Boosting Cognition
Effects of Theta-Burst Transcranial Magnetic Stimulation on the Left Superior
Parietal Lobe on Working Memory and Attention
Cato Drion 5729769
Master Brain and Cognitive Sciences Cognitive Neuroscience Track
University of Amsterdam
Cognitive Neuroscience Group University of Amsterdam Supervisor: Ilja Sligte Co-assessor: Martijn Wokke UvA-representative: Steven Scholte
31 EC
Final date: August 30th, 2012
2
OVERVIEW
OVERVIEW ... 2 ABSTRACT ... 3 INTRODUCTION ... 4 Boosting Cognition ... 4 Hemispheric balance ... 5Theta-burst transcranial magnetic stimulation ... 6
The left Superior Parietal Lobe ... 6
Effects of TBS on working memory and attention ... 9
METHODS ... 10
Subjects ... 10
Task Design ... 10
Working Memory - Capacity ... 11
Working Memory - Distractibility ... 12
Working Memory - Precision ... 13
Lateralized Attention Network Test ... 14
Procedures... 15
Training and localizing the target brain region ... 15
TBS sessions ... 16
Analysis ... 18
RESULTS ... 21
Task performance ... 21
Effect of TBS on Working Memory and Attention... 22
Working Memory Capacity ... 23
Working Memory Distractibility ... 24
Lateralized Attention Network Test ... 25
CONCLUSION AND DISCUSSION ... 28
Boosting Cognition ... 28
Attentional capture and distractibility ... 29
Individual differences ... 31
Future directions ... 33
3
ABSTRACT
This study aimed to investigate the effect of theta burst transcranial magnetic stimulation (TBS) on the left superior parietal lobe (left SPL) on working memory and attention. Six subjects performed working memory and attention tasks in intermittent-, intermediate- and continuous TBS sessions. Results, although not conclusive, point towards the hypothesis that working memory and attention can be influenced by TBS on the left SPL. Our findings suggest implications for attention networks and the role of the SPL. However, experiments with more subjects and better control for individual differences in responses to TBS are required to provide for more insights in the role of the SPL and the effects of TBS on working memory and attention.
4
INTRODUCTION
Boosting Cognition
Over the past decades, cognitive neuroscience research has demonstrated that by stimulating the human brain, cognition can be influenced. Using non-invasive techniques such as transcranial direct current stimulation (tDCS) and transcranial magnetic stimulation (TMS) on healthy human subjects, cognitive performance can be improved. Studying possibilities in this area does not only have a certain science-fiction-like appeal to it, it can also provide insights into the mechanisms underlying human cognition. Important, yet complex cognitive skills such as memory and attention can be studied in a causal way using the proper cognitive tasks and stimulation techniques. This study aimed to investigate if cognition, more specifically working memory and attention processes, can be influenced by manipulating brain activity with theta burst transcranial magnetic stimulation (TBS) on the left superior parietal lobe.
Prior research has shown that subjects improve significantly on various cognitive tasks when activity in the left hemisphere is inhibited with repetitive TMS or tDCS. Snyder et al. (2003) inhibited the left anterior temporal lobe of their subjects with 15 minutes of low-frequency magnetic pulses (low-frequency rTMS). When subjects were asked to draw a dog, a horse, or a face after the rTMS, their schema of drawing became more complex and detailed. In the same study, subjects were tested on their proofreading abilities. During and immediately after rTMS, subjects missed significantly less errors in the text they had to read (Snyder et al., 2003). In another study, Snyder, Bahramali, Hawker, & Mitchell (2006) used the same setup for rTMS procedures and tested subjects on subitizing of number estimation of 50 to 150 objects (small rectangles on a screen). The majority of subjects improved significantly during and immediately after rTMS (Snyder et al., 2006). In a study of Chi, Fregni, & Snyder (2010), tDCS was applied to both anterior temporal lobes, hereby simultaneously inhibiting the left - and stimulating the right hemisphere. Compared to sham tDCS conditions, they found that their subjects improved in a visual memory task by 110%. These studies
5 illustrate that by inhibiting the left hemisphere, performance on cognitive tasks involving detailed visual perception and visual working memory can be improved.
Hemispheric balance
This improved cognition is thought to result from the hemispheric imbalance that is created by inhibition of the left hemisphere. Inhibition of the left hemisphere would, following the interhemispheric rivalry model (Sack, Camprodon, Pascual-Leone, & Goebel, 2005), cause the right hemisphere to compensate, leading to a hemispheric imbalance with a right-hemisphere bias. Allan Snyder (Snyder, 2009) proposed that this hemispheric imbalance underlies the extraordinary skills of savants. Savant syndrome, which is often accompanied by autism, is associated with a left-hemisphere dysfunction, resulting in a right-left-hemisphere bias. Snyder argues that savant skills are latent in everyone and that they can be unlocked by creating a hemispheric imbalance. However, the question remains why hemispheric imbalance can account for improved cognition.
A right-hemisphere bias is thought to influence the global-local dichotomy in perception. Since the left hemisphere is more involved in concept formation and semantic representation whereas the right hemisphere accounts for a more literal representation, a right-hemisphere bias is proposed to underlie altered perception. Moreover, in the healthy brain, concept networks keep perception of detail mostly below the conscious level. Creating a hemispheric imbalance might therefore allow for more detailed and literal perception (Chi et al., 2010; Snyder, 2009). To illustrate this, consider the example of a Navon task (Navon, 1977) in figure 1. It is known that subjects with a disorder in the autistic spectrum, where a right-hemisphere bias is observed, have more difficulty processing global Navon stimuli (Frith & Happé, 1994). Although this might seem unfavourable, being able to focus on detail is an important feature of most savant skills (Snyder, 2009), and more detailed and literal perception can be beneficial to several cognitive skills, including visual working memory.
Furthermore, research on spatial neglect has shown over the past decades that attention networks are highly dependent on hemispheric interplay. Spatial neglect often results from damage
6 in the right hemisphere and top-down attention effects are more correlated with activity in the right – than in the left hemisphere (Corbetta & Shulman, 2011). This indicates that spatial attention is more right-hemisphere oriented and thus would benefit from a right-hemisphere bias.
Inhibiting the left hemisphere and hereby indirectly boosting the right hemisphere causes altered perception and cognition. It has already been shown that visual working memory can benefit from this, but it remains unclear which components are involved. Moreover, attention networks are likely to be affected by hemispheric imbalance. Therefore, this study aimed to investigate the changes in working memory and attention that result from manipulating hemispheric balance, by targeting an important cortical area (superior parietal lobe) in the left hemisphere with theta burst TMS.
