Studies in Cognition: The Problems Solved and Created by

Studies in Cognition The Problems Solved and Createdby Transcranial Magnetic Stimulation

E M Robertson H Theoret and A Pascual-Leone

Abstract

amp The application of transcranial magnetic stimulation(TMS) to investigate important questions in cognitiveneuroscience has increased considerably in the last fewyears TMS can provide substantial insights into the natureand the chronometry of the computations performed byspecific cortical areas during various aspects of cognition

However the use of TMS in cognitive studies has manypotential perils and pitfalls Although TMS can help bridgethe gap between psychological models and brain-basedarguments of cognitive functions hypothesis-driven carefullydesigned experiments that acknowledge the current limita-tions of TMS are critical amp

INTRODUCTION

A Problem Found

The history of science reveals a recurring pattern oftechnological advance preceding and supporting a pe-riod of scientific inquiry (Crump 2001) This conceptcan be applied to the science of cognition currentlydominated by techniques relying upon secondary meas-ures of brain activity and implicating rather than dem-onstrating the critical contribution of a brain area tobehavior Most neuroimaging methods lsquolsquocorrelatersquorsquo withvarying degrees of spatial and temporal accuracy anetwork of brain areas to a cognitive task This is trueof techniques that use reasonably direct measures ofbrain activity such as event-related potentials and ofthose that rely on less direct measures such as bloodflow or oxygen consumptionmdashfor example functionalmagnetic resonance imaging Consequently despiteelegant and ingenious experimental designs thesemethods offer little or no insight into whether a brainregion has a pivotal or merely subsidiary role in shapingbehavior

The interpretative problems of many functional imag-ing studies can be even greater because brain areasmaking a critical contribution may show no discerniblechange in activity during a task This problem of falsenegatives contributed to the widespread notion that theprefrontal cortex had a limited role in sequence learning(Clegg DiGirolamo amp Keele 1998) A subject can learna sequence of finger movements with and withoutawareness for the underlying pattern A variety ofbehavioral and imaging paradigms have sought to mapthese different types of sequence learning onto distinct

neuronal circuits (eg Grafton Hazeltine amp Ivry 1995)Generally these studies showed recruitment of theprefrontal cortex only when awareness for the sequencehad been achieved leading to the concept that theprefrontal cortex made no contribution to sequencelearning except when awareness had developed (Clegget al 1998) However studies with transcranial mag-netic stimulation (TMS) and on patients with focallesions suggest a critical role for the prefrontal cortexeven in the absence of awareness (Robertson TormosMaeda amp Pascual-Leone 2001 Gomez-Beldarrain Graf-man Pascual-Leone amp Garcia-Monco 1999 Pascual-Leone Wassermann Grafman amp Hallett 1996) Thisdiscrepancy between imaging and disruptive experi-ments appears to have now been clarified thanks toan elegant imaging study showing that the prefrontalcortex is recruited during all types of sequence learning(Willingham Salidis amp Gabrieli 2002) This exampleillustrates how major shortcomings of one techniquein this case the false negatives of functional imagingcan be overcome by the application of a complementarytechnique such as TMS (Fitzpatrick amp Rothman 2002)Traditionally the interpretative gap between correlationand causation has been bridged by carefully observingthe behavioral consequences of focal brain damage inpatients However the relationship between a behav-ioral impairment and the site of damage is at bestuncertain Behavioral changes following an insult reflectthe capacity of the rest of the brain to compensate (Kolbamp Wishaw 1998) Hence no simple relationship can beimplied between a brain area and an aspect of behaviorIn addition the inherent plasticity of the brain causesbehavioral impairments to evolve through time Suchchange is the product of a complex interplay betweenproperties of the brain and the natural history of theBeth Israel Deaconess Medical Center

copy 2003 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 157 pp 948ndash960

disease Furthermore damage is seldom well demarcat-ed or completely isolated and its location often reflectsthe vagaries of the cerebrovascular system rather thanconforming neatly to the needs of the cognitive scien-tist The fog surrounding these interpretative dilemmasbecomes even thicker with the effects of long-standingmedications

An ability to directly inquire about the causal contri-bution of different brain areas to behavior is greatlyneeded Recent years have seen this need partly as-suaged by TMS (Walsh amp Pascual-Leone 2003)

A Problem Lost

Moving beyond a merely correlative description of therelationship between brain and behavior is the freshapproach offered by TMS It allows the noninvasiveelectromagnetic stimulation of cortical sites This prin-ciple was demonstrated almost 20 years ago with invol-untary finger movement elicited by stimulation over themotor cortex (Barker Jalinous amp Freeston 1985) Themagnetic field produced around a coil can pass readilyacross the scalp and skull and induces an electricalcurrent within the brain tissue This occurs with minimalattenuation of the magnetic field Consequently signif-icant currents can be induced without having to applysubstantial voltages across the skull minimizing theactivation of pain fibers Early studies used single-pulseTMS to the occipital cortex time-locked to the presenta-tion of a visual stimulus to induce errors in the detectionof letters (Amassian et al 1989) These errors weremaximal with TMS applied between 80 and 100 msecfollowing the presentation of the visual stimulus imply-

ing that only at these times was the occipital cortexmaking a critical contribution to letter recognition(Figure 1) Similar approaches were employed to studythe role of the motor cortex in finger movements (Dayet al 1989) and of the somatosensory cortex in tactileperception (Cohen et al 1997)

This type of experiment can provide insight intowhen a given brain area is making a critical contribu-tion to a behavior In addition by applying single TMSpulses separated by a variable interstimulus interval totwo different brain areas the technique can be expand-ed to improve our understanding of the dynamicinteraction between brain areas This double-pulseparadigm has been successfully applied to explore therole of back-projections in visual awareness (Figure 1Pascual-Leone amp Walsh 2001)

One major limitation of these single- or paired-pulseTMS paradigms is that there is often insufficient infor-mation about the time at which an area makes a criticalcontribution to a given behavior TMS needs to bedelivered correctly in time and space for the verytransient disruption of brain function induced by single-pulse TMS to have a measurable behavioral effectTherefore the possibility of covering a larger timewindow with repetitive TMS (rTMS) to first explorethe lsquolsquospacersquorsquo dimension has some utility rTMS allowsthe site of stimulation to be established while latersingle-pulse experiments can give insight into thechronometry of this critical contribution to behavior(eg the role of the occipital cortex during Braillereading Figure 2) The combined use of functionalneuroimaging to provide spatial information evokedpotentials to provide chronometric information and

Figure 1 Probing chronome-try with TMS (A) A TMS pulseto the occipital probe can sup-press visual perception when itis applied between 80 and 120msec after stimulus presenta-tion (from Cracco Maccabee ampAmassian 1999) (B) The per-ception of a TMS-induced mov-ing phosphene after stimulationof motion area V5 can besignificantly suppressed by asecond TMS pulse applied tothe primary visual cortex (V1)10 to 40 msec later These datashow that fast V5ndashV1 feedbackprojections are necessary forawareness of motion (fromPascual-Leone amp Walsh 2001)

Robertson Theoret and Pascual-Leone 949

TMS to address causality may become a powerfulfuture strategy

However as with any experimental technique TMS itis not without problems Here we try to explore some ofthese pitfalls and along the way point out what has andwhat can be achieved with TMS in cognitive studies

Discomfort and Distraction

TMS is commonly described as a lsquolsquorelatively painlessmethodrsquorsquo of stimulating the brain noninvasively None-theless TMS causes significant sensory sensation that cannonspecifically interfere with task performance Thisstarts with the loud clicking sound as the stimulator isdischarged and continues with stimulation of cranialnerves and the direct activation of facial and neckmusculature These effects are particularly prominentwhen TMS targets the frontal temporal and occipitalregions (ie the brain areas most commonly targeted incognitive studies) Several approaches can be used to tryto ensure that changes in performance are attributablespecifically to the effects that TMS has upon the brain

