This work w

The views e

Administration.

Corresponde

29425.

E-mail addre

Submitted M

1935-861X/09/$

doi:10.1016/j.br

Brain Stimulation (2010) 3, 78–86

www.brainstimjrnl.com

Conditioning of transcranial magnetic stimulation:Evidence of sensory-induced respondingand prepulse inhibition

Kevin A. Johnson,a Gordon C. Baylis,b Donald A. Powell,c

F. Andrew Kozel,d Scott W. Miller,e Mark S. Georgef

aDepartments of Neuroscience and Psychiatry, Medical University of South Carolina, Charleston, South CarolinabDepartment of Psychology, University of South Carolina, Charleston, South CarolinacShirley L. Buchanan Neuroscience Laboratory, Dorn VA Medical Center, Columbia, South CarolinadDepartment of Psychiatry, University of Texas Southwestern Medical Center, Dallas, TexaseDepartment of Biostatistics, Bioinformatics, and Epidemiology, Medical University of South Carolina, Charleston, SouthCarolinafDepartments of Psychiatry, Radiology, and Neurology, Medical University of South Carolina, Charleston, South Carolina

BackgroundTranscranial magnetic stimulation (TMS) is a non-invasive method for stimulating the human cortex.Classical conditioning is a phenomenon of developed associations between stimuli. Our primaryobjective was to determine whether TMS effects could be conditioned. Prepulse inhibition representsanother relationship between two stimuli, and a secondary assessment was performed to explore thisrelationship.

MethodsAn auditory-visual conditioning stimulus (CS) was paired with the TMS unconditioned stimulus (US)over motor cortex producing a motor-evoked potential (MEP) unconditioned response (UR). Twoversions of the CS-US pairing paradigms were tested, one with a short intertrial interval (ITI) andanother with a long ITI. The short ITI paradigm had more CS-US pairings and shorter session durationthan the long ITI paradigm. Tests for conditioned responses (CRs) were performed following CS-USpairing (CS1/US1), by presenting the CS alone (CS1/US2). Reverse testing was also performedafter CS-US pairing (CS1/US1) in separate sessions, by presenting the US alone (CS2/US1).

as developed as part of the K.A.J. dissertation research and was presented at the American College of Neuropsychopharmacology.

xpressed herein are those of the authors and do not necessarily reflect the views of the US Department of Veterans Affairs or the Veterans

nce: Kevin A. Johnson, Medical University of South Carolina, Brain Stimulation Laboratory, 67 President Street, Room 502N, Charleston, SC

ss: johnsk@musc.edu

arch 28, 2009; revised August 16, 2009. Accepted for publication August 17, 2009.

-see front matter � 2010 Elsevier Inc. All rights reserved.

s.2009.08.003

TMS and classical conditioning 79

ResultsEvidence for CRs was found only with the short ITI paradigm. The magnitudes of CRs were smallerthan TMS-induced MEPs, and the CRs were found only in a percentage of tests. Prepulse inhibitionwas robustly evident for the long ITI paradigm, but not for the short ITI paradigm.

ConclusionsWe have found evidence that classical conditioning principles can be applied to brain stimulation inhumans. These findings provide a method for exploring brain and behavioral relationships in humans, aswell as suggesting approaches to enhance therapeutic uses of TMS or other forms of brain stimulation.� 2010 Elsevier Inc. All rights reserved.

Keywords transcranial magnetic stimulation; TMS; classical conditioning; prepulse inhibition

Transcranial magnetic stimulation (TMS) is a noninva-sive method for stimulating the human cortex, withemerging neuroscience and clinical applications.1,2 Anelectric discharge produces current flow through a coilplaced on the head, which induces a magnetic field thatpasses through the scalp. The magnetic field depolarizesneurons in the outer cortical layer, producing effects depen-dent on the location and parameters of stimulation. Forexample, single-pulse TMS over motor cortex can generatea motor-evoked potential (MEP) in a correspondingsomatic muscle3,4 or over visual cortex can produce theperception of phosphenes.5,6 High-frequency stimulationover Broca’s area can disrupt speech,7 and recent large clin-ical trials suggest that daily high-frequency stimulationover prefrontal cortex for several weeks may emerge asa treatment for depression.8,9 Although most research hasfocused on the effects of TMS through cortical stimulation,it should be noted that TMS is a compound stimulus alsoproducing tactile sensations at the scalp and an auditory‘‘click’’ with each pulse. Newer TMS sham methods are at-tempting to control for these additional factors.10,11