Figure 1: examples of Navon task characters. In this task a capital letter (the global stimulus) is made up of many, smaller, other letters (the local stimuli) and subjects are asked to detect either the global or local stimulus.
The left Superior Parietal Lobe
Since we aimed to study the effect of TBS on working memory and attention, we targeted the left Superior Parietal Lobe (left SPL). The SPL is important for both cognitive processes. In attention networks, the SPL is thought to be a crucial component of a dorsal-frontal network responsible for
7 top-down signals in spatial attention. This network, that is thought to be more right-hemispheric biased (Corbetta, Patel, & Shulman, 2008) runs over the dorsal posterior parietal cortex, along the intra parietal sulcus and is important for controlling to which location attention is directed (Corbetta & Shulman, 2002). In addition, the SPL is recruited when spatial attention shifts are made in a stimulus-driven (bottom-up) manner (Corbetta & Shulman, 2011). Furthermore, the SPL has already extensively been indicated to be involved in working memory, specifically in tasks involving visuospatial working memory (Cabeza & Nyberg, 2000; Pessoa, Gutierrez, Bandettini, & Ungerleider, 2002). Hamidi, Tononi, & Postle (2008) established a causal role for the left SPL in spatial working memory through a combined fMRI- and rTMS experiment. From the rationale of Snyder (2009), it would follow that the right SPL would be stimulated indirectly by inhibiting the left SPL and that by stimulating the left SPL, the right SPL would be inhibited. Therefore, we targeted the left SPL with TBS to study effects on working memory and attention.
Theta-burst transcranial magnetic stimulation
As was demonstrated in the previously mentioned studies, repetitive transcranial magnetic stimulation (rTMS) can affect cortical excitability in a non-invasive way in conscious human subjects, hereby influencing cognitive functioning. However, the effects of rTMS are subtle and they fade away fast. In 2005, Huang, Edwards, Rounis, Bhatia, & Rothwell conducted experiments with single short low-intensity bursts of rTMS at 50 Hz, ‘theta burst’ stimulation (TBS). TBS turned out to be a highly effective, longer-lasting form of rTMS.
Three variants of TBS can be differentiated. In intermittent (iTBS), 2 s of bursts is applied every 10 s for 190 s. In intermediate TBS (mTBS), 5 s of bursts is applied every 15 s for 110 s. In continuous TBS (cTBS) a continuous train of bursts is applied during 40 s. In all three variants a total of 600 pulses is applied. After testing with MEPs, it was found that intermediate TBS had no overall effect on excitability whereas continuous TBS suppressed excitability and intermittent TBS facilitated
8 excitability (Huang et al., 2005). Furthermore, 40 s of cTBS was found to suppress cortical excitability for approximately one hour, which is longer than any other form of rTMS (see figure 2).
Figure 2: Adopted from Huang et al. 2005; TBS paradigms and their effects on MEP. In panel A the temporal patterns of c-, i-c-, and mTBS bursts are shown. Panel B shows the effects of TBS on MEP over time.
An MRS study of Stagg et al. (2009) explained the mechanism behind the effect of cTBS by showing that after subjects had received cTBS, relatively more GABA was present in the target area. No significant effect was detected on glutamate or glutamine concentrations. This indicated that the effect of cTBS is dependent on increased inhibition by activating cortical GABAergic interneurons.
Although it has a high pulse frequency, TBS is experienced as less uncomfortable than other rTMS protocols by subjects (Rossi, Hallett, Rossini, & Pascual-Leone, 2009), probably because the amount and intensity of the pulses is much lower. It is considered as safe as other rTMS protocols as long as the safe parameters (stimulating at 80% of the active motor threshold) are used (Oberman, Edwards, Eldaief, & Pascual-Leone, 2012). Hence, TBS appears a safe and effective stimulation technique to influence cognitive processes for a prolonged period of time.
9
Effects of TBS on working memory and attention
Based on prior research about hemispheric imbalance, the SPL and TBS, we proposed that inhibiting the left SPL with cTBS would improve working memory and attention and reversely, stimulating the left SPL with iTBS would reduce working memory and attention. We further aimed to investigate which components of working memory would be influenced, what kind of attention effects the manipulations would have and what role the SPL fulfills in mediating these effects.
To test this, we submitted six healthy human subjects to three working memory tasks and an attention task and analysed the effect of TBS on the left SPL on their performance. Each subject received continuous-, intermittent- and intermediate TBS in distinct sessions. The working memory tasks tested working memory capacity, distractibility and precision; the attention task tested efficiency and reaction speed and allowed for differentiate alerting, orienting and conflict effects of attention.
We expected that continuous TBS on the left SPL would temporarily decrease distractibility and improve capacity and precision in working memory. On the attention task, we expected our subjects to be faster and more efficient after continuous TBS. Intermittent TBS was expected to have inverse effects on all performance measures, whereas intermediate TBS was not expected to have a net effect. Based on the known timespan of TBS effects, we expected all effects to occur in the 40 minutes after stimulation. Beyond 40 minutes after stimulation, the effect was expected to decay and we did not expect to see any effect at 60 minutes after stimulation. By analysing performance on various components of our tasks we evaluated the specific effects of stimulating and inhibiting the left SPL with TBS.
10
METHODS
Subjects
Six adult subjects (one male, five female, age range 22-31) that had already taken part in previous rTMS experiments were recruited from the Faculty of Psychology of the University of Amsterdam. Four subjects were right-handed and two were left-handed. Subjects were screened according to local MRI and TMS safety procedures and informed about the experiment. Every subject participated in a training session, an fMRI session and three distinct TBS sessions and they received monetary compensation for their participation.
All protocols and procedures of this experiment were approved by the Ethics Committee of the Psychology Department of the University of Amsterdam and all participants gave their written informed consent prior to participation in the study.
Task Design
To measure various components of working memory and attention, we designed a test battery consisting of three variants of a change detection task and one lateralized attention network test. The lateralized attention network test (LANT) was based on the LANT of Greene et al. (2008), who in turn based the LANT on the ANT of Fan et al. (2005). In the LANT, subjects fixate on the middle of a screen. On every trial, five arrows appear in a vertical row either left or right from the fixation point. Subjects need to respond as quickly as possible to the direction of the arrow in the middle, which is either up or down. Using different cue conditions and directions of the flanker arrows, orienting, alerting and conflict effects of attention can be derived from this task.
The three variations of the change detection task were based on the working memory paradigm of Pessoa et al. (2002). This change detection task consists of a fixation period, a sample screen, a short delay period and a test screen. Sample screens are made up of an array of rectangles with specific orientations on a solid background. Subjects are instructed to remember the orientation
11 of as much rectangles as possible. After the delay, the test screen appears which is made up of an equally large array of rectangles. On half of the trials, the test array is identical to the sample array (‘no change trials’) and in the other half of the trials, the orientation of one rectangle is changed (‘change trials’). Subjects indicate whether there was ‘no change’ or ‘change’ on each trial.