Control Sites and Control Tasks

One approach is to stimulate at several sites If theeffects of stimulation are observed exclusively at onesite then this gives some reassurance that the differ-ences across sites are due to the specific effects ofneuronal disruption This comparison across sites as-sumes that the nonspecific effects of TMS are equivalent

However even relatively small changes in position cancause substantial changes in the sensory effects ofstimulation Consequently many studies have also takenthe approach of observing behavior across several dis-tinct tasks following stimulation at one site (eg Beckersamp Zeki 1995)

Sham Stimulation

An alternative strategy is to use sham stimulation In thiscase although the stimulator discharges giving an audi-ble clicking sound the magnetic pulse does not traverseacross the skull This is achieved by using speciallydesigned sham stimulation coils or tilting the coil awayfrom the scalp The latter is certainly a less desirablealternative and in some instances may in fact have asimilar effect on the brain as real TMS (Lisanby GutmanLuber Schroeder amp Sackeim 2001) However currentlyavailable sham coils need to be improved because theyfail to truly mimic the peripheral sensations associatedwith TMS Without these sensations it easily becomesobvious to all but the most naotildeve subject that they areonly receiving sham stimulation

Timing

A further approach to give experimental control whichcan only be used with single-pulse TMS or short trains oflsquolsquoon-linersquorsquo rTMS is to vary the delay between a particularevent (eg the presentation of a stimulus) and stimula-tion The assumption is that nonspecific distracting

Figure 2 Repetitive versussingle-pulse TMS in the estab-lishment of causality and itstiming (A) Short trains of re-petitive TMS (10 Hz 3 sec) tooccipital cortex disrupted tactileperception in early blind sub-jects (EBB) but not in sightedvolunteers (SVR Cohen et al1997) (B) Later it was shownthat in blind subjects TMS tothe sensorimotor cortex signifi-cantly reduces tactile detectionwhen it is applied 20 msec afterstimulus presentation whereasTMS to the occipital corteximpairs tactile identificationwhen it is applied 60 msecafter stimulus presentation(Hamilton amp Pascual-Leone1998) These studies illustratethat rTMS can reveal thoseareas making a criticalcontribution to an aspect ofcognition whereas single pulseallows the time at which thiscontribution is made to bedetermined

950 Journal of Cognitive Neuroscience Volume 15 Number 7

effects of TMS will be independent whereas the behav-ioral effects will be highly dependent on the preciseinterval between the event and the stimulation Thisassumption may not be valid in all circumstances Whena stimulus and a pulse are delivered simultaneously thedistraction may be greater than when there is an intervalbetween the stimulus and the TMS pulse An elegantexample of this approach is the disruption caused to theco-ordination between saccadic eye movements andthose of the hand at a specific time following saccadeonset (van Donkelaar Less amp Drew 2000)

lsquolsquoOFF-LINErsquorsquo rTMS

A dramatic finding from the early studies using rTMS wasthe arrest of speech with high-frequency stimulation(Pascual-Leone Gates amp Dhuna 1991) Later studies useda similar design applying high-frequency stimulationwhile performing a task This is the so-called on-lineparadigm where both stimulation and task performanceoccur concurrently (Figure 3) Presumably the higher therTMS frequency the greater the disruptionof the targetedbrain region and the greater the behavioral effects How-ever the greater the potential risks and the more promi-nent the nonspecific behavioral and attentional effectscan makeresultsdifficult to interpret (Wassermann1998)

An exciting approach which has achieved some pop-ularity over the last few years is to stimulate at a site ofinterest for 5 to 10 min at 1 Hz lsquolsquobeforersquorsquo starting acognitive task (Figure 3 Table 1) This lsquolsquooff-linersquorsquo stimu-lation elegantly removes many of the nonspecific con-current effects of TMS Applied initially to investigatevisual imagery this technique has since been used

across a variety of cognitive tasks (Walsh amp Cowey2000 Kosslyn et al 1999) These studies have consis-tently demonstrated that the specific effects of rTMSupon behavior and presumably cognition outlast theinitial block of stimulation (Table 1) The neurophysio-logic bases for these long-lasting disruptive effects areuncertain Reflecting this uncertainty are the manyquestions regarding the rTMS parameters necessary toachieve a desired modulation of cortical activity

TMS PARAMETERS WHEN AND HOWTO STIMULATE

Selecting the frequency intensity and duration of stim-ulation are difficult and often arbitrary decisions unfor-tunately these can often determine the success orfailure of a study Until more objective ways of selectingparameters become available a pragmatic and adaptableapproach should be taken The following paragraphsexplore some of the pertinent issues surrounding theselection of TMS parameters

Intensity of Stimulation

Determining the intensity of stimulation required to testa hypothesis (ie interfere with a particular function at aspecific site) is a substantial problem At least two factorsinfluence the susceptibility of a brain area to stimulationmagnetic field strength and excitability of the cortex Thestrength of the magnetic field produced by a coil de-creases exponentially with distance Hence the depthtraveled from the center of the coil to the cortex largelydetermines the magnitude of the magnetic field to which

Figure 3 Applying rTMS whileperforming a task so-calledon-line stimulation (A) hasbeen a widely used approachyet it suffers from the potentialthat the nonspecific effects ofconcurrent stimulation canadversely affect performanceConsequently recent years haveseen the emergence of anlsquolsquooff-linersquorsquo paradigm (B) in whichstimulation and performanceare dissociated A new variant ofthis paradigm in which a beha-vior precedes stimulation (C)has very recently been used tostudy the cortical areas involvedin consolidation of motormemories (Muellbacher et al2002)


an area of cortex is exposed (McConnell et al 2001Kozel et al 2000) However susceptibility to stimulationdoes not merely reflect the depth of a cortical site It alsodepends upon the inherent excitability of the cortexitself which varies with the specific cortical site beingstimulated and the behavioral task being performed

In the motor cortex excitability is quantifiable Stimu-lation produces a measurable muscle twitch (Rothwell1997) Although not capable of producing physical move-ment even slight changes of electrical activity in a musclecan be measured using electromyography the loweststimulation intensity capable of producing such changesis called the lsquolsquomotor thresholdrsquorsquo (MT Rossini et al 1994)Consequently stimulation intensities can be normalizedacross subjects by using multiples of MT This methodtakes into account interindividual differences that maymodulate the efficacy of magnetic stimulation and pro-vides a composite measure related to the depth of stim-ulation and cortico-spinal excitability Using thisstandardized method may allow comparison across ex-perimental paradigms (Stewart Walsh amp Rothwell 2001)

Stimulation over nonmotor areas often does not pro-duce as readily an objective quantifiable response Con-sequently many studies have used MT as a surrogatemarker for excitability across all cortical areas Yet expect-ing to gain insight into the excitability of nonmotor areas

from stimulating the primary motor cortex may be afoolrsquos errand The underlying assumption that the effectsof TMS across cortical areas are correlated to one anotherappears to be wrong For example accompanying thestimulation of the visual cortex in some individuals isthe sensation of light called phosphenes (Cowey ampWalsh 2001) In the same manner that an MT can bedetermined phosphene thresholds (PTs) can be estab-lished in individual subjects (Boroojerdi Prager Muell-bacher amp Cohen 2000) PT has been shown to be stableover time within individuals (Stewart et al 2001) How-ever there appears to be no intraindividual correlationbetween PT and MT (Figure 4 Stewart et al 2001) Thesedifferences may be partly explained by differences inskull to cortex distances between motor and visualcortices Nonetheless studies have suggested that thescalp to cortex distance at different sites are correlated(McConnell et al 2001 Kozel et al 2000) Hence thisfactor alone seems unlikely to have broken any potentialcorrelation between the MT and the PT

Therefore MT cannot be used to gauge the biologicaleffects of TMS in cortical areas other than the motorcortex In fact the situation is even more complexbecause even within a single brain area the stimulationintensity that is required to cause cortical disruptionvaries with the behavioral task For example TMS

Table 1 Studies that Have Used an lsquolsquoOff-Linersquorsquo Paradigm to Investigate Particular Aspects of Cognition

