- Email: [email protected]
Modulating human sense of agency with non-invasive brain stimulation
Accepted Manuscript Modulating human sense of agency with non-invasive brain stimulation Nima Khalighinejad, Patrick Haggard PII:
S0010-9452(15)00139-2
DOI:
10.1016/j.cortex.2015.04.015
Reference:
CORTEX 1456
To appear in:
Cortex
Received Date: 22 January 2015 Revised Date:
10 April 2015
Accepted Date: 18 April 2015
Please cite this article as: Khalighinejad N, Haggard P, Modulating human sense of agency with noninvasive brain stimulation, CORTEX (2015), doi: 10.1016/j.cortex.2015.04.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
Modulating human sense of agency with non-invasive brain stimulation
2
Nima Khalighinejad, Patrick Haggard
4
Institute of Cognitive Neuroscience, University College London, London WC1N 3AR, UK
RI PT
3
5
Correspondence should be addressed to Nima Khalighinejad, Institute of Cognitive Neuroscience,
7
University College London, Alexandra House, 17 Queen Square, London WC1N 3AR, UK. Tel: +44
8
(0)20 7679 5434, E-mail: [email protected]
10
Running title: Human sense of agency
14
15
16
17
EP
13
AC C
12
TE D
11
M AN U
9
SC
6
18
19
20
0
ACCEPTED MANUSCRIPT Abstract
22
Human voluntary actions are accompanied by a distinctive subjective experience termed “sense of
23
agency”. We performed three experiments using transcranial direct current stimulation (tDCS) to
24
modulate brain circuits involved in control of action, while measuring stimulation-induced changes in
25
one implicit measure of sense of agency, namely the perceived temporal relationship between a
26
voluntary action and tone triggered by the action. Participants perceived such tones as shifted towards
27
the action that caused them, relative to baseline conditions with tones but no actions. Actions that
28
caused tones were perceived as shifted towards the tone, relative to baseline actions without tones.
29
This ‘intentional binding’ was diminished by anodal stimulation of the left parietal cortex (targeting
30
the angular gyrus (AG)), and, to a lesser extent, by stimulation targeting the left dorsolateral
31
prefrontal cortex (DLPFC), (Experiment 1). Cathodal AG stimulation had no effect (Experiment 2).
32
Experiment 3 replicated the effect of left anodal AG stimulation for actions made with either the left
33
or the right hand, and showed no effect of right anodal AG stimulation. The angular gyrus has been
34
identified as a key area for explicit agency judgements in previous neuroimaging and lesion studies.
35
Our study provides new causal evidence that the left angular gyrus plays a key role in the perceptual
36
experience of agency.
37
Keywords: sense of agency; angular gyrus; dorsolateral prefrontal cortex; intentional binding; tDCS
39
40
41
SC
M AN U
TE D
EP AC C
38
RI PT
21
42
43
44 1
ACCEPTED MANUSCRIPT 1. Introduction
46
Healthy human adults have the feeling that they are able to control their own actions, and, through
47
them, external events. This is referred to as the sense of agency. Sense of agency is central to
48
individual goal-directed action, and also to social responsibility and punishment (Frith, 2014; Moretto,
49
Walsh, & Haggard, 2011). Moreover, many neurological and psychiatric disorders involve
50
abnormalities of agency (de Jong, 2011; Fletcher & Frith, 2009; Kranick & Hallett, 2013).
51
Despite extensive theoretical work on agency, its neural correlates are not fully understood.
52
Neuroimaging studies found activation of AG (Farrer et al., 2003; Farrer & Frith, 2002; Farrer et al.,
53
2008) and DLPFC (Fink et al., 1999) associated with agency tasks, but the activation of these areas
54
was always greater in the conflicting, non-agency condition than in the agency condition. In a recent
55
meta-analysis of sense of agency, the single most consistent result was activation of a broadly-defined
56
temporoparietal junction area conditions associated with reduced or absent sense of agency (Sperduti,
57
Delaveau, Fossati, & Nadel, 2011). This broad ‘non-agency’ area includes AG. Computational
58
models of predictive motor control offer an important theoretical framework for understanding
59
agency. An internal forward model uses efference copies of the motor command to predict outcomes
60
(Wolpert & Ghahramani, 2000). According to these models, sense of agency arises when there is a
61
match between the predicted and actual sensory outcome of the generated action. Conversely, if
62
current sensory information does not match the model’s prediction, then the corresponding sensory
63
event cannot be self-generated, and no sense of agency is experienced (Frith, Blakemore, & Wolpert,
64
2000).
65
Farrer et al., (2008) used this framework to interpret fMRI activations of AG in particular, suggesting
66
that AG processes discrepancies between intended action and its actual consequences. Her data
67
showed increased activations of AG when a detectable temporal discrepancy was inserted between an
68
action and visual feedback of the outcome, and also when participants explicitly rejected agency over
69
the viewed outcome.
AC C
EP
TE D
M AN U
SC
RI PT
45
2
ACCEPTED MANUSCRIPT Most of these studies used explicit agency attribution tasks, in which participants judge whether they
71
did or did not cause a specific sensory event. Synofzik, Vosgerau, & Newen, (2008) noted that one
72
feels a sense of agency when acting, even without making any explicit judgements. One suitable
73
measure of this pre-reflective, sensorimotor feeling of agency is the perceived temporal relationship
74
between a voluntary action and its sensory outcome (Moore & Obhi, 2012). The perceived time of
75
voluntary actions and their sensory consequences are attracted towards each other. This ‘intentional
76
binding’ is absent, or less prominent, for involuntary movements, and for associations between
77
external events not involving voluntary actions (Cravo, Claessens, & Baldo, 2009).