We designed three variants of this paradigm, roughly resembling the ones Machizawa & Driver (2011) used. These variants were designed to test three different components of working memory: capacity, distractibility and precision. The specific task designs of our working memory and attention tasks are explained below. Trials were programmed with Presentation (Neurobehavioral Systems) and stimuli were made in Matlab (Mathworks, Inc.) and Adobe Illustrator (Adobe, Inc.).
Working Memory - Capacity
In the capacity variant of the change detection task, subjects were instructed to memorize the orientation of as much rectangles as possible. Trials consisted of a fixation period, a sample array, a short delay and a test array. For an illustration of the capacity task design see figure 3. Throughout the whole trial the background of the screen was black (colour: 0, 0, 0) and a fixation cross (font size 24) was positioned in the middle. In the fixation period the fixation cross was white (255, 255, 255) for 1000 ms, before it turned green (0, 255, 0) for 500 ms, signalling the appearance of the sample array. The sample array consisted of a variable number of white rectangles positioned around the white fixation cross. Each of the rectangles was rotated in one of twelve possible orientations. The length of the rectangles was 2.2 degrees of visual angle and the width was 0.5 degrees of visual angle. The number of rectangles, henceforth called set size, varied from 1 to 12 dependent on the subject’s performance. The sample array appeared for 250 ms. After the delay period of 400 ms, displaying only white fixation cross, the test array appeared. The test array was identical to the sample array on half of the trials (‘no change’ trials); on the other half of the trials the orientation of one rectangle was rotated by 90 degrees (‘change’ trial). The test array lasted for maximally 2000 ms, or terminated when subjects responded ‘change’ or ‘no change’ by pressing corresponding buttons. Speakers provided auditory feedback by presenting a high sound when the response was correct and
12 a low sound when the response was incorrect. Termination of the test array was followed by the fixation period (1000 ms) of the next trial. Subjects were instructed to focus on the fixation cross in the middle of the screen throughout the whole trial.
To measure working memory capacity, the set size was varied depending on performance of the subject. Through a built-in staircase procedure, performance was kept roughly at 75 % correct (75 % hits of the total trials). All subjects started out in the practise run with a set size of 8 rectangles. After every tenth trial, the percentage hits was calculated and if it was higher than 80%, an extra rectangle was added to the sample array, increasing the set size by one item. If the percentage hits was lower than 70%, one rectangle was removed from the sample array, decreasing the set size by one item. The maximum set size was set at twelve items. One run of the capacity test consisted of 50 trials.
Figure 3: Trial design of the Capacity task. This is an example of a change trial with set size 6.
Working Memory - Distractibility
In order to study distractibility in working memory we designed a change detection paradigm with three conditions: one with distractors and two without distractors. Following the same configuration as the capacity task, trials consisted of a fixation period, a sample array, a short delay and a test array. For an illustration of the distractibility task design see figure 4. The fixation period and the delay were similar to the capacity task. Exactly as in the capacity trials the sample array consisted of rectangles (2.2 degrees of visual angle long and 0.5 degrees of visual angle wide) positioned around the fixation cross and each rectangle was oriented in one out of twelve possible orientations.
Fixation 1000ms Alert 500 ms Sample array 250ms Delay 400ms Test array 2000ms
13 However, rectangles were red or green instead of white. Red (255, 0, 0) rectangles served as targets and green (0, 255, 0) rectangles served as distractors. In the distractor condition, the sample array consisted of four red and four green rectangles. On half of the distractor trials one red rectangle changed its orientation. Green rectangles never changed and were to be disregarded. In the two control conditions, trials consisted of arrays with either four or eight red rectangles without green distractors. In control trials one red rectangle changed its orientation in half of the trials. The target set size was four or eight in control conditions and always four in the distractor condition. The magnitude of the change was always 45 degrees. The sample array (250 ms) was followed by a 400 ms delay preceding the test array. The test array was visible for maximally 2000 ms or terminated when subjects responded ‘change’ or ‘no change’. Auditory feedback was provided through the speakers. One run of the distractibility test consisted of 60 trials.
Figure 4: Trial design of the Distractibility task. This is an example of a change trial in the distractor condition, with 4 red targets and 4 green distractors.
Working Memory - Precision
To monitor working memory precision, we designed a variant of the change detection task in which always one rectangle changed its orientation. Subjects responded therefore not to whether there was a change or not but in which direction (clockwise or counter clockwise) the rectangle had rotated. Trials resembled the capacity and distractor trials and stimuli were the same rectangles as the ones used in the capacity task. For an illustration of the task design see figure 5. The sample array always consisted of 8 white rectangles, making the set size 8 at every trial. In the test array, only the one rectangle that had changed orientation was shown. In half of the trials the rotation was
Fixation 1000ms Alert 500 ms Sample array 250ms Delay 400ms Test array 2000ms
14 clockwise and in the other half of the trials it was counter clockwise. The change magnitude was 15, 30 or 45 degrees, providing for three conditions of difficulty. The test array lasted for maximally 2000 ms or terminated when subjects responded ‘clockwise’ or ‘counter clockwise’. Auditory feedback was presented through speakers. One run of the precision test consisted of 60 trials.
Figure 5: Trial design of the Precision task. This is an example of a trial where the change was counterclockwise and change magnitude was 45 degrees.
Lateralized Attention Network Test
In the lateralized attention network task, subjects focused on a fixation point in the middle of a screen. On every trial, a vertical row of five arrows would appear either left or right from the fixation point. Subjects were instructed to indicate as quickly as possible whether the middle arrow pointed upwards or downwards by pressing the corresponding button. On no-cue trials, the row of arrows appeared unexpectedly. On cue trials, a cue (a star-shaped figure) occurred either at the fixation point (central cue) or left or right from fixation point (spatial cue). Central cues signalled only when the row of arrows would appear, whereas spatial cues signalled both when and where the row of arrows would appear. Furthermore, the arrows flanking the middle arrow (two above and two below) could either be congruent (pointing in the direction of the middle arrow) or they could be incongruent (pointing in the opposite direction). Reaction time (RT) and accuracy of the response were measured. For an illustration of the trial design, see figure 6. Trials started with a white (255,255,255) screen and a black (0,0,0) fixation cross in the middle of the screen. After 100 ms, a central or spatial cue or no cue was presented for 200 ms, dependent on the cue condition. The cue screen was followed by a delay screen (displaying only the fixation point) of 500 ms before the arrow
Fixation 1000ms Alert 500 ms Sample array 250ms Delay 400ms Test array 2000ms
15 display was shown. Arrows were 0.58 degrees of the screen size, with 0.06 degrees in between. Distance from the middle arrow and the spatial cue to the central fixation point was 1.06 degrees. The arrow display was shown for maximally 1700 ms and disappeared as soon as subjects responded to the direction of the middle arrow. To make the appearance of the cue or arrow display unpredictable, inter trial intervals (ITI) were varied. Possible ITI timings were 3000, 3250, 3500, 3750, 4000, 4250, 4750 or 5500ms. No auditory feedback was presented during LANT trials. One run of the LANT consisted of 48 trials.