Authors Domain AreaFrequency

(Hz)Train Duration

(min) IntensityEffect

Duration

Kosslyn et al 1999 Visual imageryperception

Area 17 1 10 90 MT NA

drsquoAlfonso van HonkHermans Postma ampde Haan 2000

Attention toangry faces

PFC 06 15 130 MT At least 10 min

Hilgetag et al 2001 Spatial attention PPC 1 10 90 MT At least 5 min

Robertson et al 2001 Motor learning DLPFCand PPC

1 5 115 MT At least 5 min

Theoret Haque ampPascual-Leone 2001

Paced fingertapping

Cerebellum 1 5 90 MT At least 5 min

Lewald et al 2002 Spatial hearing PPC 1 10 60 stim output ordm11 min

Mottaghy et al 2002 Visual workingmemory

DLPFC DMPFCVPFC

1 10 90 MT ordm5 min

Sack et al 2002 Visuospatialfunctions

Parietal1

10 80 MT NA

Shapiro Pascual-LeoneMottaghy Gangitanoamp Caramazza 2001

Grammaticalprocessing

Inferior mid-frontalgyrus

1 5 110 MT NA

All these studies used the same paradigm Studies using off-line TMS with low-frequency stimulation to study cognitive functions are based on theobserved decrease in excitability over the primary motor cortex outlasting a block of rTMS (Chen et al 1997) It is assumed that rTMS-inducedreduction in cortical excitability will similarly apply to non-motor cortical areas allowing us to target different cortical sites to investigate differentcognitive domains

DLPFC = dorsolateral prefrontal cortex DMPFC = dorsomedial prefrontal cortex MT= motor threshold PFC = prefrontal cortex PPC = posteriorparietal cortex VLPFC = ventrolateral prefrontal cortex


of visual cortical area V5 can induce phosphenes (fre-quently kinetic or moving phosphenes Pascual-Leone ampWalsh 2001) and can disrupt motion after-effects (MAETheoret Kobayashi Ganis Di Capua amp Pascual-Leone2002) Repeated determinations of either one of thesemeasures reveal reasonable within-subject reproducibil-ity but there is no intraindividual correlation of V5-PTsand V5-induced disruption of MAE (Figure 5) Can somebetter way be found to define stimulation intensity

Fixed Intensity

Using MT to determine the intensity of TMS over non-motor areas is arbitrary Consequently it might be aswell to use a fixed intensity defined by the stimulatoroutput This approach reduces the experiment durationand limits the number of TMS pulses Some studies haveused such a method (eg Lewald Foltys amp Topper2002 Corthout Uttl Walsh Hallett amp Cowey 1999

Beckers amp Zeki 1995) which should be similar to theMT technique As in the MT approach it is likely that forsome subjects the fixed intensity will be below thatcapable of inducing a behavioral effect giving addedvariability to the overall results

Intensity Corrected for Scalp to Brain Target Distance

The strength of the magnetic field produced by a coilfalls of exponentially with distance from its center sothat the depth of the cortical tissue largely determinesmagnetic (and induced electric) field strength at abrain site Consequently McConnell and collaborators(2001) suggest that the scalp to brain distance shouldbe taken into account when deciding upon stimulationintensity This method certainly would address oneimportant aspect of a brain arearsquos susceptibility tostimulation However a major weakness of this ap-proach is that it merely allows a direct calculation of

Figure 5 (A) A subjectrsquos PT measured on two separate days has a good correlation with one another (B) The PT however is poorly correlatedwith the reduction in MAE that results from applying rTMS over this region Thus the magnitude of behavioral change provoked by TMS is notrelated to cortical excitability as indexed simply by the phosphene threshold (from Theoret et al 2002) Nor is the behavioral change related to theintensity of stimulation which is determined by cortical excitability The MAE was defined as the duration of illusory motion that occurred followingadaptation to a 30-sec moving stimulus

Figure 4 Both a subjectrsquos MT (A) and PT (B) correlate between sessions Despite these consistent measures of cortical excitability there is nocorrelation between MT and PT within individual subjects (C modified from Stewart et al 2001) This implies that cortical excitability at one site isnot a good predictor of excitability at another site Thus the rationale of expressing stimulation intensity outside the motor cortex as a proportionof MT is doubtful


the strength of the magnetic field to which an area ofcortex is exposed but ignores the differential suscep-tibility of brain areas to a given stimulation intensitydepending on individual variations in anatomy andtask-related activation

Intensity Tuning Curve

In the motor cortex increasing TMS intensity results inincreased MEP amplitudes hence an inputndashouputcurve can be generated (Chen 2000) A similar proce-dure involving a set of increasing intensities whileperforming a given task would allow a stimulusndashresponse curve design to be applied to the study ofcognition If nonmotor areas behave in a similar fashionto the primary motor cortex the behavioral effects ofTMS should be modulated by different stimulationintensities Hence a parametric analysis could be ap-plied correlating different intensities of stimulationwith behavioral consequences

Behavioral Determination

Some authors have titrated the intensity of TMS for agiven experiment by using the robust behavioral effectsof stimulation on a different task In a study of visualcortex excitability Mulleners Chronicle Palmer Koeh-ler and Vredeveld (2001) applied TMS at intensities atwhich lsquolsquosubjects were unable to identify at least two ofthe three target letters correctly in the order presentedrsquorsquoIn this manner stimulation intensities were defined by abehavioral effect of TMS Similarly Rushworth Ellisonand Walsh (2001) used the disruptive effects of rTMS ona visual search task Behavioral determination of TMSintensities in cognitive studies has the advantage oftaking into account the attentional and perceptual ca-pabilities of individual subjects at the time of the exper-iment However the effects of TMS on one task may notnecessarily correlate with its effects on another distincttask (eg Figure 5)

Functional Determination

The effects of TMS on nonmotor areas can potentially beobjectively quantified using imaging techniques A num-ber of recent studies have elegantly combined functionalimaging techniques with TMS to show highly significantchanges in cortical blood flow during and followingstimulation (Paus et al 1997 Paus 2002 Bohninget al 1999) Blood flow changes within the primarymotor cortex are directly related to the intensity ofstimulation (Siebner et al 2001) Consequently thesechanges could be used as a surrogate marker of thecortical effects of TMS in areas where it is currentlyimpossible to make such measurements A similar argu-ment can also be made for the utility of electroenceph-alography (EEG) monitoring For example Paus Sipila

and Strafella (2001) have shown that single-pulse TMSapplied over the motor cortex is associated with apositive (P30) and two negative (N45 N100) scalppotentials The amplitude of the N45 component iscorrelated with the intensity of the TMS pulse andappears to be generated in the primary motor cortexTMS-induced scalp potentials could be used to measurethe effects of magnetic stimulation on nonmotor corticalareas However the relationship between a cognitivetask its potential disruption and changes in either bloodflow oxygenation or scalp potentials is uncertain

Frequency of Stimulation

In single-pulse TMS studies the interval between pulseshas to be sufficiently long to prevent interactions be-tween consecutive pulses An interval of approximately7 sec between pulses may be sufficient No studyhowever has systematically investigated this issue rTMSwhen applied in an lsquolsquoon-linersquorsquo paradigm (Figure 3) exertsgreater disruption of a targeted brain area the higher thestimulation frequency However when used in an lsquolsquooff-linersquorsquo design rTMS frequency appears to determine theneurophysiologic effects In most instances while slow-frequency rTMS (micro1 Hz) decreases cortical excitabilityhigh-frequency stimulation (para5 Hz) increases excitability(Figure 6 Maeda Keenan Tormos Topka amp Pascual-Leone 2000 Berardelli et al 1998 Pascual-LeoneGrafman amp Hallett 1994) Following the logic thatsuppression of cortical activity within a specific corticaltarget can significantly impair performance (Table 1)an increased excitability could perhaps lead to a behav-ioral lsquolsquoimprovementrsquorsquo Yet we know of no study thathas used an lsquolsquooff-linersquorsquo high-frequency rTMS (ie gt1 Hz)paradigm to convincingly demonstrate enhancedperformance