Classical conditioning is a long established and exten-sively studied phenomenon of developed associationsbetween stimuli.12,13 An unconditioned stimulus (US) natu-rally produces an unconditioned response (UR). A contin-gency is developed by pairing a conditioning stimulus(CS) with the US. After CS-US pairing, successful condi-tioning results in the CS alone now generating a conditionedresponse (CR) that resembles the UR. There is extensiveresearch examining parameters required to achieve success-ful conditioning, including the valence and relevance ofstimuli, the number of pairings needed to produce condi-tioning, and timing parameters.14-17 Classical conditioningtheory has been applied to numerous behaviors, from basicreflexes to emotional and cognitive associations.18-22 Phys-iologically paired associative stimulation, pairing TMS, andperipheral nerve stimulation demonstrates plasticitythrough changes in cortical excitability.23,24 Given suchplasticity and the range of sensory stimuli that can bepaired to produce conditioning, we wondered whether clas-sical conditioning could occur with TMS as the US.

The primary objective was to determine whether single-pulse TMS over motor cortex as the US could be paired

with a sensory CS, such that the CS subsequently wouldproduce a motor CR. To our knowledge, all prior studiesthat used classical conditioning to produce a CR haveinvolved the pairing of two sensory stimuli. As the neuro-circuitry of contingency development varies, based on thestimuli and behaviors, we did not know whether directbrain stimulation with TMS might bypass the neuro-anatomy required to generate a CR with conditioning.

A secondary objective was to explore reverse testingafter pairing of TMS as the US with a sensory CS. Ratherthan presenting the CS alone, tests here involve presenta-tion of the US alone. The magnitude of TMS-inducedelectromyography (EMG) could then be compared betweenthe CS1/US1 condition and the CS2/US1 condition. Ifthe magnitude of the CS1/US1 condition is greater than ofthe CS2/US1 condition, one possibility is that the sensorystimuli (CS1) is causing prepulse inhibition.

Prepulse inhibition represents a relationship betweentwo stimuli, whereby one stimulus inhibits the response toa following stimulus.25 Paired-pulse TMS is sometimesdescribed as a prepulse inhibition (or excitation), as oneTMS stimulus can influence the effect of a following stim-ulus.1,26 However, this is a physiologic property that doesnot require contingency learning. An auditory prepulsehas been shown to inhibit motor responses to TMS, througha proposed startle mechanism.27 It has also been proposedthat an auditory prepulse can inhibit motor responsesthrough learned association, as evidenced by a decreasingamplitude over pairings.28 The secondary objective of thisstudy was intended to further explore the building ofa sensory-TMS association; through the development ofthe learned association that sensory stimuli always predictsa TMS occurrence. The contingency rule was then brokenwith TMS presented alone, with assessment for a changein resultant motor response.

Methods and Materials

Participants

A total of 21 healthy volunteer men, age 21-40 years, wereenrolled as approved by the Medical University of South

80 K. A. Johnson et al

Carolina (MUSC) Institutional Review Board, signingwritten informed consent. Exclusionary criteria includeda history of psychiatric or neurologic disorders, presence ofsignificant medical conditions, or a finding that 120% oftheir resting motor threshold (MT) was greater than theTMS machine maximum output. Participants in the shortintertrial interval (ITI) paradigms completed up to threeconsecutive sessions, alternating between conditioning testsand prepulse inhibitions tests. Participants in the long ITIparadigms completed up to two consecutive sessions.