Figure 6: Trial Design of the Lateralized Attention Network Test. Example of a spatial cue and incongruent flankers, with the middle arrow pointing upwards.
Procedures
Training and localizing the target brain region
After completing the screenings and signing the informed consent, subjects participated in a training session to practice the change detection task by following the training tasks of Sligte, Scholte, & Lamme (2008) and to familiarize themselves with the LANT. Training sessions were held at the TMS lab to familiarize the subjects with the settings that would be used during TBS sessions. For specifics concerning these settings, see below. After subjects had been sufficiently trained (4 runs of the LANT and change detection trials until they reached a performance of 75% correct, this took approximately 2 hours per subject) they were placed in the MRI scanner where they carried out the LANT and the change detection task to map brain regions involved with attention and working memory processes. fMRI scanning took place in the Spinoza Center of the University of Amsterdam, using the Philips 3T
Fixate 100ms Cue 200 ms Delay 500ms Arrow display 1700ms Variable I T I
16 scanner. Presentation (NeuroBehavioral Systems) was used to present subjects with the tasks (for task specifics see above). Of all subjects a T1 anatomical scan was made, followed by BOLD MRI scans to map activity during the LANT and change detection trials. Total scan time was approximately 60 minutes per subject. Using BrainVoyager QX (Brain Innovation Inc.) scan images were analysed. After the usual pre-processing steps of the imaging data, including realignment, motion correction
(trilinear sinc), spatial smoothing (4mm) and temporal filtering (3 cycles), protocol files based on the logfiles of Presentation using Prolog were created. These protocols were convolved with the BOLD MRI data to create statistical activation maos. From these patterns, we derived the most activated region in the left superior parietal lobe for each subject. For each subject individually, the anatomical scan and location of the SPL were transferred into the neuronavigation computer of the TMS lab. The exact coordinates per subject in Nifti personal space are displayed in table 1.
Subject X Y Z sc0073 -17.0 21.4 111.9 sc0115 -37.0 14.6 102.9 sc0187 -29.0 16.9 90.7 sc0399 -22.1 19.1 118.4 sc1421 -48.3 12.7 83.6 sc1422 -31.4 17.8 101.6
Table 1: coordinates in Nifti personal space (mm) of the identified left SPL per subject. These locations were the targets for TBS.
TBS sessions
In three distinct TBS sessions that were held at least one week apart subjects received continuous TBS (cTBS), intermittent TBS (iTBS) and intermediate TBS (mTBS). The order in which subjects received the types of TBS was counterbalanced over subjects and subjects did not know which type of TBS they received. Subjects performed the LANT and the three working memory tasks before and after TBS manipulation. TBS sessions took place at the TMS labs of the University of Amsterdam. Subjects were placed in a chair in front of a screen of 0.39m by 0.30m that was placed at eye level. The chair was equipped with response buttons at the left and right arm rests. Tasks were presented on the screen using Presentation (NeuroBehavioral Systems). In this setting subjects performed the four tasks and received TBS. Distance to the screen was kept at 1 m throughout the session, making
17 the screen size 21.3 by 16.8 degrees of visual angle. Each session consisted of a practice run and one run of all four tasks before TBS, followed by c-, i-, or mTBS and three test runs directly after TBS. A run always started with 48 LANT trials, followed by respectively 50 capacity trials, 60 distractor trials and 60 precision trials. The practice run at the start of each session served to re-accustom subjects to the tests at the beginning of each session and allow their performance to be at baseline level at the first run. The timeline of one TBS session is depicted in figure 7. The duration of one run of all four tasks was about 20 minutes.
Timeline of TBS session Time to TBS -40 minutes -20 minutes 0 + 0 minutes + 20 minutes + 40 minutes + 60 minutes + 90 minutes Part of Session Practice Run First Run TBS Second Run Third Run Fourth Run End of Session Subject leaves Lab
Prior to administering TBS, right after the first run, the motor threshold of subjects was determined by placing the coil at the motor cortex and observing hand movement at increasing intensities, starting at 30% of machine output. Intensity of the TBS pulses was always at 80% of the subjects’ active MT. Machine outputs for TBS administration were within a range of 33% to 44%; the mean output over all subjects was 39,3%. TBS was administered through a Magstim Rapid² Stimulator (Magstim Co., UK) using a flat 70 mm figure-of-8 coil, and the left SPL was targeted using neuronavigation soft- and hardware. During TBS administration, both experimenters and subjects were given earplugs for hearing protection. Using a handle that could be fixed to the chair subjects were placed in, the coil was kept in place throughout administration. The duration of the administration varied between the three different types of TBS, whereas the number of pulses (600) was the same in all three types. The parameters used for the different forms of TBS and durations can be seen in table 2.
18
TBS parameters: Pulses: Neural Effect:
cTBS Continuous: 40 seconds of pulses. Frequency: 50 Hz Burst Frequency: 5 Hz Number of Pulses: 3 Total of 600 pulses
Inhibits cortical excitability
iTBS Intermittent: 2 seconds of pulses and 8 consecutive seconds rest, repeated 20 times (200 seconds).
Frequency: 50 Hz Burst Frequency: 5 Hz Number of Pulses: 3 Total of 600 pulses
Facilitates cortical excitability
imTBS Intermediate: 5 seconds of
pulses and 10 consecutive seconds rest, repeated 8 times (120 seconds).
Frequency: 50 Hz Burst Frequency: 5 Hz Number of Pulses: 3 Total of 600 pulses
No net effect on cortical excitability
Table 2: the different parameters of c-, i-, and imTBS.
Subjects were asked not to move, however they were not restrained in any way, during TBS administration. They were closely monitored throughout the whole session and at least one of the experimenters was a certified first aid provider. Directly after stimulation subjects proceeded with the three remaining runs of all tasks. After the last run of each session, subjects stayed in the lab and filled out a short questionnaire to indicate whether they experienced anything physical or mental related to the manipulation. They stayed in the lab under our supervision for at least 90 minutes after TBS, to ensure that no effects of TBS were still present and subjects felt well when leaving the lab. Furthermore, they were contacted the day after each session to check how they had felt after the session had ended.After the last session, subjects received a debriefing of the experiment.