In contrast a few studies applying lsquolsquoon-linersquorsquo high-frequency rTMS in short trains have actually shown im-provements in behavior Trains applied over Wernickersquos

Figure 6 The effect of rTMS on motor cortex excitability (from Maedaet al 2000) Although as a general rule slow rTMS decreases and rapidrTMS (gt1 Hz) increases cortical excitability there can be greatintersubject variability Some subjects show the opposite pattern (A)but when all participants are averaged (B) the general rule holds true


area gave a response time advantage for processing black-and-white drawings (naming Mottaghy et al 1999 andreasoning Boroojerdi et al 2001) Similar performanceimprovements have been elicited by single-pulse stimula-tion in both the healthy (Grosbras amp Paus 2002 TopperMottaghy Brugmann Noth amp Huber 1998) and diseasedbrain (Oliveri et al 1999) The precise neurologic mech-anisms underlying such behavioral improvements remainat best uncertain Potentially TMS may induce a paradox-ical functional facilitation where behavioral improvementis due to the disruption of a brain region that normallyexerts inhibitory influences on distant brain areas (Kapur1996) Similar behavioral gains have been found in studiesapplying lsquolsquooff-linersquorsquo 1-Hz rTMS presumably inhibiting thetargeted brain region and releasing distant brain areas(Hilgetag Theoret amp Pascual-Leone 2001)

Duration of Stimulation

Lengthening the duration of lsquolsquoon-linersquorsquo rTMS is pre-sumed to cause more disruption by virtue of temporalsummation of the effects of the stimulation Howeverlonger trains at high stimulation frequencies are increas-ingly risky as a seizure can be induced (Pascual-LeoneValls-Sole Wassermann amp Hallett 1994) Therefore theduration of the lsquolsquoon-linersquorsquo rTMS trains is primarily limitedby current safety guidelines aimed at minimizing the riskof side effects (Wassermann 1998) These and therelevant national guidelines should be considered man-datory reading for anyone contemplating using TMS(Hallett Wassermann Pascual-Leone amp Valls-Sole1999 Wassermann 1998 Pascual-Leone et al 1993) Itis unclear whether for safety or scientific reasons thereshould be a limit to the duration of single-pulse TMSstudies No study has addressed the question of thenumber of single TMS pulses that can be applied in agiven session

The duration of lsquolsquooff-linersquorsquo rTMS in cognitive studieshas been based on data obtained from the motor cortexThese seem to indicate that longer trains induce longer-lasting and more robust effects For example stimulatingthe primary motor cortex for 4 min at 1 Hz did notsignificantly reduce cortical excitability but increasingtrain duration caused a reduction in cortical excitability(Maeda et al 2000) However no systematic parametricstudy is available and consequently determination oftrain duration has been mostly arbitrary across studies(Table 1) The confusion is further increased by theuncertainty about whether these observations can beappropriately extrapolated to nonmotor areas

Duration of the Stimulation Effect

Little is known about the duration of the effects neuro-physiologic or behavioral of single TMS pulses or rTMStrains It is presumed that single TMS pulses exert a verylimited effect of around 100ndash500 msec After approxi-

mately 200 msec for example the function of theparietal lobe returns to normal allowing reaching move-ments to compensate for changes in target position(Desmurget et al 1999) The behavioral effects of lsquolsquoon-linersquorsquo rTMS are generally assumed to be limited to theduration of the rTMS train itself This is howeverunlikely given the experience with lsquolsquooff-linersquorsquo rTMSPresumably the behavioral effects during lsquolsquoon-linersquorsquorTMS appear more dramatic but subtle neurophysiolog-ic and behavioral consequences of stimulation probablyoutlast the rTMS train These lasting rTMS effects con-stitute the basis for the lsquolsquooff-linersquorsquo rTMS paradigmsNeurophysiologic studies (eg EEG studies during sin-gle-pulse or lsquolsquoon-linersquorsquo rTMS) should be able to providemore detailed insights into this important issue

In the motor cortex a 15-min train of rTMS atapproximately 1 Hz reduces cortical excitability for atleast the subsequent 15 min (Chen et al 1997) Twostudies have specifically addressed the question of howlong the behavioral effects can outlast the application ofrTMS In one study a 1-Hz rTMS 600 pulse train over theparietal cortex induced a shift in the lateralization ofinteraural time differences for at least 11 min (Table 1Lewald et al 2002) A similar approach also using a10-min 1-Hz train demonstrated visual working memoryimpairments following rTMS over prefrontal areas whichlasted only 5 min (Figure 7 Table 1 Mottaghy Gang-itano Sparing Krause amp Pascual-Leone 2002) Furtherbehavioral and neurophysiologic studies are criticallyneeded to gain further insights that can aid in the designof optimized experimental paradigms

Stimulation Parameters and Behavioral Task

The behavioral deficits induced by TMS do not simplydepend upon the selected stimulation parameters In-stead they are the products of an interaction between thedisruption caused by TMS to the targeted brain site theeffects to distant brain areas along functional connec-tions and the particular task being performed Hence acognitive task may be sufficiently trivial that despitesubstantive disruption to normal cortical function theremay be no observed behavioral impairment Similarly astimulation paradigm may produce a relatively subtledisruption to cortical function so that only a complexcognitive task would reveal any impairment With theinherent redundancy of the brain and its resulting highcapacity to compensate for disruption caused by TMS it isperhaps only through straining the available neuronalresources with a reasonably complex task that it becomespossible to observe behavioral impairment This relation-ship between task complexity cortical disruption andimpairment was demonstrated in a recent study exploringthe effects of rTMS of the parietal cortex (1 Hz 10 min) onvisual spatial attention (Hilgetag et al 2001) To controlfor interindividual differences in acuity and attention thevisual stimuli were adapted for each participant This


approach maximized the potential disruptive and facili-tative effects of rTMS on task performance

WHERE ARE YOU STIMULATING

One of the most substantial problems in the study ofcognition with TMS is relating the known stimulationsite in the overlying scalp with a particular brain area inother words the problem of anatomical localizationThe relationship between scalp position and a givenbrain area is variable across individuals hence theplacement of a TMS coil on the scalp according to bonylandmarks will necessarily introduce errors and interin-dividual variability in the targeted brain region (MeyerBritton Kloten Steinmetz amp Benecke 1991) A poten-tial solution to this problem is to define coil positionbased upon objective output parameters This is rela-tively easy to achieve over the primary motor cortexwhere magnetic stimulation can result in an overt re-sponse (muscle twitch) Similarly the induction ofphosphenes over the visual cortex can also be used toguide coil location (Cowey amp Walsh 2001) However incortical areas where no overt responses to TMS can beelicited appropriate placement of the coil can be asubstantially more complicated task

A possible approach is to locate brain areas relativeto those that have a reasonably certain position For

example the dorsolateral prefrontal cortex can bedefined as 5 cm anterior to the thumb representationover the primary motor cortex as measured using theTalairach atlas (Pascual-Leone Rubio Pallardo amp Cata-la 1996 Pascual-Leone Wassermann et al 1996)However when attempting to validate this procedureby comparing the Talairach position against the knownbrain anatomy from individual brain scans the finallocation of the coil relative to the underlying Brodmannrsquosarea was found to be quite variable (Herwig PadbergUnger Spitzer amp Schonfeldt-Lecuona 2001 Pascual-Leone Bartres-Faz amp Keenan 1999)