Experimental setup

Participants sat approximately 18 inches from a laptopcomputer screen and speakers (17.1-inch monitor, speakervolume set at approximately 75 dB for a single frequency440 Hz tone; Gateway M675; Gateway, Inc, Irvine, CA),with E-Prime software (version 1.1, Psychology SoftwareTools, Pittsburgh, PA) programmed to present the auditoryand visual stimuli; as well as to trigger TMS pulses andEMG recordings. A single earplug was provided for the leftear, to protect participant hearing from TMS noise. TMS(figure-eight coil connected to Magstim Super Rapidmachine; Magstim Company Ltd, Whitland, South WestWales) was fixed by a stereotaxic frame over the left motorcortex, with output power set at 120% of individual restingMT of the right abductor pollicis brevis (APB) muscle asdetermined by visual approximation.29 Two electrodes(silver/silver chloride disposable electrodes; Viasys Health-care Inc, Conshohocken, PA) were placed across the rightAPB muscle, with a third ground electrode attached tothe back of the hand. Electrodes were connected to anamplifier and filter (LabLinc V75-05 and V75-48, Coul-bourn Instruments LLC, Allentown, PA) with the followingsettings: amplifier coupling510 Hz; amplifier gain510 K;filter hi-pass cutoff530 Hz; filter lo-pass cutoff52.0KHz. The amplified EMG signal was recorded synchro-nously with stimuli presentation via LabView hardware(SCB-68, DAQCard-6062E, and 68-pin cable; NationalInstruments, Austin, TX) for recording with LabView Soft-ware (version 7 Express; National Instruments), with a re-corded range of amplified signal5210 V to 10 V (21mV to 1 mV nonamplified signal range).

Short ITI paradigm parameters

Each session involved a total of 170 events (includingpairing and test events) lasting approximately 10 minutes,limited to duration length by coil overheating. A baselinetest was acquired as the first event, with postpairing tests asevents 109, 115, 116, 121, 136, 144, 152, 169, and 170(Figure 1). The duration between events was randomly setat ITI52, 3, 4, 5, or 6 seconds. For pairing, the compoundauditory-visual stimulus was presented for 0.4 seconds,with single-pulse TMS over motor cortex occurring at theend of the 0.4 seconds. The auditory component was

a single-frequency 440 Hz tone, and the visual componentwas a white screen. Between pairings, participants vieweda black screen with white fixation cross. For comparisonwith CRs, cued ‘‘volitional’’ motor movement of the rightAPB muscle were acquired in separate paradigms (partici-pants were instructed to mimic the TMS effect as cued bya 0.4-second visual stimulus).

Long ITI paradigm parameters

Each session involved a total of 50 events (includingpairing and test events) lasting approximately 26 minutes,limited to this duration by participant comfort (Figure 1).A baseline test was acquired as the first event, with post-pairing tests as events 20, 40, 44, 47, 49, and 50. The dura-tion between events was randomly set at ITI510, 20, 30,40, 50, 60, 70, or 80 seconds. For pairing, the compoundauditory-visual stimulus was presented for 0.5 seconds,with single-pulse TMS over motor cortex occurring towardthe end at 0.4 seconds. The auditory component wasa single-frequency 440-Hz tone, and the visual componentwas a checkerboard alternating every 100 milliseconds. Tohelp participants maintain attention throughout this para-digm of long duration and long ITI, general interest photo-graphs were presented between pairings at 9-secondintervals.

Conditioning tests

In sessions with conditioning tests, pairing events involvedpresentation of compound auditory-visual stimulus (CS1)with TMS (US1). Test events for CRs involved presenta-tion of the auditory-visual stimulus (CS1) without TMS(US2) (Figure 1).

Prepulse inhibition assessment

In sessions with prepulse inhibition tests, pairing eventsagain involved presentation of compound auditory-visualstimulus (CS1) with TMS (US1). Test events hereinvolved no auditory-visual stimulation (CS2) with TMSpresented alone (US1). Assessment for prepulse inhibitioninvolved comparing the magnitude of pairing events (CS1/US1) with the magnitude of tests events (CS2/US1)(Figure 1).

Statistical analysis

For sessions with conditioning tests, each test was visuallyassessed for the presence or absence of a CR as evidencedby EMG activity above noise levels with temporalappropriateness.