Analysis
Performance on the LANT, capacity and distractibility tasks was analysed. Precision trials were excluded from analysis since subjects performed consequently at chance level on all runs; therefore this task was deemed too hard and not further analysed.
For capacity, Cowan’s K (Cowan, 2001) was calculated that indicated performance corrected for set size. This K was calculated per subject per run. The equation can be viewed below (see equation 1).
19
( )
( )
Equation 1: calculation of working memory capacity. The percentage correct over a run was divided by the product of the mean set size of that run and the total set size times 2, minus 1.
Performance on the distractibility task was compared between conditions using a Student’s T-test. To calculate distractibility, the percentage correct on distractor trials was divided by the mean percentage correct over the non-distractor trials, where every percentage correct was multiplied by the set size of targets of those trials (see equation 2). This was done for every run, per subject. The performance on the distractor trials, corrected for set size, was also analysed in itself.
( )
Equation 2: calculation of working memory distractibility. The percentages correct over the three different conditions were first multiplied by the target set size. Then the performance on the distractor trials was divided by the combined performance on the two non-distractor trials.
The LANT was analysed in terms of reaction times and efficiency. The medians of the reaction times in milliseconds over different runs were analysed. The efficiency on LANT runs was calculated as the percentage correct over that run divided by the median of the reaction time in seconds per run. The various conditions (no, central and spatial cue and congruent and incongruent flanker arrows) were compared to each other in terms of efficiency, using a Student’s T-test. The central cue accounts for an alerting attention effect since it provides temporal information about the appearance of the arrow display. The spatial cue accounts for an orienting effect by giving spatial information about the appearance of the arrow display. Congruent and incongruent flankers account for conflict monitoring in attention. By subtracting the results of different conditions from each other, the orienting, alerting and conflict effects were studied. For an overview of the effects in the LANT results, see table 3.
20
LANT effects of conditions
Alerting Effect Efficiency (central cue) – Efficiency (no cue)
Orienting Effect Efficiency (spatial cue) – Efficiency (central cue)
Conflict Effect Efficiency (congruent flanker trials) – Efficiency (incongruent flanker trials)
Table 3: the explanation of the different effects of the LANT conditions, derived from Fan et al. (2005).
The performance on capacity, distractibility and LANT runs was analysed over runs and plotted for the three TBS sessions. The average performance over subjects was analysed over the four runs for each of the TBS conditions, leading for each task to a set of four time points that could be compared between the three TBS types. For each task, the average performance of the first and fourth run was subtracted from the average performance of the second and third run to create a value that represented performance on the two runs after TBS, corrected to baseline. Henceforth these values are denotes as Effect Scores (see equation 3). The Effect Scores were compared between cTBS and iTBS using permutation tests (n permutations = 1,000,000).
( ) ( )
Equation 3: calculation of Effect Scores, used for each task. Effect Scores reflect performance on the two trials after TBS, corrected to baseline.
21
RESULTS
Task performance
The precision variant of the working memory tasks was excluded from analysis since subjects performed consequently at chance level. Performance on capacity and distractibility variants and on the LANT was distributed within a common range. On the capacity task, subjects’ working memory capacity was on average at a K of 3.769 (Cowan’s K; Cowan, 2001). The average set size over all trials and subjects was 8.0 (minimum set size 3.4, maximum set size 12). Performance on the distractibility trials differed between the three conditions of this task (see figure 8). On average, subjects performed better in the condition without distractors and with set size 4 than in the condition with set size 4 and four distractors (p < 0.01, paired two-tail Student’s T-test) and subjects performed better in the condition with set size 4 and four distractors than in the condition with set size 8 and no distractors (p < 0.01, paired two-tail Student’s T-test).
Figure 8: Performance on the different conditions of distractibility
In line with expectations, LANT performance differed between the various cue and flanker conditions. In figure 9, the reaction times and efficiency scores are plotted for all conditions. On average, median reaction times (RT) were longer in no cue trials than in central cue trials (p < 0.01, paired two-tail Student’s T-test), and longer in central cue trials than in spatial cue trials (p < 0.01,
0.86 0.79 0.68 0.00 0.20 0.40 0.60 0.80 1.00 Set size 4 No Distractors Set size 4 Distractors Set size 8 No Distractors
Performance on Distractibility
Conditions
22 paired two-tail Student’s T-test). The average median RT was longer in incongruent flanker trials than in congruent flanker trials (p < 0.01, paired two-tail Student’s T-test). Efficiency scores showed the same trend. Efficiency scores were higher in congruent flanker trials than in incongruent flanker trials (p < 0.01, paired two-tail Student’s T-test). Efficiency scores in the spatial cue trials were higher than in the central- and no cue trials (p < 0.01, paired two-tail Student’s T-test).
Effect of TBS on Working Memory and Attention
The effect of TBS was analysed for the working memory capacity task, the distractibility task and the LANT. For all three tasks, performance was plotted as a function of the runs (run 1 before TBS and
0 100 200 300 400 500 600 700 No Central Spatial A ve rag e R T ( m s)
LANT RT
Congruent Incongruent 0 0.5 1 1.5 2 2.5 3 No Central Spatial Eff ic ie n cy (Acc u rac y/ R T i n s)LANT Efficiency
Congruent IncongruentFigure 9: In the upper panel, LANT Reaction Times are shown (Y-axis represents average of median RT in milliseconds). In the bottom panel, LANT Efficiency scores are shown (Y-axis represents the percentage correct divided by the RT in seconds). In both panels, blue bars represent congruent trials and red lines incongruent trials. Cue conditions are shown on the X-axis.
23 runs 2, 3 and 4 after TBS). For capacity the K-scores were plotted, for distractibility the D-scores and the performance on distractor trials only were plotted and for the LANT the RT and Efficiency scores were plotted, as well as the attention effects on efficiency as explained in table 3 (see Methods). We expected that performance on all tasks would be elevated in the second and third runs following cTBS, whereas it would be decreased in those runs following iTBS. In the fourth run performance was expected to have returned to baseline (the first run) again. We did not expect to see any effect on performance of mTBS.
Working Memory Capacity
Figure 10 shows that performance on the capacity trials declined over the second and third run of iTBS sessions, whereas it slightly improved in the second and third run after cTBS. In the fourth run, both trends seemed to reverse and performance approached the baseline level of the first run. This was in line with expectations, although the change in performance on mTBS sessions was not according to expectations. Permutation testing of the capacity Effect Scores, comparing them between cTBS and iTBS sessions, resulted in a p value of 0.1 (n permutations = 1,000,000).