78
The neural bases of such feelings of agency are poorly understood. One neuroimaging study found a
79
neural correlate of intentional binding in the medial frontal cortex (Kühn, Brass, & Haggard, 2013). A
80
‘virtual lesion’ study showed that theta-burst stimulation over a slightly more anterior medial frontal
81
location reduced the intentional binding effect (Moore, Ruge, Wenke, Rothwell, & Haggard, 2010).
82
On the other hand, other lesion (Sirigu et al., 2004) and stimulation (Desmurget et al., 2009) studies
83
suggested an important role of parietal cortex in intentional action and agency, though these studies
84
did not use binding. To our knowledge, no previous causal study has investigated the influence of
85
both frontal and parietal areas on sense of agency using implicit measures. We therefore performed
86
three transcranial direct current stimulation (tDCS) experiments, to modulate excitability of key brain
87
circuits underlying the control of action, while measuring the effects on sense of agency, using
88
intentional binding. Our experiments investigated the respective contributions of parietal and frontal
89
areas to intentional binding as a proxy measure of agency (Experiment 1), their susceptibility to both
90
up- and down-regulation (Experiment 2), and their hemispheric specialisation (Experiment 3).
91
Based on the existing neuroimaging data investigating explicit agency judgement (Farrer et al., 2003;
92
Farrer & Frith, 2002; Farrer et al., 2008), we predicted that anodal stimulation of putative AG should
93
also influence the sense of agency, as measured by intentional binding. Importantly, such a result
94
would identify a causal role for AG in sense of agency, but would not conclusively identify how AG
95
computes agency. We also investigated the role of prefrontal areas in sense of agency. Studies of
96
frontal contributions to sense of agency are more equivocal. Neurostimulation (Moore et al., 2010)
AC C
EP
TE D
M AN U
SC
RI PT
70
3
ACCEPTED MANUSCRIPT 97
and neuroimaging (Kühn et al., 2013) studies of intentional binding found evidence for medial
98
prefrontal involvement, but studies of explicit agency judgements in tasks requiring a choice between
99
alternative actions (Chambon, Wenke, Fleming, Prinz, & Haggard, 2013) identified a more lateral prefrontal focus. DLPFC has also been identified as a key area for initiation (Jahanshahi et al., 1995)
101
and monitoring of voluntary action (Rowe, Hughes, & Nimmo-Smith, 2010). Given the relative
102
inaccessibility of medial prefrontal cortex to neurostimulation, we focussed here on the lateral
103
prefrontal cortex. The stimulations targeted primarily the left hemisphere, and participants made
104
actions with their right hand (experiments 1,2), or with either hand (experiment 3).
105
2. Materials and Methods
106
2.1. Participants
107
In total 55 healthy volunteers, 18-35 years of age (25 females) were recruited from the Institute of
108
Cognitive Neuroscience subject data pool for three separate experiments. All participants were right
109
handed, had normal or corrected to normal vision, had no history or family history of seizure, epilepsy
110
or any neurologic or psychiatric disorder and did not have any metallic or electronic object in the
111
head. Participants affirmed that they had not participated in any other brain stimulation experiment in
112
the last 48 hours, nor had consumed alcohol in the last 24 hours. The sample for Experiment 1
113
consisted of 18 participants (8 females), Experiment 2 consisted of 19 participants (10 females) and
114
Experiment 3 consisted of 18 participants (7 females). One participant failed to finish Experiment 2
115
due to lack of concentration, and was therefore excluded. Experimental design and procedure were
116
approved by the UCL research ethics committee, and followed the principles of the Declaration of
117
Helsinki. Transcranial stimulation followed established safety procedures (Nitsche et al., 2003;
118
Poreisz, Boros, Antal, & Paulus, 2007). Participants were paid a minimum amount for participating in
119
each session of the experiment. Participants were paid a small additional bonus at the end of the last
120
session.
121
2.2. Behavioural task
AC C
EP
TE D
M AN U
SC
RI PT
100
4
ACCEPTED MANUSCRIPT We used intentional binding paradigm as an implicit measure of agency (Fig. 1). The task was based
123
on previous studies (Haggard, Clark, & Kalogeras, 2002), and was programmed in LabVIEW 2012
124
(Austin, Texas). Participants viewed a clock hand rotating on a computer screen which was located
125
60cm in front of the participants in a quiet room. The initial clock position was random. Clock
126
rotation was initiated by participants pressing the return key on a keyboard. Each full rotation lasted
127
2560ms. Participants were instructed to look at the centre of the clock. They made voluntary actions,
128
when instructed, by pressing the return key with their right index finger (Experiments 1, 2), or by
129
pressing F9 or F4 with their right or left index finger, respectively (Experiment 3). Participants chose
130
for themselves when to make these voluntary actions. After each key press, the clock hand stopped at
131
a random location, participants made a time judgement according to condition (see later). Each
132
experimental session consisted of four types of trials, presented in separate blocked and randomised
133
conditions. At the beginning of each block, brief instructions for the relevant condition were displayed
134
on the screen. In the baseline action condition, participants had to press the key at a time of their own
135
free choice. The clock hand stopped after 1500-2500ms (at random), and participants then judged the
136
clock hand position at the time of their key press, entering their response on the keyboard. In this
137
condition, the participant’s actions produced no sensory outcome. In the baseline tone condition,
138
participants were instructed to look at the clock but not to press any key. While the clock was rotating,
139
a pure tone (1000Hz, 100ms duration) was played over a loudspeaker, 1750-4000ms (at random) after
140
the onset of the trial. Participants were then asked to judge the clock hand position at the time of the
141
tone. In the operant action condition, participants pressed a key at a time of their own choosing, and
142
each keypress produced a tone after 250ms. Participants judged the clock hand position at the moment
143
of pressing the key. Finally, the operant tone condition was similar to the operant action condition,
144
with the difference that participants had to judge the clock hand position at the time of the tone. Each
145
condition was tested in a separate block of 30 trials. The order of the blocks was randomised and there
146
was a 1 minute break between each block.
