10 Hz). An oscillation within this frequency is known to develop in a visual cortex deprived of sensory input, and thus it is likely that the TMS pulse acts as a phase reset, causing a stronger oscillation thereafter. Trials with different TMS strengths could not be performed in an interleaved manner because stimulator output strength could not be controlled by remote. Therefore, some variations in activity levels between traces result from fluctuations in cell activity over time, which were usually correlated with changes in EEG pattern (arousal state) and are typical for this kind of anaesthesia (Li et al. 1999). This could be one explanation for a discontinuous increase in the strength of facilitation or suppression with increasing TMS strength, including cases where the TMS effect was weaker at a stimulus strength of 90 or 100 % compared to 80 %. On the other hand, there might be a shift in the balance of suppressive and facilitating processes. On average, there was no decline of the TMS effect at high stimulus strength, but probably some saturation at around 80–90 %.
Figure 2. Effect of TMS pulses of different strengths on spontaneous or visually evoked activity
TMS was applied either during spontaneous activity (B-D) or during visually evoked activity by uncorrelated flicker of a bright bar (A). Data from two simple cells are shown in A and D, and from two complex cells in B and C. The cortical recording depths were 777 µm (A), 1100 µm (B), 1427 µm (C) and 185 µm (D). For further explanation see Results.
Download figure to PowerPoint
Figure 3A shows the spontaneous activity averaged for 20 cells. The mean spontaneous activity level prior to the TMS pulse was set to zero to emphasise increased and decreased post-stimulus episodes (black filled areas). Mean (±s.d.) spontaneous activity was 12.8 ± 8.06 spikes s−1 (n= 72 records, range 3.6–40.9 spikes s−1). For cells with very low spontaneous activity (< 3 spikes s−1), a flickering bar stimulus within the receptive field elevated activity to 13.9 ± 8.6 spikes s−1 (n= 47 records, range 4.9–50.1 spikes s−1). The thin line depicts the standard deviation for every data point of the mean spontaneous activity. While moderate stimuli (20–30 %) caused short early facilitation (up to 100 ms) followed by a long-lasting suppression (about 1 s), stronger stimuli (60–100 %) generated sequences of stronger and longer facilitation (up to 500 ms) that were interrupted by an early suppression of activity (100–200 ms). The ensuing late suppression was prolonged (could exceed 5 s). The early suppression was less pronounced in this grand average due to the inclusion of cells with and without early inhibition. Nevertheless, even the standard deviation shows a clear dip at around this time. The very early peak of activity directly following the TMS artefact is in part due to the facilitation of activity, but is also contaminated by a recharge artefact of the MagStim rapid, which occurred at around that time with strong stimuli (70–100 %). This recharge artefact is visible in the upper traces of Fig. 2A-D. Therefore, the very early part of the facilitation was occluded from further quantitative analysis.
Figure 3. Grand average and statistics of TMS effect on spontaneous activity
A, grand average of records obtained with different TMS strengths during spontaneous activity. The mean level of spontaneous activity prior to TMS was set to zero to emphasise increases and decreases in activity (black filled areas). The thin line depicts the standard deviation. B, scatter plots showing the strength and time of increase (triangles) and decrease (squares) in activity relative to the pre-TMS level of spontaneous activity. A threshold was set at +15 and −15 spikes s−1, corresponding to two times the standard deviation of spontaneous activity. Data of the same records as those averaged in A are shown for different strengths of TMS pulse. PSTHs (A) and scatter plots (B) were obtained from 20 cells, in only 18 of which was a range of 80–100 % calculated. Total number of records analysed: 10–30 %, 36; 40–50 %, 36; 60–70 %, 37; 80–100 %, 32.
Download figure to PowerPoint
The scatter plots shown in Fig. 3B show the temporal distribution of changes in activity based on measurements of single trials. To include only significant changes, a threshold for changes in activity was set to ± 15 spikes s−1, which is close to two times the standard deviation for spontaneous activity without TMS. It is evident that facilitation prevailed during the first 500 ms from TMS onset (TMS at 0), while the late suppression was almost continuous, but strongest after 500 ms. Facilitation and suppression increased with increasing stimulus strength and the peak of late suppression also shifted from about 0.5 s to about 1.2 s. With a 40–50 % stimulus strength, the late suppression caused less strong reductions in activity, but data points are much more frequent at around 1 s than later. Suppression of activity (in spikes s−1) seems to be weak compared to facilitation, however, it has to be taken into account that spontaneous activity rates were low and spike activity cannot be negative. The data points indicating an increase in activity after about 1 s should not be interpreted as a real excitation. Single bursts of activity occurred during the late suppression or at its termination possibly as a result of a rebound process, but suppression of spontaneous activity prevailed at that time.
