The Ferrier Lecture 2004 what can transcranial magnetic stimulation tell us about how the brain works?

Abstract

Transcranial magnetic stimulation (TMS) is a technique whereby parts of the cerebral cortex and underlying white matter can be excited by a brief electrical current induced by a similarly brief, rapidly fluctuating magnetic field which is itself produced by rapidly discharging a current through an insulated coil held against the scalp. When combined with magnetic resonance structural and functional images of the subject's brain, the stimulation can be directed at specific cortical areas. Over a period of only 15 years, TMS has revealed hitherto unsuspected aspects of brain function, such as the role of distant parts of the brain in recovery from stroke, and has helped to resolve several previously intractable disputes, such as the neuronal basis of conscious awareness. This article describes and discusses the origins and nature of TMS, its applications and limitations, and its especial usefulness in conjunction with other techniques of evaluating or imaging brain activity.

Figures

Figure 1

On the left is Ferrier's (1886) drawing of the lateral surface of the left hemisphere of a macaque monkey with 15 numbered areas where stimulation produced particular movements. On the right is his drawing of the human brain showing where the ‘same’ functional areas would be. He proved to be close to the mark with most of them.

Figure 2

(a) Diagram of the circuit of a simple magnetic stimulator. (b) Outline of a simple round stimulating coil. The arrows indicate the direction of current flow in the coil. To its right is shown a schematic of the induced electric field in the tissue directly beneath the coil, which has been moved sideways for clarity. The arrows show the direction of flow of induced current, which is opposite to that in the stimulating coil. The length of the arrows crudely indicates the magnitude of the induced current, which declines towards the centre of the coil and away from its perimeter. (c) Schematic of the induced current in the brain beneath a figure-of-eight coil. The maximum current is in the vicinity of the ‘virtual cathode’, shown by the asterisk. (d) The time-course of the magnetic field and induced electric field waveforms (solid and dotted lines, respectively) beneath the centre of a simple round coil simulating coil driven by a standard rate Magstim 200 machine. (e) The same when driven by a Magstim Rapid stimulator. (a), (d) and (e) from Barker (1999). (b), (c) from Cowey & Walsh (2001).

Figure 3

(a) From models of TMS-induced electrical fields, one can infer the region of stimulation by studying the behavioural effects of closely spaced coil positions. In the schematic illustrated, if only the central position yields a particular effect, despite the fact that the induced fields must overlap, the effective locus must be beneath the centre (from Walsh & Cowey 2000, with permission). (b) An axial image through the head on which the positions of the bridge of the nose and the entrance to the ear canals are marked and can be tracked by the Polaris/Brainsight system if the head moves. The TMS coil is simultaneously tracked, allowing the stimulation to be delivered to a known anatomical region, for example, visual area V5/MT, which is marked on the image at the bottom right.

Figure 4

To the left are eight positions (A–H, at 1 cm intervals) on the scalp, all 1 cm to the left of the midline and starting 2 cm above the inion. On the right are shown the phosphenes drawn by the subject (AC) with respect to the central fixation point. Stimulation was 10 Hz for 0.5 s at 75% of maximum output of a Magstim Rapid stimulator and using a 70 mm figure-of-eight coil. When the same stimulation was applied over the other hemisphere at the point marked MT, the resulting speckled and moving phosphene is shown to the lower right of the fixation point. All the phosphenes were reproducible and retinotopically consistent.

Figure 5

(a) Visual suppression curves of three subjects. The proportion of correct identifications of three dark letters briefly flashed on a bright background is plotted as a function of the delay between visual stimulus onset and the subsequent application of a TMS pulse delivered over the primary visual cortex with a simple round coil. (From Amassian et al. 1989, with permission.) (b) Identification of a single letter chosen at random from a possible five letters. The letter was present for approximately 4 ms at the fixation point. The graph shows the proportion of letters correctly identified as a function of the delay for a subject and shows three dips in performance as well as the blink artefact when the single TMS pulse was delivered 70 ms before the letter. Each point is the mean of 30 trials and error bars denote the s.e.m. (Redrawn from Corthout et al. 1999a,.)

Figure 6

(a) Simplified example of the difference between parallel (pop-out) and serial (conjunction) search. In practice, many more items are displayed. (b–d) Mean normalized response time (±s.e.) of five subjects when a single pulse of TMS was applied over the right parietal cortex at various delays after the stimulus was presented. For each subject, the average response for each delay was normalized to the no-TMS condition. (b) There was no effect of TMS on reaction times in the pop-out task. (c) The solid line shows that stimulation around l00 ms after presenting the display during a conjunction task resulted in significantly longer RTs when the target was present (compared with the no-stimulation condition, dotted line). Other TMS delays had no significant effect. The broken line shows that TMS on the conjunction task, when the target was absent, significantly increased reaction times when applied 160 ms after the stimulus onset. (d) After over-training on the task, which renders TMS ineffective, the effect was restored when new conjunction stimuli were introduced.