147
This common basic design was slightly changed according to the demands of each specific
148
experiment. In Experiment 3, text below and above the clock instructed participants to reply with
AC C
EP
TE D
M AN U
SC
RI PT
122
5
ACCEPTED MANUSCRIPT either their left or right index fingers. The order of hands was randomised in each block. The block
150
length was increased by 40 trials to allow sufficient trials for analysis of each hand’s data.
151
Before each experiment, participants were trained and familiarised with the task. They were reminded
152
to look at the centre of the clock, to avoid following the clock hand with their eyes, to be spontaneous
153
in their key presses and to be as precise as possible in their judgements, in particular not confining
154
themselves to those numbers 5,10,15… marked on the clock face. Each experimental session
155
consisted of four blocks and took approximately 20 minutes. The short duration of each individual
156
session was planned to coincide with the known effective period of tDCS.
157
2.3. tDCS
158
Direct current stimulation was delivered by StarStim noninvasive wireless neurostimulator
159
(Neuroelectrics, Barcelona, Spain). Circular rubber electrodes (25cm2) were covered in saline-soaked
160
sponges, installed in a 27 channel neoprene cap, and connected to a wireless current generator. tDCS
161
was then controlled by Neuroelectrics Instrument Controller (NIC v1.2) through a separate computer.
162
Current strength was set at 1mA in all experiments, generating a current density of 0.04mA/cm2 at the
163
scalp surface. For each experiment, all participants underwent three separate sessions of tDCS, two
164
effective stimulations and one sham session. The order of the sessions was randomised and
165
counterbalanced across participants. There was a minimum of 48 hours (and a maximum of 1 week)
166
between each stimulation session to minimise any potential carry over effects of tDCS (Nitsche et al.,
167
2008). The duration of stimulation in each session was set at 25 minutes, including 30s to ramp-up
168
and down the stimulating current. For the sham condition, electrical current was only applied during
169
the first and last 30 seconds of the stimulation, so as to induce the same cutaneous sensation as real
170
stimulation, and thus blind the participants as to stimulation condition. During the first 5min of each
171
stimulation, participants were asked to relax on their seats and close their eyes. This delay was
172
designed to allow potential neuro-modulatory effects to build up (Zwissler et al., 2014). Next,
173
participants began the behavioural task while stimulation continued (Fig. 2). All participants finished
174
the behavioural task approximately after 20min, the same time as the end of stimulation. In case
175
participants finished the task prior to the end of stimulation they were asked to remain seated until the
AC C
EP
TE D
M AN U
SC
RI PT
149
6
ACCEPTED MANUSCRIPT end of the stimulation. In case the task outlasted the stimulation, they continued to perform the task
177
without further stimulation. The task period never exceeded the stimulation period by more than 2
178
minutes.
179
Fig. 2 shows tDCS montages of the three experiments. In Experiment 1, the anodal electrode was
180
placed on the left DLPFC (F3 according to the 10/20 international EEG electrode placement) or
181
putative left AG (position P3) (Okamoto et al., 2004; Spitoni et al., 2013) in separate sessions. During
182
the sham session, the position of the stimulating electrode was counterbalanced between F3 and P3. In
183
all three sessions, the return electrode (cathodal) was placed on the right supraorbital area. For
184
Experiment 2, anode and cathode were placed on the putative left AG in separate sessions while the
185
return electrode was placed on the right supraorbital area. This arrangement was retained during the
186
sham session. Experiment 3 used a biparietal montage (Cohen Kadosh, Soskic, Iuculano, Kanai, &
187
Walsh, 2010; Hecht, Walsh, & Lavidor, 2010). For anodal stimulation of the putative left AG, the
188
anode was placed over P3 and cathode was placed over P4. This arrangement was reversed for anodal
189
stimulation of the putative right AG. For sham stimulation, the anode was pseudorandomly placed
190
either at P3 or P4. After each session participants were asked as part of debriefing if they had
191
experienced any notable effects of stimulation. No effects were reported other than mild tingling
192
sensations localised to the electrodes.
193
2.4. Data analysis
194
The difference between the judged clock hand position and the actual onset of the corresponding
195
event was calculated, giving a judgement error for each trial. A perceptual delay was represented by a
196
positive judgement error, and an anticipation by a negative judgement error. The mean and standard
197
deviation of the judgement errors across trials were then measured for each condition. Action binding
198
was defined as the shift of action toward its outcome, and was calculated by subtracting each
199
participant’s mean judgement error in the baseline action from that in the operant action condition.
200
Likewise, tone binding was defined as a shift in the perceived time of a tone towards the action that
201
caused it. Tone binding was calculated by subtracting each participant’s mean judgement error in
AC C
EP
TE D
M AN U
SC
RI PT
176
7
ACCEPTED MANUSCRIPT baseline tone condition from that in the operant tone condition. Thus, perceptual association of an
203
action with a subsequent tone produced a positive value for action binding, and a negative value for
204
tone binding. We analysed action and tone binding separately, since there is evidence from both
205
cognitive studies (Wolpe, Haggard, Siebner, & Rowe, 2013), and previous neurostimulation studies
206
(Moore et al., 2010), that they are driven by distinct mechanisms.