For sessions with prepulse inhibition tests, peak-to-peakMEP amplitudes for each session were grouped into fourcategories: baseline (the first event of the paradigm, a testevent involving TMS alone before any pairing); primary

Figure 1 Experimental setup. (a), A short ITI paradigm session involves 170 events (160 pairing events and 10 test events) presented overapproximately 10 minutes. (b), A long ITI paradigm session involves 50 events (43 pairing events and seven test events) presented overapproximately 26 minutes. (c), Pairing events for the short ITI paradigm involve an auditory-visual stimulus (CS) followed by TMS(US), separated by an ITI of 2-6 seconds. (d), Pairing events for the long ITI paradigm involve an auditory-visual stimulus (CS), followedby TMS (US), separated by an ITI of 10-80 seconds. (e), Conditioning tests involve presentation of the auditory-visual stimulus alone (CSalone), followed by an EMG recording for potential conditioned responses. (f), Prepulse inhibition tests involve presentation of TMS alone(US alone), followed by an EMG recording of the MEP. ITI5intertrial interval; CS5conditioning stimulus; TMS5transcranial magneticstimulation; US5unconditioning stimulus; EMG5electromyography; MEP5motor-evoked potential.

TMS and classical conditioning 81

82 K. A. Johnson et al

pairings (all pairing events before the first nonbaseline testevent); end pairings (all pairing events after the firstnonbaseline test event); and tests (all nonbaseline testevents, which are interspersed between end pairings).A hierarchical linear model was used to account for thecorrelations present in the data (more than one trial forsome subjects, repeated measurements on the same subjectsover time). As significant skew was detected in the data,a natural log transform of the data was used to reduce theskew, and render the data more normally distributed. Thestandard error estimates reported in the results were reversetransformed back to match mean amplitude values.A compound symmetry model was used to model thecorrelation of repeated measures on the same subjects; thiswas accomplished partially because of the design of theexperiment and partially based on data characteristics(small sample size). This analysis was implemented byusing the mixed procedure in SAS (version 9.1; SASInstitute, Inc, Cary, NC).

Results

Short ITI conditioning tests

Ten participants completed two sessions of this paradigm,with five other participants completing one session (fora total of 25 sessions). Three sessions were excludedbecause of low TMS-induced MEPs in the second-half ofpairings (average amplitude,0.05 mV), likely indicatingcoil movement from optimal position or other proceduralfailure. Thus a total of 22 sessions were analyzed.A selected timeframe was set for expected occurrence ofconditioned responses, from 0.1 second after the onset ofthe CS up to 1.0 second after US presentation (a 1.3-secondwindow). Visual inspection in this window noted 1 EMGspike of 22 preconditioning tests, indicating a smallpotential of some nonconditioning related noise in thedata. Testing for CRs, visual inspection noted 28 EMGspikes of 198 postconditioning tests. Considering distribu-tion across trials, 11 of 22 sessions and 9 of 15 participantshad at least one positive test (Figure 2).

Long ITI conditioning tests

Four participants completed two sessions of this paradigm,with two other participants completing one session (fora total of 10 sessions). No sessions were excluded becauseof low TMS-induced MEPs in the second-half of pairings(average amplitude,0.05 mV). A selected timeframe wasset for expected occurrence of conditioned responses, from0.1 second after the onset of the CS up to 1.0 second afterUS presentation (a 1.3-second window). Visual inspectionin this window noted no EMG spikes of eight precondition-ing tests (two sessions did not have preconditioning tests),indicating a low potential of some nonconditioning-related

noise in the data. Testing for CRs, visual inspection noted 1EMG spike of 60 postconditioning tests. The magnitude ofthis spike was quite small (0.02 mV) relative to the averageEMG amplitude of TMS for second half of that sessionswith positive spikes (0.94 mV).

Short ITI Prepulse inhibition assessment

Ten participants completed one session of this paradigm,with five other participants completing two sessions (fora total of 20 sessions). Three sessions were excludedbecause of low TMS-induced MEPs in all pairings (averageamplitude,0.05 mV), likely indicating coil movementfrom optimal position or other procedural failure. Thusa total of 17 sessions were analyzed. For each classification,mean values (with estimated standard error) are as follows:baseline50.60 mV (0.08); primary pairings50.52 mV(0.07); end pairings: 0.64 mV (0.09); and tests50.70 mV(0.10). Tukey-adjusted P-values for the various compari-sons are as follows: baseline versus tests50.72; primarypairings versus end pairings50.53; primary pairings versustests50.75; and end pairings versus tests ..99. None ofthese or the other comparisons were significant at theP5.05 level (Figure 3).