Figure 7

Effects of viewing distance and parietal TMS on neglect. The results on the right, top, show the proportion of ‘right side is longer’ responses for bisected lines in near space (unfilled triangles) and far space (unfilled circles) with normal subjects. Error bars show ±1 s.d. Significantly fewer of the responses were ‘right-is-longer’, which demonstrates the well known right pseudo-neglect. The difference in the subjects' responses at the two viewing distances was significant (*p<0.05). Below, right, are shown the effects of stimulating the right PPC on the same subjects' responses. They are again normalized and show significant shifts in the perceived midpoint during TMS; in near space the right side was now perceived as longer, filled symbols (*p<0.05).

Figure 8

Examples of first-order (a) and second-order (c) global motion using flickering random dots. In (a), the tokens (small clusters of pixels) differ in mean luminance from the surround. In (c), the tokens differ from the surround in contrast but not in mean luminance. Patient RA was impaired in his affected hemifield on the first-order task (b), whereas patient FD was impaired in his affected hemifield on the second-order task (d). Their cortical lesions, established by MRI, are shown on a lateral view (FD) and coronal view (RA). The dots on the lateral surface of FD indicate the likely position of area MT. Abbreviations: cs, central sulcus; syl, Sylvian sulcus; sts, superior temporal sulcus.

Figure 9

Six normal subjects were tested on the same tasks presented to patients FD and RA and shown in figure 8. TMS delivered above area V5/MT or V2/V3 impaired performance on both tasks. Stimulation over the vertex had no significant effect.

Figure 10

(a) The temporal sequence of events in the priming task where the subjects had to judge the direction of global motion. Subjects were presented with a fixation spot for 500 ms followed by four virtual squares each containing prominent random dots moving in the same direction in three squares and a different direction in the fourth. The stimuli were presented for 144 ms followed by a blank screen until the subject responded. Response was followed by a further 500 ms of blank screen after which rTMS was applied for 500 ms before the fixation point for the next trial. (b) The priming effect is abolished by TMS delivered over the motion area V5/MT. (Modified from Campana et al. (2002).)

Figure 11

Examples of a mental rotation task where the subject has to judge which of the images cannot be fitted to all the others. In a similar, but not identical task, Klimesch et al. (2003) showed that rTMS above the medial prefrontal or parietal cortex at a frequency just above each subject's alpha frequency improved performance on the task.

Figure 12

Phosphenes elicited by medial occipital TMS (a) in a normally sighted observer (b) in retinally blind subject PS, and (c) in hemianopic patient GY. The coordinates give the site of stimulation in dorsal-lateral order. For example, 2,1 indicates that the coil was centred 2 cm rostral to the inion and 1 cm lateral to the midline. Note that as the coil is moved rostrally from the inion, the phosphenes migrate inferiorly and that as the coil is moved away from the midline the phosphenes migrate farther into the contralateral visual field. In subject PS, the phosphenes remain resolutely in the central few degrees of the visual field despite stimulation being delivered between 2 and 5 cm above the inion and up to 2 cm lateral. Moving phosphenes are shown in the three right-hand figures, (a) in a normally sighted observer, (b) in retinally peripherally blind subject PS, and (c) in cortically hemianopic patient GY with TMS in his undamaged hemisphere. All three subjects reported moving phosphenes. Reproduced from Cowey & Walsh (2000), with permission.

Figure 13

The effects of stimulating the motion area V5/MT in the ‘blind’ hemisphere of subject GY on his blindsighted discrimination of moving versus stationary random dot arrays (TMS 10 Hz, 0.6 s, 0.5 s stimulus display). At all three speeds (4, 32 and 64° s−1) he was impaired by the TMS above V5/MT but not at a control sight. The dashed line indicates chance performance out of 70 trials and at the two lower speeds he did not differ significantly from chance in the TMS condition.

Figure 14

The positions of phosphenes produced in a normal subject (AC) and a retinally blind subject (PS) by rTMS delivered over the scalp positions shown to the left of the plots of the phosphenes. Phosphenes were readily elicited in both subjects, although those in PS were much more confined to the centre of personal space.

Figure 15

When event-related potentials were recorded over the entire skull, there was no evoked response to auditory stimuli in the caudal occipital region of the sighted subject AC (dashed line), but there was a clear and significant response in the same region in the retinally blind subject (PS), corresponding to his synaesthetic experience of visual phosphenes in his central visual field in response to sudden and meaningful sounds (thick black line). The thin black line shows a smaller occipital response to sounds that failed to produce a phosphene. At the top are shown the evoked responses in auditory cortex recorded at position CZ. The symbols OZ, O1 and O2 refer to standard occipital EEG positions over visual cortex at the back of the head.

Figure 16

(a) schematic of the experimental design of the V5–Vl interaction study. The brain MRI image from one of the subjects displays the site of stimulation for induction of stationary (VI) and moving (V5/MT) phosphenes. The location on the subject's scalp of the centre of the figure-of-eight coil is superimposed on the subject's brain as reconstructed from an anatomical MRI. (b) Mean responses of all eight subjects to combined stimulation of V5 and Vl. The V5–Vl TMS asynchrony is on the x-axis: negative values indicate that VI was stimulated before V5, and positive values indicate that VI was stimulated after V5. The subjects made one of four judgements on each trial. (i) A phosphene elicited by V5 TMS was present and moving, (ii) a phosphene was present, but the subject could not decide whether it moved, (iii) the phosphene was present but stationary, (iv) no phosphene was experienced. TMS over VI between 10 and 30 ms after TMS over V5 affected the perception of the phosphene (see text for more details).