207
Some participants were excluded because of highly variable time judgement. A standard deviation of
208
judgement error across trials of over 250ms in any condition was used as a marker of poor time
209
perception. As in previous intentional binding experiments (Haggard, Aschersleben, Gehrke, & Prinz,
210
2002), these participants were excluded. On this basis, two participants were excluded from
211
Experiment 1, two from Experiment 2, and four from Experiment 3. Importantly, these exclusion
212
criteria are orthogonal to the mean judgement errors used for statistical inference.
213
In Experiments 1 and 2, inferential statistics were based on one-way repeated measures ANOVA,
214
with paired-sample t-tests for follow-up testing. Because our ANOVA had only 3 levels, Bonferroni
215
correction was not required for follow-up testing after a significant result (Cardinal & Aitken, 2013;
216
Meier, 2006). Additionally, in Experiment 1, we used linear discriminant analysis to determine which
217
percepts were most strongly affected by stimulation condition. Experiment 3 used repeated measures
218
ANOVA, with the additional factor of acting hand (right hand responses vs. left hand responses). A
219
final pooled analysis was performed to compare effects of stimulation common to all conditions.
220
3. Results
221
3.1. Experiment 1: frontal vs parietal anodal stimulation
222
This experiment compared the effects of frontal (targeting left DLPFC) and parietal (targeting left
223
AG) cortex stimulation on intentional binding for actions and tones. One-way repeated measures-
224
ANOVA with the factor of stimulation type (anodal frontal vs. anodal parietal vs. sham) showed that
225
action binding was not significantly affected by the type of stimulation (F(2, 30)=1.90, p=0.17,
226
η2=0.11). However, an identical ANOVA on tone binding showed significant differences (F(2,
227
30)=4.30, p=0.02, η2=0.22). Follow-up testing showed that tone binding was significantly reduced by
AC C
EP
TE D
M AN U
SC
RI PT
202
8
ACCEPTED MANUSCRIPT anodal stimulation of the putative left AG compared to sham (t(15)=2.67, p=0.02, d=0.43). Anodal
229
stimulation of the left DLPFC showed a clear trend to reduce tone binding, which approached the
230
border of conventional significance (t(15)=2.07, p=0.06, d=0.42). There was no significant difference
231
between the frontal and parietal stimulation (t(15)=-0.10, p=0.92, d=0.02) (Fig. 3; see also
232
supplementary table A.1-6).
233
We additionally performed the same analysis using median rather than mean judgement error in each
234
condition (see supplementary table B.1-3), since median measures are more robust than means to the
235
influence of outliers. The patterns of statistical significance were unchanged.
236
Finally, to confirm that anodal stimulation of putative AG affected primarily the operant tone
237
condition and not the baseline tone, the effect of stimulation type on participants’ judgement error in
238
baseline tone conditions was assessed using repeated-measure one-way ANOVA. There was no
239
significant main effect of stimulation type on baseline tone condition (F(2, 30)=0.58, p=0.56,
240
η2=0.04). This suggests that stimulation influenced a neurocognitive process that is present primarily
241
in the operant condition.
242
We additionally applied multivariate linear discriminant analysis (Krzanowski, 2000) to identify the
243
linear combination of action binding and tone binding variables that optimally discriminates the
244
different stimulation conditions. Linear discriminant analysis significantly differentiated the three
245
stimulation conditions (Wilks' Lambda=0.59, approx. F(4,58)=4.36, p<0.01). Inspection of canonical
246
coefficients showed that this difference was primarily due to tone binding (standardized canonical
247
coefficient 1.86) rather than action binding (-0.93) (The scores of the individual participants on the
248
first discriminant variate are shown in supplementary Fig. A.1). Post-hoc comparisons between
249
conditions showed a highly significant difference between parietal and sham stimulation (p<0.01;
250
standardised coefficients -1.03 for action binding, 2.19 for tone binding), and also a significant
251
difference between frontal and sham stimulation (p=0.04; standardised coefficients -0.82 for action
252
binding, 1.55 for tone binding). Interestingly, the frontal effect thus involved a slightly larger action
253
binding coefficient, considered relative to the tone binding coefficient, than did the parietal effect,
AC C
EP
TE D
M AN U
SC
RI PT
228
9
ACCEPTED MANUSCRIPT though no inferential statistics can be applied to this ratio. Frontal and parietal stimulation did not
255
differ significantly (p=0.74).
256
3.2. Experiment 2: anodal vs cathodal parietal stimulation
257
If anodal stimulation boosts activity in the left AG then cathodal stimulation of the same area should
258
lead to its suppression. Thus, if anodal stimulation decreases intentional binding, cathodal stimulation
259
should increase it. To test this hypothesis, putative left AG was exposed to anodal, cathodal or sham
260
stimulation in different sessions, and effects on intentional binding were evaluated.
261
One-way repeated measures-ANOVA with the factor of stimulation type (anodal parietal vs. cathodal
262
parietal vs. sham) was used for analysis. Action binding was not significantly affected by the type of
263
stimulation (F(2, 30)=0.50, p=0.61, η2=0.03). Tone binding was also unaffected by type of stimulation
264
(F(2, 30)=0.45, p=0.64, η2=0.03), contrary to our predictions from Experiment 1. Nevertheless, the
265
numerical effect of anodal stimulation of the putative left AG was in the same direction as Experiment
266
1, namely a decreased tone binding compared to sham and cathodal stimulation (Fig. 4; see also
267
supplementary table A.7-12).
268
We additionally performed the same analysis using median rather than mean judgement error in each
269
condition (see supplementary table B.4-6). The patterns of statistical significance were unchanged.