Long ITI Prepulse inhibition assessment

Six participants completed one session of this paradigm. Nosessions were excluded because of low TMS-induced MEPsin all pairings (all sessions had an average MEP amplitudeof at least 0.05 mV). Data for each session were groupedinto four categories: baseline (TMS alone before pairing);primary pairings (compound stimulus and TMS beforetests); end pairings (compound stimulus and TMS inter-spersed between tests); and tests (TMS alone after pairing).For each classification, mean values (with estimatedstandard error) are as follows: baseline50.51 mV (0.12);primary pairings50.42 mV (0.09); end pairings: 0.42 mV(0.09); and tests50.75 mV (0.18). Tukey-adjusted P-valuesfor the various comparisons are as follows: baseline versusprimary pairing50.06; baseline versus end pairings50.12;baseline versus tests50.17; primary pairings versus endpairings . .99; primary pairings versus tests50.04; andend pairings versus tests50.02. It should be noted thatthe differences in sample size is related to the standarderrors for each category, which could contribute to thefailure to find significant effects, specifically for baseline(Figure 4).

Discussion

We found evidence for conditioned responding using theshort ITI paradigm parameters. These CRs differed fromTMS-induced MEPs and cued ‘‘volitional’’ APB musclemovement by the smaller magnitude, but the similar

Figure 2 Results for the short ITI conditioning tests. Average rectified EMG recording for sampling of 28 TMS-induced MEPs, acquiredfrom five participants (top). Average rectified EMG recording for all 28 detected conditioned responses, found with 9 of 15 participants(middle). Average rectified EMG recording for sample of 28 cued ‘‘volitional’’ APB muscle movement, acquired from 5 participants(bottom). ITI5intertrial interval; EMG5electromyography; TMS5transcranial magnetic stimulation; MEP5motor-evoked potential;APB5abductor pollicis brevis.

TMS and classical conditioning 83

temporal profiles are indicative of a cued, biologicresponse. No repeatable evidence for CRs was found withthe long ITI paradigm parameters. Consistent with our longITI paradigm results, another study failed to find CRs froma compound auditory-visual stimulus (CS1/US2), aftersensory stimuli pairing with TMS (CS1/US1).28 Failuresto find CRs again could relate to the ITI parameter, as theITI used in the study by Luber et al.28 (20610 seconds)is greater than our short ITI paradigm. Our long ITI para-digm also had much fewer pairings than the short ITI para-digm, which is one potential parameter that could effect thedevelopment of conditioning. However, the study by Luberet al.28 failed to find CRs even with repeated session overdays. As our short and long ITI paradigms differed inseveral regards (ITI, paradigm duration, number of pair-ings, CS characteristics, and attentional task between pair-ings), future work to identify the set of parameters that

optimally generates conditioned responding would beimportant.

Based on the temporal characteristics of the conditionedresponses, we presume that the contingency is developed inthe motor cortex rather than through sensory feedback(although not conclusively established by blocking sensoryfeedback). There is evidence from eyeblink conditioningthat the motor cortex can indeed play a role in contingencydevelopment.30-32 We assumed that the US valence, andthus the strength of conditioning, would be increasedwith high TMS output (120% MT). However, TMS canhave disruptive effects on cognitive processes33,34 andincreased TMS output affects a larger cortical region.35,36

Thus, reducing the magnitude of stimulation could poten-tially permit more robust conditioning. The valence of theCS stimulus could also be modified, either in characteristicsor by selecting a different sensory modality.

Figure 3 Results for the short ITI prepulse inhibition tests.Baseline is the average MEP magnitude of the first event of theparadigm, a test event involving TMS alone before any pairing(n517 events). Primary pairings column is the average MEPmagnitude for all pairing events before the first nonbaseline testevent (n51819 events). End pairings column is the averageMEP magnitude for all pairing events after the first nonbaselinetest event (n5901 events). Tests column is the average MEP forall nonbaseline test events, which are interspersed between endpairing (n5153 events). The MEP magnitude is determinedpeak-to-peak, and variance is displayed as standard error.ITI5intertrial interval; MEP5motor-evoked potential;TMS5transcranial magnetic stimulation.