Another approach is to use a frameless stereotacticsystem to provide lsquolsquoon-linersquorsquo information about thelocation of the coil (Gugino et al 2001 Herwig et al2001 Paus 1999) A structural brain MRI is obtainedprior to the TMS session and is displayed on a com-puter monitor Sensors are attached to the stimulatingcoil and to the subjectrsquos head both monitored by aposition sensor This information is sent to a computerwhich after a calibration procedure displays the posi-tion and orientation of the coil on the MRI Image-guided TMS permits constant visualization of coilplacement in relation to the subjectrsquos brain Howeverthe assumption that the gross anatomical features ofthe cerebral cortex (eg mid-frontal gyrus) are relatedto its functional subdivisions (eg Brodmannrsquos areasie BA 46) is certainly questionable Fortunatelyframeless stereotaxy can also be used in conjunctionwith functional neuroimaging allowing activated sitesto be targeted for stimulation This reduces the influ-ence of interindividual anatomical variability and nolonger assumes a correspondence between anatomicallandmarks of the cerebral cortex and task-related func-tional activations

Although these approaches offer enhanced anatom-ical precision this does not necessarily translate intoproviding TMS studies of cognition with substantiallyimproved accuracy Even relatively conservative esti-mates suggest that single-pulse TMS effects corticaltissue over 1 cm from the center of the coil (WilsonThickbroom amp Mastaglia 1993 Brasil-Neto et al1992) Inevitably with rTMS this affected zone is likelyto be larger consequently the high precision offeredby navigational systems is swamped by a more substan-tial source of error which is inherent within thetechnique At this point we should view TMS asproviding a method to dissociate and explore aspectsof human behavior and cognition without necessarilygiving a particularly detailed account of the preciseanatomical zone effected by TMS For example it hasbeen possible to dissociate working memory systemsacross the prefrontal cortex with the dorsolateral areaproviding a critical contribution to spatial while non-spatial working memory was supported by the ventro-lateral area (Figure 7 Mottaghy et al 2002) Improvedmore focal TMS coils are needed to gain sufficient

Figure 7 Shows the effect upon working memory performance oflsquolsquooff-linersquorsquo rTMS at different sites over the prefrontal cortex (Dm =dorsomedial prefrontal cortex Dl = dorsolateral prefrontal cortexV = ventrolateral prefrontal cortex) Two distinct working memorytasks were used one involved recalling the position of a stimulus(spatial) while the other required recollection of a personrsquos face (face)The mean performance rates in percent with error bars (standard errorof the mean) are shown for each stimulation site and each workingmemory task (spatial and faces) at the four different time points(base = baseline post1 = immediately after rTMS post2 = 5 minafter rTMS post3 = 10 min after rTMS) Significant decreases inperformance ( p lt 05) following rTMS are marked () The moredorsal sites of stimulation cause impairment in spatial workingmemory In contrast it is the ventral sites which are responsible fornonspatial (eg facial) working memory suggesting a relativesegregation of spatial and nonspatial working memory across theprefrontal cortex (Mottaghy et al 2002)


spatial resolution to derive all the possible benefits ofusing a combination of frameless stereotaxy and TMSSuch coils are going to be essential when TMS is beingused to alter cortical excitability as is generally the casein cognitive studies

TMS can also be used to measure cortical excitabilityWhen applied in this context remarkable spatial reso-lution can be achieved with current devices For exam-ple while learning a sequence of finger movements aTMS study has shown that the representation of thehand over the primary motor cortex expands consider-ably until subjects become aware of the underlyingpattern (Pascual-Leone Grafman et al 1994) Whenawareness is achieved the hand representation sudden-ly contracts down to baseline levels Similar changeshave been observed in piano players and athletes(Pearce Thickbroom Byrnes amp Mastaglia 2000 Pasc-ual-Leone et al 1995) Being able to resolve thesechanges is a testament to the spatial accuracy of TMSwhen used to measure rather than alter cortical excit-ability These studies have been extended to includemeasurement of changes in intracortical excitabilityduring the acquisition of skills (Nordstrom amp Butler2002) providing insights into the neural mechanismssupporting procedural learning

WHAT ARE YOU STIMULATING

Each area of the brain is coupled through anatomicalconnections and projections with a vast number of otherareas Hence stimulating an area of the brain may havefunctional consequences not only at that site but through-out a neuronal circuit Potentially this makes the inter-pretation of any behavioral impairment fraught withdifficulty for it may represent the ability of the rest ofthe brain to compensate for disruption either within anarea or across a circuit In accord with this view arefunctional imaging studies demonstrating substantial ac-tivity changes even in brain areas distant from the actualsite of stimulation (Bohning et al 1999 Paus et al 1997)Even these studies may underestimate the circuit affectedby magnetic stimulation Projections arising from an areabeing stimulated are likely to be activated orthodromi-cally leading to an increase in metabolic demands in adistal site reflected in a rise in blood flow (Paus et al1997 Wong amp Moss 1992) In addition anatomicalprojections to this site are also likely to be activatedantidromically Although this activation may influencedistal sites it will be without a direct change in synapticactivity Consequently there may be neither changes inmetabolic demands nor blood flow Perhaps the differ-ential effects upon blood flow produced by orthodromicand antidromic stimulation may explain why a largernetwork of areas is not visualized (Wong amp Moss 1992)For example stimulation of the frontal eye field (Paus etal 1997) did not result in a significant increase in bloodflow to the dorsolateral prefrontal cortex as may have

been anticipated (Buttner amp Fuhry 1995 FunahashiBruce amp Goldman-Rakic 1991) Alternatively theseblood flow patterns may provide a description of thefunctional connectivity as opposed to an accurate ana-tomical description of the human brain Regardless anentire circuit not just a single brain area is effected bystimulation making the interpretation of behavioral ef-fects difficult

Cortical areas distant from the primary site of stimu-lation have shown not only blood flow changes but alsoalterations in excitability (Munchau Bloem IrlbacherTrimble amp Rothwell 2002 Gerschlager Siebner ampRothwell 2001) The primary motor cortex has a re-duced excitability following 20 min of 1-Hz stimulation at80 of MT over the premotor cortex These two areasare richly interconnected consequently it is not toosurprising that altered activity within the premotorcortex produces detectable changes within the primarymotor cortex Nonetheless these observations serve todemonstrate that rTMS does not merely affect theneuronal activity of a single site but rather a networkSuch distant effects may also influence subcortical struc-tures These problems are substantially reduced whenusing single pulses of TMS to disrupt cognition (Pascual-Leone Walsh amp Rothwell 2000)

Although these studies demonstrate the activation andpotential disruption of an entire circuit this is very farindeed from showing that distant effects are responsiblefor the observed behavioral effects of TMS It is quitepossible that the principal component responsible forbehavioral changes remains the primary area being stim-ulated with other more distant sites having at best only amarginal influence upon a particular behavior For exam-ple studies examining the contribution of the prefrontalcortex to working memory have shown quite a surprisingdegree of specificity (Mottaghy et al 2002) Stimulationover the dorsolateral region of the prefrontal cortexproduces specific deficits in spatial working memoryleaving verbal working memory relatively preservedStimulation over the ventrolateral prefrontal cortex re-sults in the opposite behavioral results These effects arespatially specific despite the interconnections amongareas of the prefrontal cortex However in some circum-stances the distant effects of TMS may best explain theobserved behavioral effects These explanations couldinclude the release of neurotransmitters from a distantsite (Strafella Paus Barrett amp Dagher 2001) or itsfunctional release from tonic inhibition following thedirect stimulation of a brain area (Hilgetag et al 2001Oliveri et al 1999) Interpreting the effects of TMS uponbehavior is a substantial problem We always run theconsiderable risk of offering a post hoc explanation forany pattern of behavior resulting from TMS by invoking acombination of both primary and distant site effects Wemay only be freed from this problem by using a hypoth-esis-driven approach What is true generally in science isalso true of TMS studies Paradigms should be set up


specifically to refute a particular concept or in an attemptto dissociate between possible contributions to behav-ior Although such a hypothesis-driven approach islaudable with such an immature discipline as cognitiveneuroscience there still needs to be room for a moredata-driven approach with some experiments beingtruly exploratory in nature In such studies the simplestexplanation should perhaps be given the greatest cre-dence whether this involves known aspects of theprimary or distant sites