270
3.3. Experiment 3: left vs right parietal stimulation
271
Experiment 3 aimed to replicate the effects of Experiment 1, and additionally investigated the
272
lateralisation of intentional binding using a biparietal montage. The biparietal montage may provide a
273
higher local current density, because of the relatively short path between anode and cathode. In
274
addition, this montage controls for any possible effect of cathodal stimulation of prefrontal areas that
275
may occur with the conventional supraorbital placement of the cathode. Therefore, putative left and
276
right AG were exposed to anodal stimulation in separate sessions, and a third session involved sham
277
stimulation. We also investigated whether the putative AG involvement in intentional binding is hand-
278
specific or hemisphere-specific, by asking participants to make actions with either the left or right
AC C
EP
TE D
M AN U
SC
RI PT
254
10
ACCEPTED MANUSCRIPT hand, choosing randomly on each trial. Analysis of action binding showed no significant main effect
280
of stimulation type (F(2, 26)=0.06, p=0.94, η2=0.01) or acting hand F(1, 13)=0.10, p=0.76, η2=0.01)
281
and no significant interaction (F(2, 26)=0.89, p=0.42, η2=0.06). Analysis of tone binding showed a
282
highly significant main effect of stimulation (F(2, 26)=5.93, p<0.01, η2=0.31). Follow-up testing
283
showed that anodal stimulation of the putative left AG significantly decreased tone binding relative to
284
both sham stimulation (t(13)=2.55, p=0.02, d=0.40), and relative to anodal stimulation of the putative
285
right AG (t(13)=2.90, p=0.01, d=0.56). No significant difference was observed between anodal
286
stimulation of the putative right AG and sham (t(13)=-0.75, p=0.47, d=0.10) (Fig. 5; see also
287
supplementary table A.13-18). Acting hand had no significant main effect on tone binding (F(1,
288
13)=0.01, p=0.94, η2<0.01) and no interaction was observed between the stimulation and acting hand
289
(F(2, 26)=0.15, p=0.86, η2=0.01).
290
We additionally performed the same analysis using median rather than mean judgement error in each
291
condition (see supplementary table B.7-9). The patterns of statistical significance were unchanged.
292
To check whether the decrease in tone binding was primarily due to shifts in the operant tone, or in
293
the baseline tone condition, participants’ judgement errors in the baseline tone condition were
294
compared across the stimulation groups. Analysis showed no significant main effect of stimulation
295
type on baseline tone condition (F(2, 26)=0.50, p=0.60, η2=0.04).
296
3.4. Pooled data: anodal parietal vs sham stimulation
297
Anodal stimulation of the putative left AG was common to all three experiments reported here, as was
298
a sham stimulation condition, although the experiments differed in other respects. Therefore, we
299
pooled the data in these specific conditions across the 46 participants (21 female) from the three
300
experiments, in a single analysis. We found that anodal stimulation of putative left AG significantly
301
reduced the perceptual shift of tone toward action compared to sham (t(45)= 3.28, p<0.01, d=0.35),
302
but had no effect on action binding (t(45)=-1.37, p=0.18, d=0.22).
303
The anodal montage in the third session used a different return (cathode) location compared to the
304
first and second experiments. Our decision to pool data of the three studies was based on the common
AC C
EP
TE D
M AN U
SC
RI PT
279
11
ACCEPTED MANUSCRIPT placement of the anode across these three experiments, and the general observation that cathodal
306
effects on cognitive function are rare (Jacobson, Koslowsky, & Lavidor, 2012). In that case,
307
differences in cathode location may be relatively unimportant, and need not prevent pooling across
308
studies. However, because we cannot entirely exclude some contribution of cathodal location to our
309
main results, we ran a further pooled analysis using the left anodal stimulation conditions of
310
experiments 1 and 2 only, which share a supraorbital cathode location, but excluding experiment 3.
311
This analysis again found that anodal stimulation of putative left AG significantly reduced the
312
perceptual shift of tone toward action compared to sham (t(31)= 2.45, p=0.02, d=0.35), but had no
313
effect on action binding (t(31)=-1.48, p=0.15, d=0.32).
314
To investigate the generality of the anodal AG effect across experiments, we performed a mixed
315
ANOVA with a between-subject factor of experiments (1, 2 or 3), and a repeated-measures factor of
316
stimulation type (anodal left AG vs. sham). The main effect of stimulation type (F(1, 43)=10.7,
317
p<0.01, η2=0.20) recapitulating the pooled t-test reported above. There was no significant main effect
318
of experiment (F(2, 43)=0.80, p=0.45, η2=0.04). Importantly, there was no hint of interaction between
319
experiment and stimulation (F(2,43)=0.96, p=0.39, η2=0.04).
320
Our main inferences above are based on comparing experimental stimulation with sham. We therefore
321
additionally investigated whether sham stimulation had different effects in the three experiments. We
322
found no significant difference among the sham conditions for the 3 experiments for action binding
323
(F(2,43)=2.20, p=0.12), or tone binding (F(2,43)=1.10, p=0.33).
324
4. Discussion
325
4.1. Stimulation-induced modulation of the left parietal cortex
326
We performed a series of three tDCS experiments to investigate the neural circuits responsible for the
327
sense of agency, as measured by the perceptual association between the time of a voluntary action and
328
the time of a resulting auditory tone. We found a significant decrease in the binding of outcomes
329
towards actions after anodal stimulation of the putative left AG (Experiment 1). Anodal stimulation of
330
the left DLPFC also decreased action and tone binding compared to sham. DLPFC affected tone
AC C
EP
TE D
M AN U
SC
RI PT
305
12
ACCEPTED MANUSCRIPT binding as much as AG stimulation, but its effects were less consistent across participants.