84 K. A. Johnson et al

For the short ITI parameters in the prepulse inhibitionassessment, no statistically significant differences werefound between MEP magnitudes, with or without a pre-TMS auditory-visual stimulus. However, for the long ITIparameters, MEP magnitudes were significantly less witha pre-TMS auditory-visual stimulus than with TMS alone.This suggests that the pre-TMS auditory-visual stimulus

Figure 4 Results for the long ITI prepulse inhibition tests. Base-line is the average MEP magnitude of the first event of the para-digm, a test event involving TMS alone before any pairing (n56events). Primary pairings column is the average MEP magnitudefor all pairing events before the first nonbaseline test event(n5108 events). End pairings column is the average MEP magni-tude for all pairing events after the first nonbaseline test event(n5150 events). Tests column is the average MEP for all nonbase-line test events, which are interspersed between end pairing (n536events). The MEP magnitude is determined peak-to-peak, andvariance is displayed as standard error. . ITI5intertrial interval;MEP5motor-evoked potential; TMS5transcranial magneticstimulation.

produces prepulse inhibition with the long ITI parameters.We suspect this prepulse inhibition is an inherent phenom-enon as MEP magnitude does not change between primaryand end pairings. However, a more rigorous experimentaldesign (increasing the number of baseline measures andadding a control set of nonpaired TMS over time) would beneeded to address this question directly.

Whether an inherent condition or conditioning process,converging evidence demonstrates that sensory stimuli canresult in prepulse inhibition of TMS.27,28 Both this studyand a previous study27 have shown that a sensory stimulus(auditory-visual or auditory) before TMS produces an MEPless than what is produced by TMS alone. In the currentstudy, the frequently presented sensory stimuli predictedthe occurrence of TMS. We hypothesized that a changein MEP magnitude over time with successive pairing wouldindicate conditioned learning; however, we failed to findsuch an effect (Figures 3 and 4, no significant differencebetween primary pairings and end pairings). The previousstudy describes the sensory stimuli as a startle stimulus,27

although specific parameters differ from the current study.Thus the inhibition of the TMS effect may be an inherentchange in cortical excitability induced by the sensory stim-ulus, rather than a change in excitability related to expec-tancy (or lack of expectancy) for TMS. The findings inboth studies were similarly ITI dependent, as the previousstudy found prepulse inhibition for long ITI (20, 30, and40 seconds), but not for short ITIs (5 and 10 seconds). Inter-estingly, both studies suggest a minimum timeframebetween sensory presentations required to detect changesin cortical excitability. Similar to our long ITI paradigmresults, the study by Luber et al.28 also found inhibitedMEPs with the compound auditory-sensory stimulus.Unlike our study, the study by Luber et al.28 noteda progression of inhibition over successive pairings, indi-cating a conditioning process.

We noted that prepulse inhibition is seen with the longerITI, but the generation of conditioned responding is not.Conversely, prepulse inhibition is not seen with the shorterITI, but the generation of conditioned responding is. Wethus speculate that there may be a reverse relationshipbetween the capacity to generate prepulse inhibition andCRs, with ITI as a determining factor. Given the confoundsof sample size and differences in experimental designs,a more focused experimental design would be needed toclarify this suspected finding.

Classical conditioning with TMS could have manyimportant applications, such as a screening tool for drugsblocking conditioning or as a method for enhancing brainstimulation therapy. Given the limited conditioned respond-ing observed, refinement of these pilot paradigms toincrease the magnitude and occurrence of CRs would behelpful in developing such applications. However, prepulseinhibition appears to be a more robust phenomenon asfound here in the long ITI parameter sample, and alsoreplicated in similar research. Thus, prepulse inhibition

TMS and classical conditioning 85

may have more immediate applications without significantrefinement of paradigm parameters.

Numerous derivative studies can be developed applyingclassical conditioning theory to TMS. Here we used single-pulse TMS over motor cortex as the US. However, brainstimulation alone could be tested as a CS for pairing withanother US-UR event, although controlling for the TMSauditory and tactile effects could be a challenge. TMScould also be tested both as the CS and US, following aftercellular work whereby subthreshold stimulation can beconditioned to produces effects.19,37 Single-pulse TMS hasbeen shown to temporarily disrupt cognitive function, andcould potentially be used to disrupt prior conditioning ifthe relevant neuroanatomy occurs cortically. There islimited data that repetitive TMS has some efficacy in treat-ing PTSD symptoms,38 and an improved understanding ofTMS and conditioning could lead to more informed treat-ment approaches for PTSD or other disorders of abnormalconditioning. Lastly, TMS possibly could gauge condi-tioning in humans, as cortical excitability has been shownto change with contingency development.30 This currentwork demonstrates the potential of TMS as a neurosciencetool and the interaction of psychologic factors on brainstimulation. The effects of TMS are dependent onnumerous factors, including anatomic positioning, powerof stimulation, frequency of stimulation, and individualbrain characteristics. Although we specifically examinedclassical conditioning with single-pulse TMS over motorcortex in healthy men, future work would be required todetermine whether the principles of classical conditioningcan be applied to other approaches of brain stimulation.