CONCLUSIONS

We have described and where possible explained someof the perils and pitfalls of applying TMS in cognitivestudies It may appear to the unwary reader that we havebeen cheated In exchange for a single problem we nowhave many distinct problems each requiring a solutionHowever the single problem was one of determiningthe contribution made by a cortical area to an aspect ofbehavior by no means trivial and one which has plaguedmuch of contemporary cognitive neuroscience This hasbeen replaced admittedly with many more problemsbut these are simpler Given time many of these will besolved or shown not to be as problematic as we hadonce feared Some have already been solved or can atleast be ameliorated by using carefully designed experi-ments Nonetheless these difficulties should not have usturn away in despair Rather we need to increasinglyunderstand and meet the challenges set by integratingTMS into cognitive studies The alternative is to riskloosing a potentially unique opportunity to deepen ourunderstanding of human cognition

Acknowledgments

The financial support of the National Alliance for Research inSchizophrenia and Depression (EMR) the Canadian Institutesof Health Research (HT) and the Goldberg Foundation isgratefully acknowledged

Reprint requests should be sent to Alvaro Pascual-LeoneLaboratory for Magnetic Brain Stimulation Behavioral Neurol-ogy Unit Beth Israel Deaconess Medical Center 330 BrooklineAvenue Kirstein Building KS-452 Boston MA 02215 USA orvia e-mail apleonebidmcharvardedu

REFERENCES

Amassian V E Cracco R Q Maccabee P J Cracco J BRudell A amp Eberle L (1989) Suppression of visualperception by magnetic coil stimulation of humanoccipital cortex Electroencephalography and ClinicalNeurophysiology 74 458 ndash462

Barker A T Jalinous R amp Freeston I L (1985) Non-invasivemagnetic stimulation of human motor cortex Lancet 11106 ndash1107

Beckers G amp Zeki S (1995) The consequences ofinactivating areas V1 and V5 on visual motion perceptionBrain 118 49ndash60

Berardelli A Inghilleri M Rothwell J C Romeo S CurraA Gilio F Modugno N amp Manfredi M (1998) Facilitationof muscle evoked responses after repetitive corticalstimulation in man Experimental Brain Research122 79ndash84

Bohning D E Shastri A McConnell K A Nahas ZLorberbaum J P Roberts D R Teneback C VincentD J amp George M S (1999) A combined TMSfMRI study ofintensity-dependent TMS over motor cortex BiologicalPsychiatry 45 385 ndash394

Boroojerdi B Phipps M Kopylev L Wharton C M CohenL G amp Grafman J (2001) Enhancing analogic reasoningwith rTMS over the left prefrontal cortex Neurology 56526 ndash528

Boroojerdi B Prager A Muellbacher W amp Cohen L G(2000) Reduction of human visual cortex excitability using1-Hz transcranial magnetic stimulation Neurology 541529 ndash1531

Brasil-Neto J P Cohen L G Panizza M Nilsson J RothB J amp Hallett M (1992) Optimal focal transcranialmagnetic activation of the human motor cortex Effects ofcoil orientation shape of induced current pulse andstimulus intensity Journal of Clinical Neurophysiology 9132 ndash136

Buttner U amp Fuhry L (1995) Eye movements CurrentOpinion in Neurology 8 77ndash82

Chen R (2000) Studies of human motor physiology withtranscranial magnetic stimulation Muscle and NerveSupplement 9 S26 ndashS32

Chen R Classen J Gerloff C Celnik P Wassermann E MHallett M amp Cohen L G (1997) Depression of motorcortex excitability by low-frequency transcranial magneticstimulation Neurology 48 1398 ndash1403

Clegg B A DiGirolamo G J amp Keele S W (1998) Sequencelearning Trends in Cognitive Sciences 2 275ndash281

Cohen L G Celnik P Pascual-Leone A Corwell B Falz LDambrosia J Honda M Sadato N Gerloff C CatalaM D amp Hallett M (1997) Functional relevance ofcross-modal plasticity in blind humans Nature 389180 ndash183

Corthout E Uttl B Walsh V Hallett M amp Cowey A (1999)Timing of activity in early visual cortex as revealed by tran-scranial magnetic stimulation NeuroReport 10 2631 ndash2634

Cowey A amp Walsh V (2001) Tickling the brain Studyingvisual sensation perception and cognition by transcranialmagnetic stimulation Progress in Brain Research 134411 ndash425

Cracco R Q Cracco J B Maccabee P J amp Amassian V E(1999) Cerebral function revealed by transcranial magneticstimulation Journal of Neuroscience Methods 86 209ndash219

Crump T (2001) A brief history of science As seen throughthe development of scientific instruments (pp 78ndash81)London UK Constable amp Robinson

drsquoAlfonso A A van Honk J Hermans E Postma A amp deHaan E H (2000) Laterality effects in selective attention tothreat after repetitive transcranial magnetic stimulation atthe prefrontal cortex in female subjects NeuroscienceLetters 280 195 ndash198

Day B L Rothwell J C Thompson P D Maertens deNoordhout A Nakashima K Shannon K amp MarsdenC D (1989) Delay in the execution of voluntary movementby electrical or magnetic brain stimulation in intact manEvidence for the storage of motor programs in the brainBrain 112 649ndash663

Desmurget M Epstein C M Turner R S Prablanc CAlexander G E amp Grafton S T (1999) Role of theposterior parietal cortex in updating reaching movementsto a visual target Nature Neuroscience 2 563 ndash567


Fitzpatrick S M amp Rothman D L (2002) Meeting reportChoosing the right MR tools for the job Journal ofCognitive Neuroscience 14 806 ndash815

Funahashi S Bruce C J amp Goldman-Rakic P S (1991)Neuronal activity related to saccadic eye-movements in themonkeyrsquos dorsolateral prefrontal cortex Journal ofNeurophysiology 65 1464 ndash1483

Gerschlager W Siebner H R amp Rothwell J C (2001)Decreased corticospinal excitability after subthreshold 1 HzrTMS over lateral premotor cortex Neurology 57 449 ndash455

Gomez-Beldarrain M Grafman J Pascual-Leone A ampGarcia-Monco J C (1999) Procedural learning is impaired inpatients with prefrontal lesions Neurology 52 1853 ndash1860

Grafton S T Hazeltine E amp Ivry R (1995) Functionalmapping of sequence learning in normal humans Journalof Cognitive Neuroscience 7 497ndash510

Grosbras M H amp Paus T (2002) Transcranial magneticstimulation of the human frontal eye field Effects on visualperception and attention Journal of CognitiveNeuroscience 14 1109 ndash1120

Gugino L D Romero R Ramirez M Titone DPascual-Leone A Grimson E Weisenfeld N Kikinis R ampShenton M E (2001) The use of transcranial magneticstimulation co-registered with MRI The effect on responseprobability Clinical Neurophysiology 112 1781 ndash1792

Hallett M Wassermann E M Pascual-Leone A amp Valls-SoleJ (1999) Repetitive transcranial magnetic stimulation TheInternational Federation of Clinical NeurophysiologyElectroencephalography and Clinical NeurophysiologySupplement 52 105 ndash113

Hamilton R H amp Pascual-Leone A (1998) Cortical plasticityassociated with Braille learning Trends in CognitiveSciences 2 168 ndash174

Herwig U Padberg F Unger J Spitzer M amp Schonfeldt-Lecuona C (2001) Transcranial magnetic stimulation intherapy studies Examination of the reliability of lsquolsquostandardrsquorsquocoil positioning by neuronavigation Biological Psychiatry50 58ndash61

Hilgetag C C Theoret H amp Pascual-Leone A (2001)Enhanced visual spatial attention ipsilateral to rTMS-inducedlsquolsquovirtual lesionsrsquorsquo of human parietal cortex NatureNeuroscience 4 953 ndash957

Kapur N (1996) Paradoxical functional facilitation in brain-behaviour research A critical review Brain 119 1775 ndash1790