Protocol II: the effect of TMS on visually induced activity
In this case a moving bright or dark bar was combined with a single TMS pulse. The moving bar is better suited to stimulate area 17 cells and allows the stimulation of the peripheral and central aspects of the receptive field subsequently, and, when using opposite motion directions, aspects of direction specificity could be studied (but is not a topic of this report because of the small sample of direction-specific cells). Orientation tuning was not investigated systematically because it is too time consuming when different strengths and the timing of the TMS pulse have to be tested. TMS was only tested for bars of optimal orientation. Usually, only three different stimulus strengths were tested for one cell (20–30 %, 40–50 % and 60–70 %; a few cells were tested with 80–100 %) because this stimulus protocol was intended to test primarily the effect of different temporal intervals between TMS and the visual response. Therefore, the TMS pulse was given at different time points during the motion trajectory of the bar stimulus, thus appearing at different times relative to the response induced by the visual stimulus. Initially, the TMS pulse was given at the onset of bar motion, and visual activity followed the TMS pulse by some 100 ms. The temporal difference between TMS and the onset of the visual response was then reduced in a stepwise manner. First, when the temporal interval between TMS and the visual response was large, it was reduced in steps of 100 ms. Closer to the visual response and during it, the steps were reduced to 50 ms and less to achieve a better temporal resolution for effects directly following the TMS. The same procedure was usually performed for both visual responses, those elicited by forward motion of the bar and those evoked by backward motion. Each interval between TMS and the visual response was tested 32 times, and trials with visual stimulation alone, TMS alone and a combination of both were interleaved. This was done to test, by subtraction of the different activities, whether the TMS pulse by itself elicited suprathreshold activity or whether it facilitated subthreshold visual inputs.
Figure 4 and Figure 5 give examples of interactions between TMS and visual activity for two different cells. In a reduced form, Fig. 4 shows only four different situations: pure visual stimulation (A), pure TMS (B) and combined visual stimulation and TMS at different time points of TMS (C and D). Only the response to forward motion of the bar is shown. TMS alone at moderate strength (30 %) evoked only a few action potentials early after the pulse (asterisk, Fig. 4B), as was found for other applications during spontaneous activity. TMS pulses given prior to the visual response caused a facilitation of the visual response within a time window of less than 100 ms from TMS onset (Fig. 4C and D). In both cases the facilitation exceeded the algebraic sum of activity evoked by visual stimulation (Fig. 4A) and TMS alone (Fig. 4B), indicating that subthreshold visual activity had been pushed above threshold. This holds both for the peak response and for the previously weak activity evoked at the periphery of the receptive field (compare Fig. 4A and D, time point indicated by the diamond).
Figure 4. Effect of TMS on visual responses to moving bars
Visual response of a cortical neuron (simple cell) to a bar moving across its receptive field. For simplicity, only the response to one direction of bar motion is shown here (see inset of motion trajectory). A, visual stimulation alone. B, TMS alone. C and D, TMS combined with visual stimulation, with TMS given at two different times. Note that the activity evoked by TMS alone (asterisks) is less than the increase of visual activity during combined TMS and moving bar (see response components labelled with asterisks and diamonds). Arrows indicate the TMS artefact.
Download figure to PowerPoint
Figure 5 shows part of a sequence of different timings (nine out of 14) of the TMS pulse for another cell and for the full stimulus cycle with forward and backward motion of the bar. The upper rows show the mean responses for combined TMS and visual stimulation, the lower rows show the controls with visual stimulation alone. The diagram to the lower right shows the activity after TMS alone; actually no significant change of spontaneous activity is evident in this case. The PSTHs demonstrate two effects that are further quantified in Fig. 6. After a long delay, the TMS pulse inhibited visual activity. In records 1–8 the second visual response (to backward motion) is partly diminished (see downward pointing arrows in Fig. 5). Facilitation of the visual response was achieved when the TMS pulse was given close to the visual response (see upright arrows in diagrams 5, 7, 8, 11, 12 and 13). For this cell, facilitation of visual activity was stronger for the second response, which was the stronger response in the control condition with pure visual stimulation. This might be an indication for a stronger facilitation of activity with more numerous active, but subthreshold, excitatory inputs available for the preferred direction of the stimulus, or the other way round, more numerous inhibitory inputs active for the non-preferred stimulus direction strengthened by TMS. The sample of direction-specific cells tested with the same set of intervals between TMS and response onset, however, is too small to allow any statistically significant conclusions to be drawn.