332
Nevertheless, our discriminant analysis showed a significant effect of DLPFC stimulation compared
333
to sham, when action and tone binding were considered together.
334
Our tDCS stimulation of putative AG could have widespread effects across the inferior parietal cortex
335
(IPC), since tDCS has quite low spatial specificity. For example, anterior parts of the IPC may also be
336
affected. IPC is routinely activated in neuroimaging studies when participants judge whether their
337
own action, or some other cause, is responsible for a specific sensory event. In a study by Farrer and
338
Frith (2002), the IPC area was more active when participants attributed a visual event to another
339
person, rather than to themselves. Similarly, a PET study (Farrer et al., 2003) observed that neural
340
activity in IPC increased with the level of discrepancy between the executed and the observed action
341
on the screen. In an fMRI study (Farrer et al., 2008), the subjective feeling of loss of control
342
correlated with BOLD response in the AG, as did the awareness of temporal discrepancy between
343
action and feedback. The authors of those studies suggested that AG houses the comparison between
344
the efference copy of the intended action and the actual sensory outcome. Any mismatch between
345
these signals will then give rise to the explicit awareness of non-agency, or an external source of
346
action.
347
Our overall results are consistent with this view. We found that anodal stimulation of the putative AG
348
decreases intentional binding, our proxy measure for agency. Anodal stimulation is generally thought
349
to increase the activity of the cortical region immediately under the electrode. However, AG
350
activation is routinely associated with lack of agency, rather than with experience of positive agency
351
(Farrer & Frith, 2002; Farrer et al., 2008; Sperduti et al., 2011). Therefore, excitation of a neural
352
substrate of non-agency might be expected to decrease intentional binding. The conventional polarity-
353
specific (anode-boosting, cathode-suppressing) framework of tDCS was developed on the basis of
354
effects in primary motor cortex stimulation (Nitsche & Paulus, 2000). Its applicability to non-primary
355
areas and cognitive processing has recently been questioned (Horvath, Forte, & Carter, 2014).
356
Nevertheless, our findings are broadly compatible with the conventional polarity-specific view.
AC C
EP
TE D
M AN U
SC
RI PT
331
13
ACCEPTED MANUSCRIPT 4.2. Stimulation-induced modulation of the left frontal cortex
358
Since DLPFC is normally thought to facilitate intentional action, one may question why prefrontal
359
anodal tDCS did not increase intentional binding. Rowe et al., (2010) questioned whether DLPFC
360
played any important role in initiation of simple voluntary actions, such as those tested here, and
361
suggested a role in monitoring sequential action patterns instead. Fink et al., (1999) observed
362
activation of DLPFC using PET when an intentional action and its sensory outcome were
363
incompatible. Anodal tDCS over DLPFC might correspond to an increased coding for action-outcome
364
conflict, even though our task did not explicitly manipulate action-outcome compatibility. Our
365
discriminant analysis found some evidence consistent with this interpretation. However, this effect
366
was investigated in only one experiment, and achieved statistical significance in multivariate analysis,
367
but not in univariate analyses of action binding and tone binding separately. Therefore, further
368
research is required before a strong statement about frontal tDCS effects on sense of agency can be
369
made.
370
4.3. Polarity-specific effects of tDCS
371
Experiment 2 aimed to investigate whether parietal stimulation effects were polarity-specific. On one
372
model, tDCS would simply add neural noise, irrespective of polarity. On another model, anodal
373
stimulation would upregulate putative non-agency coding in AG, while cathodal stimulation should
374
down-regulate it. The result of anodal left AG stimulation in experiment 2 followed the expected
375
trend for tone binding, but did not reach statistical significance. Replication of statistically significant
376
results is an important and controversial issue in modern neuroscience (Cumming, 2005). All effects
377
measured in experiments represent a combination of the underlying ‘true’ effect, and noise.
378
Importantly, when a nonzero true effect indeed exists, but is modest in size, it is quite likely for the
379
effect to reach statistically significant levels in one study, but not in another. Thus, absence of a
380
significant anodal tDCS effect in experiment 2 does not prove that no true effect exists: we return to
381
this point later.
AC C
EP
TE D
M AN U
SC
RI PT
357
14
ACCEPTED MANUSCRIPT Experiment 2 found no significant difference between cathodal and sham stimulation, although we
383
had predicted that cathodal stimulation might enhance intentional binding. Although inhibitory
384
cathodal effects on motor function are well established, a recent review of 34 studies found that
385
cathodal inhibitory effects on cognitive function are rare (Jacobson, Koslowsky, et al., 2012). Another
386
possible reason for the absence of any significant cathodal AG effect in Experiment 2 could be the
387
placement of the anode electrode on the supraorbital area. This location is standard for tDCS studies
388
of action (Nitsche et al., 2008). However, it causes a strong current density close to the frontopolar
389
and prefrontal areas, where the anode is located. These areas may also contribute to intentional action
390
(Brass & Haggard, 2007). Thus, our montage for cathodal stimulation of AG in experiment 2 involved
391
anodal stimulation at a frontopolar site, which may not be strictly neutral for sense of agency. Future
392
studies could address this issue by using extracephalic cathode placement.
393
4.4. Hemispheric specialisation of the sense of agency
394
Experiment 3 avoided the potential confound of frontopolar stimulation using a biparietal montage.