Kolb B amp Wishaw I Q (1998) Brain plasticity and behaviorAnnual Review of Psychology 49 43ndash64

Kosslyn S M Pascual-Leone A Felician O Camposano SKeenan J P Thompson WLGanis G Sukel K E ampAlpertN M (1999) The role of area 17 in visual imagery Convergentevidence from PET and rTMS Science 284 167 ndash170

Kozel F A Nahas Z de Brux C Molloy M LorberbaumJ P Bohning D Risch S C amp George M S (2000) Howcoil ndashcortex distance relates to age motor threshold andantidepressant response to repetitive transcranial magneticstimulation Journal of Neuropsychiatry and ClinicalNeuroscience 12 376 ndash384

Lewald J Foltys H amp Topper R (2002) Role of the posteriorparietal cortex in spatial hearing Journal of Neuroscience22 RC207

Lisanby S H Gutman D Luber B Schroeder C ampSackeim H A (2001) Sham TMS Intracerebralmeasurement of the induced electrical field and theinduction of motor-evoked potentials Biological Psychiatry49 460 ndash463

Maeda F Keenan J P Tormos J M Topka H amp Pascual-Leone A (2000) Modulation of corticospinal excitability byrepetitive transcranial magnetic stimulation ClinicalNeurophysiology 111 800 ndash805

McConnell K A Nahas Z Shastri A Lorberbaum J PKozel F A Bohning D E amp George M S (2001)The transcranial magnetic stimulation motor thresholddepends on the distance from coil to underlying cortex Areplication in healthy adults comparing two methods ofassessing the distance to cortex Biological Psychiatry 49454 ndash459

Meyer B U Britton T C Kloten H Steinmetz H ampBenecke R (1991) Coil placement in magnetic brainstimulation related to skull and brain anatomyElectroencephalography and Clinical Neurophysiology81 38ndash46

Mottaghy F M Gangitano M Sparing R Krause B J ampPascual-Leone A (2002) Segregation of areas related tovisual working memory in the prefrontal cortex revealed byrTMS Cerebral Cortex 12 369 ndash375

Mottaghy F M Hungs M Brugmann M Sparing RBoroojerdi B Foltys H Huber W amp Topper R (1999)Facilitation of picture naming after repetitive transcranialmagnetic stimulation Neurology 53 1806 ndash1812

Muellbacher W Ziemann U Wissel J Dang N Kofler MFacchini S Boroojerdi B Poewe W amp Hallett M (2002)Early consolidation in human primary motor cortex Nature415 640 ndash644

Mulleners W M Chronicle E P Palmer J E Koehler P J ampVredeveld J W (2001) Suppression of perception inmigraine Evidence for reduced inhibition in the visualcortex Neurology 56 178 ndash183

Munchau A Bloem B R Irlbacher K Trimble M R ampRothwell J C (2002) Functional connectivity of humanpremotor and motor cortex explored with repetitivetranscranial magnetic stimulation Journal of Neuroscience22 554 ndash61

Nordstrom M A amp Butler S L (2002) Reduced intracorticalinhibition and facilitation of corticospinal neurons inmusicians Experimental Brain Research 144 336 ndash342

Oliveri M Rossini P M Traversa R Cicinelli P FilippiM M Pasqualetti P Tomaiuolo F amp Caltagirone C(1999) Left frontal transcranial magnetic stimulationreduces contralesional extinction in patients with unilateralright brain damage Brain 122 1731 ndash1739

Pascual-Leone A Bartres-Faz D amp Keenan J P (1999)Transcranial magnetic stimulation Studying the brainndashbehaviour relationship by induction of lsquolsquovirtual lesionsrsquorsquoPhilosophical Transactions of the Royal Society of LondonSeries B Biological Sciences 354 1229 ndash1238

Pascual-Leone A Gates J R amp Dhuna A (1991) Inductionof speech arrest and counting errors with rapid-ratetranscranial magnetic stimulation Neurology 41 697 ndash702

Pascual-Leone A Grafman J amp Hallett M (1994) Explicit andimplicit learning and maps of cortical motor output Science265 1600 ndash1601

Pascual-Leone A Houser C M Reese K Shotland L IGrafman J Sato S Valls-Sole J Brasil-Neto J PWassermann E M Cohen L G Hallett M (1993) Safety ofrapid-rate transcranial magnetic stimulation in normalvolunteers Electroencephalography and ClinicalNeurophysiology 89 120 ndash130

Pascual-Leone A Nguyet D Cohen L G Brasil-Neto J PCammarota A amp Hallett M (1995) Modulation of muscleresponses evoked by transcranial magnetic stimulationduring the acquisition of new fine motor skills Journal ofNeurophysiology 74 1037 ndash1045

Pascual-Leone A Rubio B Pallardo F amp Catala M D(1996) Rapid-rate transcranial magnetic stimulation of leftdorsolateral prefrontal cortex in drug resistant depressionLancet 348 233 ndash237

Pascual-Leone A Valls-Sole J Wassermann E M amp Hallett M


(1994) Responses to rapid-rate transcranial magneticstimulation of the human motor cortex Brain 117 847ndash858

Pascual-Leone A amp Walsh V (2001) Fast backprojectionsfrom the motion to the primary visual area necessary forvisual awareness Science 292 510 ndash512

Pascual-Leone AWalshVampRothwell J C (2000) Transcranialmagnetic stimulation in cognitive neurosciencemdashvirtuallesion chronometry and functional connectivity CurrentOpinion in Neurobiology 10 232 ndash237

Pascual-Leone A Wassermann E M Grafman J amp HallettM (1996) The role of the dorsolateral prefrontal cortexin implicit procedural learning Experimental BrainResearch 107 479 ndash485

Paus T (1999) Imaging the brain before during and aftertranscranial magnetic stimulation Neuropsychologia 37219 ndash224

Paus T (2002) Combination of transcranial magneticstimulation with brain imaging In A Toga amp J Mazziotta(Eds) Brain mapping The methods (2nd edpp 691 ndash705) San Diego Elsevier Science

Paus T Jech R Thompson C J Comeau R Peters T ampEvans A C (1997) Transcranial magnetic stimulation duringpositron emission tomography A new method for studyingconnectivity of the human cerebral cortex Journal ofNeuroscience 17 3178 ndash3184

Paus T Sipila P K amp Strafella A P (2001) Synchronizationof neuronal activity in the human primary motor cortex bytranscranial magnetic stimulation An EEG study Journal ofNeurophysiology 86 1983 ndash1990

Pearce A J Thickbroom G W Byrnes M L amp MastagliaF L (2000) Functional re-organisation of the corticomotorprojection to the hand in skilled racquet playersExperimental Brain Research 130 238 ndash243

Robertson E M Tormos J M Maeda F amp Pascual-Leone A(2001) The role of the dorsolateral prefrontal cortex duringsequence learning is specific for spatial informationCerebral Cortex 11 628ndash635

Rossini P M Barker A T Berardelli A Caramia M DCaruso G Cracco R Q Dimitrijevic M R Hallett MKatayama Y Lucking C H Maertens de Noordhout AMarsden C Murray N Rothwell J Swash M amp TombergC (1994) Non-invasive electrical and magnetic stimulationof the brain spinal cord and roots Basic principles andprocedures for routine clinical application Report of anIFCN committee Electroencephalography and ClinicalNeurophysiology 91 79ndash92

Rothwell J C (1997) Techniques and mechanisms of action oftranscranial magnetic stimulation of the human motorcortex Journal of Neuroscience Methods 74 113 ndash122

Rushworth M F Ellison A amp Walsh V (2001)Complementary localization and lateralization of orientingand motor attention Nature Neuroscience 4 656 ndash661

Sack A T Sperling J M Prvulovic D Formisano E GoebelR Di Salle F Dierks T amp Linden D E (2002) Trackingthe mindrsquos image in the brain II Transcranial magneticstimulation reveals parietal asymmetry in visuospatialimagery Neuron 35 195ndash204