Figure 10: The effect of TBS on WM capacity. The average K scores (y-axis) are plotted over the four runs (x-axis), where run 1 is before TBS and runs 2, 3 and 4 are after TBS. Different lines represent the different TBS conditions.
2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 4.3 4.5 1 2 3 4 Cap ac ity (K )
Working Memory Capacity
cTBS iTBS mTBS
24 Working Memory Distractibility
In the distractibility task (figures 11 and 12), it appeared that in the third run after cTBS, performance inclined with reference to baseline. In the third run after iTBS performance declined with reference to baseline. In the fourth run after TBS both effects decayed for the D-score (figure 11), yet the performance on distractor trials only (figure 12) remained elevated even in the fourth run after cTBS. The performance in the third runs after c- and iTBS followed our predictions about the effects of TBS. The performance on the second runs however was not in line with expectations. When the Effect Scores on distractibility performance (figure 11) were compared between cTBS and iTBS using permutation testing (n permutations = 1,000,000), this resulted in a p-value of 0.3. For performance on distractor trials only (figure 12), this permutation test resulted in a p-value of 0.3 as well.
Figure 11: The effect of TBS on WM distractibility. The average performance (D-score) is plotted over the four runs (x-axis), where run 1 is before TBS and runs 2, 3 and 4 are after TBS. Different lines represent the different TBS conditions. Performance was calculated by dividing percentages correct of distractor trials by percentages correct of non-distractor trials while correcting for set size.
0.58 0.60 0.62 0.64 0.66 0.68 0.70 0.72 0.74 0.76 0.78 0.80 1 2 3 4 Di str ac tibil ity
Distractibility
cTBS iTBS mTBS25
Figure 12: The effect of TBS on WM distractibility. The average performance on distractor trials only is plotted over the four runs (depicted on the x-axis), where run 1 is before TBS and runs 2, 3 and 4 are after TBS. Different lines represent the different TBS conditions. The y-axis represents the performance on distractor trials only, expressed as the percentage correct responses.
Lateralized Attention Network Test
The effect of TBS on RT and Efficiency in the LANT is shown in figure 13 and 14. The average RT over runs seemed to decrease in the third run after cTBS, whereas the average RT seemed to remain similar in the second and third run after iTBS. When the Effect Scores were compared between cTBS and iTBS sessions using permutation testing, the resulting p-value was 0.6. Efficiency increased in the second run after cTBS and decreased in the third run after iTBS (figure 14). The p-value resulting from permutation testing of the efficiency Effect Scores of cTBS and iTBS sessions was 0.2.
Figure13: The red line represents the average RT on iTBS runs and the blue line represents the average RT on cTBS runs. Runs are depicted on the x-axis. Run 1 is before TBS and 2, 3 and 4 are after TBS.
0.68 0.70 0.72 0.74 0.76 0.78 0.80 0.82 0.84 0.86 1 2 3 4 % Co rr e ct o n D istr ac to r Tr ial s
Distractor Trials
cTBS iTBS mTBS 470 475 480 485 490 495 1 2 3 4 R T ( m s)LANT overall RT
cTBS iTBS26
Figure 14: The red line represents the average efficiency score on iTBS runs and the blue line represents the average efficiency score on cTBS runs. Runs are depicted on the x-axis. Run 1 is before TBS and 2, 3 and 4 are after TBS.
Results of the alerting, orienting and conflict effects of the LANT in terms of efficiency are shown in figure 15, 16 and 17, respectively. The alerting effect, representing the effect of the central cue, seemed to increase after both cTBS and iTBS. However, efficiency scores were higher in the iTBS sessions than in the cTBS sessions. When Effect Scores of the alerting effect were compared between cTBS and iTBS using permutation testing, this resulted in a p-value of 0.5. The orienting effect, which reflects the effect of the spatial cue, showed an increase in the third run after cTBS, but not after iTBS (p = 0.6, n permutations 1,000,000). The conflict effect, representing the effect of (in)congruency of the flanker arrows, increased in the second run after cTBS whereas it decreased in the second run after iTBS. In the third run the effect was similar again for both TBS conditions. The Effect Scores differed significantly between cTBS and iTBS (p < 0.05, n permutations 1,000,000).
1.9 1.95 2 1 2 3 4 Eff ic ie n cy (Acc u rac y/ R T i n s)
LANT overall Efficiency
cTBS iTBS
27
Figure 15: The Alerting Effect plotted over the four runs of cTBS and iTBS sessions. Runs are depicted on the x-axis. Run 1 is before TBS and 2, 3 and 4 are after TBS.
Figure 16: The Orienting Effect plotted over the four runs of cTBS and iTBS sessions. Runs are depicted on the X-axis. Run 1 is before TBS and 2, 3 and 4 are after TBS.
Figure 17: The Conflict Effect plotted over the four runs of cTBS and iTBS sessions. Runs are depicted on the X-axis. Run 1 is before TBS and 2, 3 and 4 are after TBS.
0 0.05 0.1 0.15 0.2 0.25 0.3 1 2 3 4 A ve rag e E ff ic ie n cy Ce n tr al C u e Tr ial s - A ve rag e E ff ic ie n cy N o Cu e Tr ial s
LANT Alerting Effect
cTBS iTBS 0 0.1 0.2 0.3 0.4 1 2 3 4 A ve rag e E ff ic ie n cy S p atial Cu e Tr ial s - A ve rag e E ff ic ie n cy Ce n tr al C u e Tr ial s
LANT Orienting Effect
cTBS iTBS 0 0.1 0.2 0.3 0.4 1 2 3 4 A ve rag e E ff ic ie n cy Co n gr u e n t Tr ial s - A ve rag e E ff ic ie n cy In co n gr u e n t Tr ial s
LANT Conflict Effect
cTBS iTBS
28
CONCLUSION AND DISCUSSION
Boosting Cognition
Since none of the results of the working memory tasks were significant and of the LANT only the conflict effect showed a significant result, we cannot infer one straightforward conclusion from this study. However, the results we display here can be interpreted in the context of our expectations. When looking at the results on working memory it appears that stimulating the left SPL with iTBS lowers performance on two aspects. In both the capacity and distractibility variant, performance drops slightly in the runs after iTBS. After cTBS it appears that performance on both tasks increases. This is in line with our expectations. In the fourth run, where TBS effects are assumed to have ceased, performance returns towards, or even past, baseline levels for the D-scores and K-scores. Although these results are subtle and the differences between c- and iTBS Effect Scores are not significant, these findings could indicate that working memory capacity and distractibility can be slightly boosted when the left SPL is inhibited with cTBS. The small number of subjects in this study could be a factor that prevented results from reaching significance. Possibly, if this study had been carried out with more subjects, we could have concluded that working memory can be boosted with cTBS on the left SPL since our current results do point towards this hypothesis.