395
This produces a higher current density in a small region surrounding the electrodes (Cohen Kadosh et
396
al., 2010; Nathan, Sinha, Gordon, Lesser, & Thakor, 1993), compared to the conventional supra-
397
orbital location. This might result in a more focal stimulation. More importantly, the biparietal
398
montage excludes the possibility that the significant effects of anodal AG stimulation in experiments
399
1 and 2 were in fact caused by cathodal frontopolar stimulation. Specifically, if the effects in
400
experiments 1 and 2 were merely due to cathodal frontopolar stimulation, then no effect of stimulation
401
should be found in experiment 3. The biparietal montage also allowed us to investigate lateralisation
402
of agency by varying both tDCS polarity and the hand used for action. Similar approaches have been
403
used previously in other studies (Bardi, Kanai, Mapelli, & Walsh, 2013; Jacobson, Goren, Lavidor, &
404
Levy, 2012).
405
The results of the third experiment replicated our previous findings. Anodal stimulation of putative
406
left AG significantly decreased tone binding compared to both sham and cathodal stimulation of the
AC C
EP
TE D
M AN U
SC
RI PT
382
15
ACCEPTED MANUSCRIPT same area. The tDCS effect was statistically equivalent whether the action was made with the left or
408
the right hand. No effects were observed with anodal stimulation of the putative right AG.
409
Experiment 3 does not support the alternative interpretation of experiments 1 and 2 based on a
410
putative cathodal frontopolar stimulation. In contrast, experiment 3 supports the interpretation of an
411
anodal left AG effect. We cannot conclusively rule out some contribution of frontopolar stimulation to
412
our results, but we can rule in a specific contribution of the left AG.
413
Experiment 3 adds several important elements to the previous studies. First, it demonstrates an
414
involvement of AG in a task involving randomised, stimulus-driven selection between alternative
415
actions, as opposed to mere repetition of a simple action. Second, it suggests that left, but not right
416
AG is responsible for action-outcome binding for actions made by either hand. We found no
417
interaction between stimulation and hand used for action. Previous neuroimaging studies have
418
reported activation corresponding to non-agency judgements in both left and right AG. Interestingly,
419
right AG activations appeared to dominate (Farrer et al., 2003; Farrer & Frith, 2002; Farrer et al.,
420
2008), in contrast to our finding. However, in a more recent fMRI study, Lee & Reeve, (2013)
421
reported higher activity in the left AG during non-self-determined behaviour, consistent with our
422
hypothesis in Experiment 1 that anodal AG stimulation activates a neural code for ‘non-agency’.
423
Finally, hemispheric specialisation of agency could plausibly depend on the task used, and the type of
424
agency judgement. Previous neuroimaging studies generally used explicit judgements of agency, and
425
often used complex manual actions with visual feedback (Farrer & Frith, 2002; Farrer et al., 2008;
426
Sperduti et al., 2011). We are not aware of any neuroimaging study investigating the hemispheric
427
lateralisation of low-level implicit measures of agency.
428
4.5. Limitations
429
The results of experiment 3 by themselves could not distinguish between an effect of anodal
430
stimulation of the putative left AG from an effect of cathodal stimulation of the putative right AG.
431
However, this result does allow us to exclude a model in which tDCS simply acts to increase neural
432
noise, irrespective of polarity. Moreover, our experiment 1 found some evidence of a left-hemisphere
AC C
EP
TE D
M AN U
SC
RI PT
407
16
ACCEPTED MANUSCRIPT anodal tDCS effect, while our experiment 2 found no evidence of any cathodal effect (though in the
434
left hemisphere, rather than the right). Cathodal stimulation effects in cognitive tasks are reported to
435
be weak (Jacobson, Koslowsky, et al., 2012). Therefore, we provisionally favour an interpretation of
436
experiment 3 based on a left parietal anodal effect, rather than a right-hemisphere cathodal effect.
437
Further research would be required to draw a definitive conclusion.
438
Our study is further limited because we did not control for cases of crossmodal binding in the absence
439
of active movement. Therefore, we cannot exclude the possibility that AG stimulation influenced
440
some general feature of time perception, as opposed to temporal processing specific to agency.
441
However, several studies have shown stronger binding between voluntary actions and outcomes than
442
between other, similarly paired, events, including involuntary movements and outcomes (Engbert,
443
Wohlschläger, Thomas, & Haggard, 2007; Haggard et al., 2002) or pairs of sensory stimuli (Haggard,
444
Aschersleben, Gehrke, & Prinz, 2002; Haggard, Martin, Taylor-Clarke, Jeannerod, & Franck, 2003).
445
Moreover, other studies have investigated effects of parietal tDCS on time perception in general, in
446
the absence of action: and these studies found no effect (Woods et al., 2014). Thus, the weight of
447
other studies suggests that the intentional binding phenomenon reflects a distortion of perceptual
448
timing that is, at least partly, specific to voluntary action.
SC
M AN U
TE D
449
RI PT
433
4.6. Dissociation between action binding and tone binding
451
Anodal stimulation over putative left AG was a common condition in all 3 experiments. Accordingly,
452
we could perform a pooled analysis of intentional binding results to compare this to the sham
453
stimulations that were also included in each experiment. This analysis showed a highly significant
454
reduction in tone binding with anodal stimulation of the putative left AG. We found no overall effect
455
on action binding. Dissociations between action binding and tone binding have been reported
456
previously (Wolpe et al., 2013), so it is possible that left parietal cortex is concerned primarily with
457
tone binding, rather than with action binding. This conclusion would be consistent with previous
458
studies suggesting that the AG processes mismatches in action outcomes (Farrer et al., 2008). On the
459
other hand, recent studies of explicit agency judgement suggest that AG also processes prospective,
AC C
EP
450
17
ACCEPTED MANUSCRIPT premotor information arising during action selection (Chambon, Moore, & Haggard, 2014; Chambon
461
et al., 2013).