Shapiro K A Pascual-Leone A Mottaghy F M GangitanoM amp Caramazza A (2001) Grammatical distinctions in theleft frontal cortex Journal of Cognitive Neuroscience 13713 ndash720

Siebner H R Takano B Peinemann A Schwaiger MConrad B amp Drzezga A (2001) Continuous transcranialmagnetic stimulation during positron emission tomographyA suitable tool for imaging regional excitability of the humancortex Neuroimage 14 883 ndash890

Stewart L M Walsh V amp Rothwell J C (2001) Motor andphosphene thresholds A transcranial magnetic stimulationcorrelation study Neuropsychologia 39 415 ndash419

Strafella A P Paus T Barrett J amp Dagher A (2001)Repetitive transcranial magnetic stimulation of the humanprefrontal cortex induces dopamine release in the caudatenucleus Journal of Neuroscience 21 RC157

Theoret H Haque J amp Pascual-Leone A (2001) Increasedvariability of paced finger tapping accuracy followingrepetitive magnetic stimulation of the cerebellum inhumans Neuroscience Letters 306 29ndash32

Theoret H Kobayashi M Ganis G Di Capua P ampPascual-Leone A (2002) Repetitive transcranial magneticstimulation of human area MTV5 disrupts perception andstorage of the motion aftereffect Neuropsychologia 402280 ndash2287

Topper R Mottaghy F M Brugmann M Noth J amp HuberW (1998) Facilitation of picture naming by focal transcranialmagnetic stimulation of Wernickersquos area ExperimentalBrain Research 121 371 ndash378

van Donkelaar P Less J H amp Drew A S (2000) Transcranialmagnetic stimulation disrupts eye ndashhand interactions in theposterior parietal cortex Journal of Neurophysiology 841677 ndash1680

Walsh V amp Cowey A (2000) Transcranial magneticstimulation and cognitive neuroscience Nature ReviewsNeuroscience 1 73ndash79

Walsh V amp Pascual-Leone A (2003) Transcranial magneticstimulation A neurochronometrics of mind CambridgeMIT Press

Wassermann E M (1998) Risk and safety of repetitivetranscranial magnetic stimulation Report and suggestedguidelines from the International Workshop on the Safety ofRepetitive Transcranial Magnetic Stimulation June 5ndash71996 Electroencephalography and ClinicalNeurophysiology 108 1ndash16

Willingham D B Salidis J amp Gabrieli J D E (2002) Directcomparison of neural systems mediating conscious andunconscious skill learning Journal of Neurophysiology 881451 ndash1460

Wilson S A Thickbroom G W amp Mastaglia F L (1993)Transcranial magnetic stimulation of the motor cortex innormal subjects The representation of two intrinsic handmuscles Journal of Neurological Sciences 118 134 ndash144

Wong M amp Moss R L (1992) Modulation of single unitactivity in the rat amygdala by neurotransmitters estrogenpriming and synaptic input from the hypothalamus andmidbrain Synapse 10 94ndash120


disease Furthermore damage is seldom well demarcat-ed or completely isolated and its location often reflectsthe vagaries of the cerebrovascular system rather thanconforming neatly to the needs of the cognitive scien-tist The fog surrounding these interpretative dilemmasbecomes even thicker with the effects of long-standingmedications

An ability to directly inquire about the causal contri-bution of different brain areas to behavior is greatlyneeded Recent years have seen this need partly as-suaged by TMS (Walsh amp Pascual-Leone 2003)

A Problem Lost

Moving beyond a merely correlative description of therelationship between brain and behavior is the freshapproach offered by TMS It allows the noninvasiveelectromagnetic stimulation of cortical sites This prin-ciple was demonstrated almost 20 years ago with invol-untary finger movement elicited by stimulation over themotor cortex (Barker Jalinous amp Freeston 1985) Themagnetic field produced around a coil can pass readilyacross the scalp and skull and induces an electricalcurrent within the brain tissue This occurs with minimalattenuation of the magnetic field Consequently signif-icant currents can be induced without having to applysubstantial voltages across the skull minimizing theactivation of pain fibers Early studies used single-pulseTMS to the occipital cortex time-locked to the presenta-tion of a visual stimulus to induce errors in the detectionof letters (Amassian et al 1989) These errors weremaximal with TMS applied between 80 and 100 msecfollowing the presentation of the visual stimulus imply-

ing that only at these times was the occipital cortexmaking a critical contribution to letter recognition(Figure 1) Similar approaches were employed to studythe role of the motor cortex in finger movements (Dayet al 1989) and of the somatosensory cortex in tactileperception (Cohen et al 1997)

This type of experiment can provide insight intowhen a given brain area is making a critical contribu-tion to a behavior In addition by applying single TMSpulses separated by a variable interstimulus interval totwo different brain areas the technique can be expand-ed to improve our understanding of the dynamicinteraction between brain areas This double-pulseparadigm has been successfully applied to explore therole of back-projections in visual awareness (Figure 1Pascual-Leone amp Walsh 2001)

One major limitation of these single- or paired-pulseTMS paradigms is that there is often insufficient infor-mation about the time at which an area makes a criticalcontribution to a given behavior TMS needs to bedelivered correctly in time and space for the verytransient disruption of brain function induced by single-pulse TMS to have a measurable behavioral effectTherefore the possibility of covering a larger timewindow with repetitive TMS (rTMS) to first explorethe lsquolsquospacersquorsquo dimension has some utility rTMS allowsthe site of stimulation to be established while latersingle-pulse experiments can give insight into thechronometry of this critical contribution to behavior(eg the role of the occipital cortex during Braillereading Figure 2) The combined use of functionalneuroimaging to provide spatial information evokedpotentials to provide chronometric information and

Figure 1 Probing chronome-try with TMS (A) A TMS pulseto the occipital probe can sup-press visual perception when itis applied between 80 and 120msec after stimulus presenta-tion (from Cracco Maccabee ampAmassian 1999) (B) The per-ception of a TMS-induced mov-ing phosphene after stimulationof motion area V5 can besignificantly suppressed by asecond TMS pulse applied tothe primary visual cortex (V1)10 to 40 msec later These datashow that fast V5ndashV1 feedbackprojections are necessary forawareness of motion (fromPascual-Leone amp Walsh 2001)









Sham Stimulation


Timing






















(Hz)Train Duration


Duration


Area 17 1 10 90 MT NA








Paced fingertapping




DLPFC DMPFCVPFC



Parietal1

10 80 MT NA




1 5 110 MT NA





Fixed Intensity


















































CONCLUSIONS


Acknowledgments



REFERENCES





























































































Sham Stimulation


Timing






















(Hz)Train Duration


Duration


Area 17 1 10 90 MT NA








Paced fingertapping




DLPFC DMPFCVPFC



Parietal1

10 80 MT NA




1 5 110 MT NA





Fixed Intensity


















































CONCLUSIONS


Acknowledgments



REFERENCES








































































































(Hz)Train Duration


Duration


Area 17 1 10 90 MT NA








Paced fingertapping




DLPFC DMPFCVPFC



Parietal1

10 80 MT NA




1 5 110 MT NA





Fixed Intensity


















































CONCLUSIONS


Acknowledgments



REFERENCES





























































































(Hz)Train Duration


Duration


Area 17 1 10 90 MT NA








Paced fingertapping




DLPFC DMPFCVPFC



Parietal1

10 80 MT NA




1 5 110 MT NA





Fixed Intensity


















































CONCLUSIONS


Acknowledgments



REFERENCES























































































Fixed Intensity


















































CONCLUSIONS


Acknowledgments



REFERENCES
































































































































CONCLUSIONS


Acknowledgments



REFERENCES



















































































































CONCLUSIONS


Acknowledgments



REFERENCES








































































































CONCLUSIONS


Acknowledgments



REFERENCES































































































CONCLUSIONS


Acknowledgments



REFERENCES























































































CONCLUSIONS


Acknowledgments



REFERENCES



















































































































































































Documents

Studies in Cognition: The Problems Solved and Created by