Alternatively however, it could be the case that subjects did not improve because of TBS but because of a learning curve within each session. However, the practice trials at the beginning of each session, and the training session each subject received prior to the experiment, would argue against this explanation. In some of the after-session questionnaires, a few subjects declared that they felt more alert in the runs directly after TBS and some subjects reported that they were tired at the end of a session. Hence one could suggest that alertness and fatigue effects could have caused the performance curve in cTBS sessions. However, if these factors caused the effects in the cTBS sessions, it is remarkable that the performance curve is reversed in iTBS sessions. Unfortunately, the mTBS sessions, in which we did not expect any effect on performance, could not provide for more
29 clarity since performance in mTBS sessions was highly inconsistent. Based on our results, we cannot unambiguously confirm the hypothesis that by manipulating the left SPL with cTBS cognition can be boosted, however, our findings do account for some interesting possible explanations.
Attentional capture and distractibility
If the subtle changes in performance on working memory tasks we observe are indeed caused by effects of cTBS and iTBS, the question remains how inhibition and stimulation of the left SPL can account for better and worse working memory performance, respectively. A possible explanation for the fact that stimulating the left SPL increases distractibility and decreases working memory capacity could be that a more active left SPL accounts for a more widened external focus. More widened external focus increases susceptibility for attentional capture and this in turn increases distractibility. This would result in worse performance on capacity tasks, as a more widened focus might make subjects more distracted by increased set sizes. Therefore, the involvement of the left SPL in attentional capture would explain the better performance on the working memory tasks after cTBS and the decrease in performance after iTBS.
The role of the left SPL in attentional capture should then also be reflected in the results on the attention tasks. At first glance, the LANT results suggest that stimulating the left SPL does not enhance overall attention, whereas inhibiting the left SPL does. This was in line with our expectations, although results of overall attention were not significant. Not in line with our expectations however, was the remarkable finding that the alerting effect shows higher scores for all runs in the iTBS sessions than in the cTBS sessions. Although this is not significant, it suggests a stronger effect of the alerting cue when the left SPL is stimulated. The plotted orienting effect displays the inverse; all scores are higher in cTBS sessions than in iTBS sessions, indicating a stronger effect of the spatial cue when the left SPL is inhibited. For the conflict effect a clear and significant difference is visible between cTBS and iTBS sessions that implicates that the effect of (in)congruent flankers is more pronounced after inhibiting the left SPL. The finding that the alerting effect of the
30 LANT seems more pronounced when the left SPL is stimulated is supports the explanation that the SPL is involved in attentional capture. Attention is more efficiently captured by the central cue if the left SPL is more active. The finding that the other attention effects display the inverse (more pronounced orienting and conflict effects after cTBS) can also be explained. Whereas the alerting effect involves a bottom-up process, spatial orienting and conflict monitoring require top-down attention processes. If the SPL is involved in attentional capture, inhibition of the left SPL with cTBS would allow for more efficient orienting and conflict monitoring, which is reflected in our results.
This possible explanation of our findings does imply the left SPL in bottom-up processes. Although earlier literature considers the SPL as part of a top-down network (Corbetta & Shulman, 2002), more recent studies support this claim. De Fockert, Rees, Frith, & Lavie (2004) conducted an fMRI study and found that SPL activity correlated with appearance of distractors in a visual search task. They proposed that the SPL is a neural correlate for attentional capture. Remarkably, in a study by Kanai, Dong, Bahrami, & Rees (2011), subjects showed more susceptibility for attentional capture in the same visual search task when their left SPL was targeted with cTBS. The authors therefore conclude that the left SPL is a top-down controller of attention, directly contradicting our findings. In a more recent review (Corbetta & Shulman, 2011) the SPL is thought to be recruited in both goal-driven and stimulus-goal-driven spatial attention shifts. It could thus be the case that the SPL is involved in both top-down and bottom-up processes. Since the top-down effects of attention are more right-hemisphere biased (Corbetta et al., 2008), the interplay of right-hemispheres is an important factor to consider. In a recent study on visual short term memory, tDCS was used to stimulate the right hemisphere. Subjects were better able to suppress distractors in a working memory task as a result of the stimulation (Tseng et al., 2012). In our study we inhibited the left hemisphere, and this would, according to the theory of Allen Snyder (Snyder, 2009), accomplish the same effect. Considering these findings altogether, one could suggest that our results could either point to the left SPL being involved in attentional capture, or to the right SPL functioning as a top-down controller. However,
31 the possibility that both claims can be justified and an interaction between hemispheres and attention networks accounts for these effects seems plausible and deserves further investigation.
Individual differences
Similar TBS studies can result in contradicting outcomes, as is demonstrated by comparing our results with the results of Kanai et al. (2011). In that study however, no control measurements were taken for individual differences in responses to TBS. Between our subjects we found highly variable effects of TBS manipulations. This was to be expected, based on a recent study that analyzed MEP responses to TBS in 56 subjects (Hamada, Murase, Hasan, Balaratnam, & Rothwell, 2012). Results of this study revealed a high variability in MEP responses to cTBS and iTBS between subjects. Only a quarter of the subjects responded to both cTBS and iTBS as expected, that is, showing facilitating effects of iTBS on cortical excitability and inhibiting effects of cTBS. Half of the subjects showed the expected effects in only either cTBS or iTBS. According to the authors, these differences result from individual differences in recruitment of cortical interneurons (Hamada et al., 2012).
We explored the individual differences of our subjects in terms of the effects of TBS on task performance. We calculated the Effect Scores on the three tasks for each subject individually (see figure 18). Based on our expectations, cTBS sessions would result in positive Effect Scores (for the Effect Score calculations see Methods, formula 3). This is visible in three subjects for the capacity and distractibility task, and in four subjects for LANT Efficiency. iTBS was expected to result in negative Effect Scores. This was found in four subjects for the capacity task and the LANT and in two subjects for the distractibility task. Remarkably, not one subject responded consistently on all tasks, yet on each task at least one subject responded according to expectations to both cTBS and iTBS (see figure 18). Consistent with the findings of Hamada et al. (2012), the probability that one of our subjects responds to cTBS and iTBS as expected, is close to chance level. This might also have applied to the study of Kanai et al. (2011), explaining their deviating results. When interpreting results of TBS studies, it is important to take possible individual differences into account.
32
Figure 18: individual differences in TBS effects on performance. Effect scores are plotted per subject and compared between cTBS and iTBS for all tasks.