Figure 6. Effect of TMS presented at different times and at different strengths on visual responses
Series of PSTHs showing changes in the visual response amplitude due to TMS pulses of different strengths (A, 30 %; B, 50 %; C, 70 %) applied at different times relative to the visual response (traces 1–14). The diagrams were obtained by subtracting the activity evoked by visual stimulation alone from the activity elicited by combined TMS and visual stimulation. The traces thus show the enhancement (grey areas) and reduction (black areas) of activity caused by the additional TMS (the artefacts appear as unfilled regions). Arrows point to a sequence of suppression and facilitation, possibly resembling a rebound phenomenon. All measurements were performed in the same cell; part of the original PSTHs are shown in Fig. 5. Grey bars below A label episodes of visual activity.
Download figure to PowerPoint
Figure 6 quantifies and summarises all measurements performed with the cell already introduced in Fig. 5. The traces in the diagrams show the difference of activity elicited by combined TMS and visual stimulation compared to visual stimulation alone for TMS tested at 30, 50 and 70 %. For quantification of the TMS effects, we subtracted the activity recorded during TMS application without visual stimulation (not shown here because there was no effect of TMS alone; only the TMS artefact would then be missing). Peaks filled in grey resemble facilitation of the visual responses, those in black show the reduction of visual activity. As already seen in the PSTHs of Fig. 5, facilitation was only found when the TMS pulse was just in front of the visual response (traces 5–9 for the first response and traces 11–14 for the second response), while suppression of the second response was observed when TMS was several hundred milliseconds prior to the visual response (traces 1–9). Late suppression was almost identical for TMS strengths of 30, 50 and 70 %, but facilitation of the first visual response increased with TMS strength. A suppression of activity during the first visual response is also visible in the upper traces of the diagrams (9–14), although the TMS pulse followed. This is an indication that the TMS pulse given in the previous trial still had a suppressive effect, lasting even for 5–6 s. The visual response itself is not visible in these traces because only the changes are shown, and these were strongest during the visual response. An indication of the time during which the visually elevated activity occurred is indicated by the grey bars below Fig. 6A).
The short peaks of suppression (black) in the traces of Fig. 6 could further raise the illusion that suppression occurred only at distinct times after TMS. This is related to the fact that considerable reduction (and enhancement) of activity can only occur at the time of visually induced activity. The level of spontaneous activity before and after the visual response was too low to allow a considerable reduction of activity (there is nothing to reduce) and no subthreshold visual inputs are available to be pushed above spike threshold. Visually induced responses had amplitudes between 27.4 and 172.3 spikes s−1 (mean 82.7 ± 29.2 spikes s−1, n= 48 cells). To get a better impression of the time course of facilitation and suppression, we plotted all changes in visual activity exceeding ± 50 spikes s−1 for all records obtained with different time points of TMS, and aligned the time axis to TMS onset (time 0). The scatter plots shown in Fig. 7 show the data obtained from 48 cells, grouped in four ranges of TMS strength. The complete set of TMS strength could not be tested for every cell, and some settings were tested twice and included in the sample (20–30 %: 39 cells 290 records; 40–50 %: 48 cells, 256 records; 60–70 %: 28 cells, 99 records; 80–100 %: 7 cells, 14 records). Similar to the scatter plots shown for the spontaneous activity, visually induced activity was also facilitated up to 500 ms following the TMS pulse, with the strongest facilitation occurring during the first 200 ms. Early suppression occurs preferentially at around 100 ms post-TMS, while late (long-lasting) suppression peaks at around 1 s. Only a few measurements were made with 80–100 % TMS, but these also show early facilitation and late suppression of visual activity. However, because of the small sample size, a real quantification for this range of TMS strength is impossible.
Figure 7. Statistics for the effect of TMS on the visual responses evoked by a moving bar stimulus
For individual measurements, the scatter plots show the strength and time of changes in activity elicited by single TMS pulses of different strengths (20–30 %: 39 cells, 290 records; 40–50 %: 48 cells, 256 records; 60–70 %: 28 cells, 99 records; 80–100 %: seven cells, 14 records). The time axis is aligned to TMS onset and the threshold for changes in activity was set to ± 50 spikes s−1. Facilitation following a suppression within 60 ms (rebound) is shown by diamonds. Facilitation missing a preceding suppression is shown with triangles. Squares are used to plot the suppression of activity.
Download figure to PowerPoint
Some increase of activity was also found later than 500 ms relative to TMS onset; however, this facilitation was usually accompanied by a preceding suppression (see arrows in Fig. 6) and may be the result of a rebound process. Therefore, we distinguished between facilitation with and without a preceding suppression: the diamonds in Fig. 7 indicate a rise in activity following a drop in activity within 60 ms, while the triangles label facilitation without preceding suppression. The diamonds are largely found at long delays, between 200 and 1200 ms, indicating that rebound responses occurred primarily with inhibition during the long-latency suppression of activity. At short delays, facilitation of activity without preceding suppression dominates in the scatter plots of Fig. 7.