Acknowledgments

We acknowledge the late Dr. Oakley Ray, who firstcontacted one of the authors (M.S.G.) to inquire aboutconditioning and TMS. We also acknowledge the guidanceof the dissertation committee: Mark S. George, Gordon C.Baylis, Daryl E. Bohning, F. Andrew Kozel, AntonietaLavin, Jacqueline F. McGinty, and Ronald E. See. Parts ofthe experimental setup were developed with assistancefrom John Walker (MUSC, Biomedical Engineering).

References

1. Terao Y, Ugawa Y. Basic mechanisms of TMS. J Clin Neurophysiol

2002;19(4):322-343.

2. Anand S, Hotson J. Transcranial magnetic stimulation: neurophysio-

logical applications and safety. Brain Cogn 2002;50(3):366-386.

3. Sohn YH, Hallett M. Motor evoked potentials. Phys Med Rehabil Clin

N Am 2004;15(1):117-131. vii.

4. Chen R. Studies of human motor physiology with transcranial

magnetic stimulation. Muscle Nerve Suppl 2000;9:S26-S32.

5. Ray PG, Meador KJ, Epstein CM, Loring DW, Day LJ. Magnetic stim-

ulation of visual cortex: factors influencing the perception of phos-

phenes. J Clin Neurophysiol 1998;15(4):351-357.

6. Cowey A, Walsh V. Magnetically induced phosphenes in sighted, blind

and blindsighted observers. Neuroreport 2000;11(14):3269-3273.

7. Epstein CM, Lah JJ, Meador K, Weissman JD, Gaitan LE, Dihenia B.

Optimum stimulus parameters for lateralized suppression of speech

with magnetic brain stimulation. Neurology 1996;47(6):1590-1593.

8. Gershon AA, Dannon PN, Grunhaus L. Transcranial magnetic stimu-

lation in the treatment of depression. Am J Psychiatry 2003;160(5):

835-845.

9. O’Reardon JP, Solvason HB, Janicak PG, et al. Efficacy and safety of

transcranial magnetic stimulation in the acute treatment of major

depression: a multisite randomized controlled trial. Biol Psychiatry

2007;62(11):1208-1216.

10. Arana AB, Borckardt JJ, Ricci R, et al. Focal electrical stimulation as

a sham control for repetitive transcranial magnetic stimulation: does it

truly mimic the cutaneous sensation and pain of active prefrontal

repetitive transcranial magnetic stimulation? Brain Stimulation 2008;

1(1):44-51.

11. Borckardt JJ, Walker J, Branham RK, et al. Development and evalua-

tion of a portable sham TMS system. Brain Stimulation 2008;1(1):

52-59.

12. Aguado L. Neuroscience of Pavlovian conditioning: a brief review.

Span J Psychol 2003;6(2):155-167.

13. Logan CA. When scientific knowledge becomes scientific discovery:

the disappearance of classical conditioning before Pavlov. J Hist

Behav Sci 2002;38(4):393-403.

14. Carter RM, Hofstotter C, Tsuchiya N, Koch C. Working memory and

fear conditioning. Proc Natl Acad Sci U S A 2003;100(3):

1399-1404.

15. McLaughlin J, Skaggs H, Churchwell J, Powell DA. Medial prefrontal

cortex and pavlovian conditioning: trace versus delay conditioning.

Behav Neurosci 2002;116(1):37-47.

16. Wasserman EA, Miller RR. What’s elementary about associative

learning? Annu Rev Psychol 1997;48:573-607.

17. Kremer EF. The Rescorla-Wagner model: losses in associative

strength in compound conditioned stimuli. J Exp Psychol Anim Behav

Process 1978;4(1):22-36.