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 sc001 sc002 sc003 sc004 sc005 sc006 A ve rag e p e rfor m an ce r u n 2 an d 3 - A ve rag e p e rfor m an ce r u n 1 an d 4
Capacity
cTBS iTBS -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 sc001 sc002 sc003 sc004 sc005 sc006 A ve rag e p e rfor m an ce r u n 2 an d 3 - A ve rag e p e rfor m an ce r u n 1 an d 4Distractibility
cTBS iTBS -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 sc001 sc002 sc003 sc004 sc005 sc006 A ve rag e p e rfor m an ce r u n 2 an d 3 - A ve rag e p e rfor m an ce r u n 1 an d 4LANT Efficiency
cTBS iTBS33
Future directions
Since we cannot infer one straightforward conclusion from our results, future studies with a larger number of subjects are in order to research how hemispheres and attention networks interact to account for effects of TBS on cognition. Considering the high variability in the way subjects respond to TBS manipulations, it would be advisable to predict beforehand, for each subject individually, what effects cTBS and iTBS will have on cortical excitability and behavior. An important question to take into account is whether responses to TBS are generalizable over different cortical areas and behavioral measures. Therefore, TBS studies combined with fMRI in humans and with
electrophysiology in rodents would contribute to this promising line of research, providing insights into the underlying mechanisms of the effects of TBS on cognition.
34
LITERATURE
Cabeza, R., & Nyberg, L. (2000). Neural bases of learning and memory: functional neuroimaging evidence. Current opinion in neurology, 13(4), 415–21. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10970058 Chi, R. P., Fregni, F., & Snyder, A. W. (2010). Visual memory improved by non-invasive brain stimulation. Brain
research, 1353(1998), 168–75. doi:10.1016/j.brainres.2010.07.062
Corbetta, M., Patel, G., & Shulman, G. L. (2008). The reorienting system of the human brain: from environment to theory of mind. Neuron, 58(3), 306–24. doi:10.1016/j.neuron.2008.04.017
Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nature reviews. Neuroscience, 3(3), 201–15. doi:10.1038/nrn755
Corbetta, M., & Shulman, G. L. (2011). Spatial neglect and attention networks. Annual review of neuroscience (Vol. 34, pp. 569–99). doi:10.1146/annurev-neuro-061010-113731
Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–114. doi:10.1017/S0140525X01003922
Frith, U., & Happé, F. (1994). Autism: beyond “theory of mind”. Cognition, 50(1-3), 115–32. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12212920
Hamada, M., Murase, N., Hasan, A., Balaratnam, M., & Rothwell, J. C. (2012). The Role of Interneuron Networks in Driving Human Motor Cortical Plasticity. Cerebral cortex (New York, N.Y. : 1991).
doi:10.1093/cercor/bhs147
Hamidi, M., Tononi, G., & Postle, B. R. (2008). Evaluating frontal and parietal contributions to spatial working memory with repetitive transcranial magnetic stimulation. Brain research, 1230, 202–10.
doi:10.1016/j.brainres.2008.07.008
Huang, Y.-Z., Edwards, M. J., Rounis, E., Bhatia, K. P., & Rothwell, J. C. (2005). Theta burst stimulation of the human motor cortex. Neuron, 45(2), 201–6. doi:10.1016/j.neuron.2004.12.033
Kanai, R., Dong, M. Y., Bahrami, B., & Rees, G. (2011). Distractibility in daily life is reflected in the structure and function of human parietal cortex. The Journal of neuroscience : the official journal of the Society for Neuroscience, 31(18), 6620–6. doi:10.1523/JNEUROSCI.5864-10.2011
Machizawa, M. G., & Driver, J. (2011). Principal component analysis of behavioural individual differences suggests that particular aspects of visual working memory may relate to specific aspects of attention. Neuropsychologia, 49(6), 1518–26. doi:10.1016/j.neuropsychologia.2010.11.032
Navon, David (University of Haifa, I. (1977). Forest Before Trees: The Precedence of Global Features in Visual Perception. Cognitive Psychology, 9, 353–383.
Oberman, L., Edwards, D., Eldaief, M., & Pascual-Leone, A. (2012). Safety of Theta Burst Transcranial Magnetic Stimulation: A systematic review of the literature. Journal of Clinical Neurophysiology, 28(1), 67–74. doi:10.1097/WNP.0b013e318205135f.Safety
Pessoa, L., Gutierrez, E., Bandettini, P., & Ungerleider, L. (2002). Neural correlates of visual working memory: fMRI amplitude predicts task performance. Neuron, 35(5), 975–87. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/12372290
Rossi, S., Hallett, M., Rossini, P. M., & Pascual-Leone, A. (2009). Safety, ethical considerations, and application guidelines for the use of transcranial magnetic stimulation in clinical practice and research. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology, 120(12), 2008– 39. doi:10.1016/j.clinph.2009.08.016
35 Sack, a T., Camprodon, J. a, Pascual-Leone, a, & Goebel, R. (2005). The dynamics of interhemispheric
compensatory processes in mental imagery. Science (New York, N.Y.), 308(5722), 702–4. doi:10.1126/science.1107784
Sligte, I. G., Scholte, H. S., & Lamme, V. a F. (2008). Are there multiple visual short-term memory stores? PloS one, 3(2), e1699. doi:10.1371/journal.pone.0001699
Snyder, A. (2009). Explaining and inducing savant skills: privileged access to lower level, less-processed information. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 364(1522), 1399–405. doi:10.1098/rstb.2008.0290
Snyder, A., Bahramali, H., Hawker, T., & Mitchell, D. J. (2006). Savant-like numerosity skills revealed in normal people by magnetic pulses. Perception, 35(6), 837–845. doi:10.1068/p5539
Stagg, C. J., Wylezinska, M., Matthews, P. M., Johansen-Berg, H., Jezzard, P., Rothwell, J. C., & Bestmann, S. (2009). Neurochemical effects of theta burst stimulation as assessed by magnetic resonance spectroscopy. Journal of neurophysiology, 101(6), 2872–7. doi:10.1152/jn.91060.2008
Tseng, P., Hsu, T.-Y., Chang, C.-F., Tzeng, O. J. L., Hung, D. L., Muggleton, N. G., Walsh, V., et al. (2012). Unleashing Potential: Transcranial Direct Current Stimulation over the Right Posterior Parietal Cortex Improves Change Detection in Low-Performing Individuals. Journal of Neuroscience, 32(31), 10554–10561. doi:10.1523/JNEUROSCI.0362-12.2012
de Fockert, J., Rees, G., Frith, C., & Lavie, N. (2004). Neural correlates of attentional capture in visual search. Journal of cognitive neuroscience, 16(5), 751–9. doi:10.1162/089892904970762