18. Ivkovich D, Stanton ME. Effects of early hippocampal lesions on

trace, delay, and long-delay eyeblink conditioning in developing

rats. Neurobiol Learn Mem 2001;76(3):426-446.

19. Carew TJ, Walters ET, Kandel ER. Classical conditioning in a simple

withdrawal reflex in Aplysia californica. J Neurosci 1981;1(12):

1426-1437.

20. Gorn GJ. The effects of music in adverstising on choice behavior:

a classical conditioning approach. J Marketing 1982;46(1):94-101.

21. Kellaris JJ, Cox AD. The effects of background music in advertising:

a reassessment. J Consumer Res 1989;16:113-118.

22. Domjan M. Pavlovian conditioning: a functional perspective. Annu

Rev Psychol 2005;56:179-206.

23. Wolters A, Sandbrink F, Schlottmann A, et al. A temporally asym-

metric Hebbian rule governing plasticity in the human motor cortex.

J Neurophysiol 2003;89(5):2339-2345.

24. Stefan K, Kunesch E, Cohen LG, Benecke R, Classen J. Induction of

plasticity in the human motor cortex by paired associative stimulation.

Brain 2000;3:572-584. 123 Pt.

25. Braff DL, Geyer MA, Swerdlow NR. Human studies of prepulse inhi-

bition of startle: normal subjects, patient groups, and pharmacological

studies. Psychopharmacology (Berl) 2001;156(2-3):234-258.

26. Chen R, Tam A, Butefisch C, et al. Intracortical inhibition and facili-

tation in different representations of the human motor cortex. J Neuro-

physiol 1998;80(6):2870-2881.

27. Furubayashi T, Ugawa Y, Terao Y, et al. The human hand motor area is

transiently suppressed by an unexpected auditory stimulus. Clin Neu-

rophysiol 2000;111(1):178-183.

28. Luber B, Balsam P, Nguyen T, Gross M, Lisanby SH. Classical condi-

tioned learning using transcranial magnetic stimulation. Exp Brain Res

2007;183(3):361-369.

29. Mishory A, Molnar C, Koola J, et al. The maximum-likelihood

strategy for determining transcranial magnetic stimulation motor

threshold, using parameter estimation by sequential testing is faster

86 K. A. Johnson et al

than conventional methods with similar precision. J ECT 2004;20(3):

160-165.

30. Aou S, Woody CD, Birt D. Increases in excitability of neurons of the

motor cortex of cats after rapid acquisition of eye blink conditioning.

J Neurosci 1992;12(2):560-569.

31. Aou S, Woody CD, Birt D. Changes in the activity of units of the cat

motor cortex with rapid conditioning and extinction of a compound

eye blink movement. J Neurosci 1992;12(2):549-559.

32. Birt D, Aou S, Woody CD. Intracellularly recorded responses of

neurons of the motor cortex of awake cats to presentations of

Pavlovian conditioned and unconditioned stimuli. Brain Res 2003;

969(1-2):205-216.

33. Mottaghy FM, Gangitano M, Krause BJ, Pascual-Leone A. Chronom-

etry of parietal and prefrontal activations in verbal working memory

revealed by transcranial magnetic stimulation. Neuroimage 2003;

18(3):565-575.

34. Pourtois G, Vandermeeren Y, Olivier E, de Gelder B. Event-related

TMS over the right posterior parietal cortex induces ipsilateral

visuo-spatial interference. Neuroreport 2001;12(11):2369-2374.

35. Bohning DE, He L, George MS, Epstein CM. Deconvolution of trans-

cranial magnetic stimulation (TMS) maps. J Neural Transm 2001;

108(1):35-52.

36. Hernandez-Garcia L, Lee S, Grissom W. An approach to MRI-based

dosimetry for transcranial magnetic stimulation. Neuroimage 2007;

36(4):1171-1178.

37. Murphy GG, Glanzman DL. Cellular analog of differential classical

conditioning in Aplysia: disruption by the NMDA receptor

antagonist DL-2-amino-5-phosphonovalerate. J Neurosci 1999;

19(23):10595-10602.

38. Burt T, Lisanby SH, Sackeim HA. Neuropsychiatric applications of

transcranial magnetic stimulation: a meta analysis. Int J Neuropsycho-

pharmacol 2002;5(1):73-103.