462
Both online prospective and retrospective processes contribute to the intentional binding phenomenon
463
(Moore and Haggard, 2008). The experimental design used here cannot identify the independent
464
contribution of each process. However, binding of action towards outcome may rely more on
465
prospective processes during action selection, while perceptual shift of outcome toward action may
466
depend on retrospective, more inferential processes, triggered by reafferent signals about action
467
outcome (Chambon, Moore, & Haggard, 2014). Future studies may address this issue by designing
468
new paradigms which dissociate prospective and retrospective components of agency and examine the
469
role of dLPFC and AG in each of these components.
470
4.7. Conclusion and clinical implications
471
Sense of agency is an important and distinctive feature of human voluntary action. We used a causal
472
intervention (tDCS) and an implicit perceptual measure of sense of agency (intentional binding) to
473
examine the role of different brain areas in sense of agency. Anodal stimulation of parietal cortex
474
consistently reduced the binding of tones towards actions. We hypothesised that the angular gyrus
475
might contribute to the sense of agency by monitoring the linkage of actions to outcomes, or,
476
alternatively and equivalently, failures of such linkage. Anodal stimulation of this area may
477
correspond to artificial boosting of a mismatch detection process.
478
Sense of agency is altered following several classes of psychiatric and neurological disorders. In
479
particular, patients with apraxia following lesions to the left parietal fail to recognise the source of a
480
viewed manual gesture (Sirigu, Daprati, Pradat-Diehl, Franck, & Jeannerod, 1999). This deficit is
481
formally equivalent to an overestimation of agency in an explicit judgement task, consistent with
482
damage to a neural centre detecting mismatches. The posterior form of ‘alien hand syndrome’ is also
483
associated with contralateral parietal lesions (e.g., (Kloesel, Czarnecki, Muir, & Keller, 2010)).
484
Interestingly, these patients show involuntary and spontaneous movements of the contralateral limb,
485
but may correctly perceive that they are not agents over these actions. The capacity for voluntary
AC C
EP
TE D
M AN U
SC
RI PT
460
18
ACCEPTED MANUSCRIPT 486
movement is often preserved. Quantitative assessment of the implicit sense of agency in parietal
487
patients would be of considerable value in understanding the neural basis of sense of agency.
488
Competing interests
490
The authors declare no competing financial interests.
491
Acknowledgements
492
This work was supported by the European Research Council Advanced Grant HUMVOL. PH was
493
additionally supported by a Professorial Research Fellowship from the ESRC. The funding sources
494
had no involvement, in study design; in the collection, analysis or interpretation of data; in the writing
495
of the report; or in the decision to submit the article for publication.
496
We thank Andrea Desantis for helpful comments during the preparation of the manuscript.
SC
M AN U
497
502
503
EP
501
AC C
500
TE D
498
499
RI PT
489
504
505
506
507 19
ACCEPTED MANUSCRIPT 508
References
509
Bardi, L., Kanai, R., Mapelli, D., & Walsh, V. (2013). Direct current stimulation (tDCS) reveals
510
parietal asymmetry in local/global and salience-based selection. Cortex, 49(3), 850–860.
511
http://doi.org/10.1016/j.cortex.2012.04.016 Brass, M., & Haggard, P. (2007). To do or not to do: the neural signature of self-control. The Journal
RI PT
512 513
of Neuroscience: The Official Journal of the Society for Neuroscience, 27(34), 9141–9145.
514
http://doi.org/10.1523/JNEUROSCI.0924-07.2007
517 518 519
SC
516
Buxbaum, L. J., Sirigu, A., Schwartz, M. F., & Klatzky, R. (2003). Cognitive representations of hand posture in ideomotor apraxia. Neuropsychologia, 41(8), 1091–1113.
Cardinal, R. N., & Aitken, M. R. F. (2013). ANOVA for the Behavioral Sciences Researcher.
M AN U
515
Psychology Press.
Chambon, V., Moore, J. W., & Haggard, P. (2014). TMS stimulation over the inferior parietal cortex
520
disrupts prospective sense of agency. Brain Structure & Function.
521
http://doi.org/10.1007/s00429-014-0878-6
Chambon, V., Wenke, D., Fleming, S. M., Prinz, W., & Haggard, P. (2013). An Online Neural
TE D
522 523
Substrate for Sense of Agency. Cerebral Cortex, 23(5), 1031–1037.
524
http://doi.org/10.1093/cercor/bhs059
Cohen Kadosh, R., Soskic, S., Iuculano, T., Kanai, R., & Walsh, V. (2010). Modulating neuronal
EP
525
activity produces specific and long-lasting changes in numerical competence. Current
527
Biology: CB, 20(22), 2016–2020. http://doi.org/10.1016/j.cub.2010.10.007
528 529 530
AC C
526
Cravo, A. M., Claessens, P. M. E., & Baldo, M. V. C. (2009). Voluntary action and causality in temporal binding. Experimental Brain Research, 199(1), 95–99. http://doi.org/10.1007/s00221-009-1969-0
531
Cumming, G. (2005). Understanding the Average Probability of Replication Comment on Killeen
532
(2005). Psychological Science, 16(12), 1002–1004. http://doi.org/10.1111/j.1467-
533
9280.2005.01650.x
20
ACCEPTED MANUSCRIPT 534
De Jong, B. M. (2011). Neurology of widely embedded free will. Cortex; a Journal Devoted to the
535
Study of the Nervous System and Behavior, 47(10), 1160–1165.
536
http://doi.org/10.1016/j.cortex.2011.06.011
537
Desmurget, M., Reilly, K. T., Richard, N., Szathmari, A., Mottolese, C., & Sirigu, A. (2009). Movement intention after parietal cortex stimulation in humans. Science (New York, N.Y.),
539
324(5928), 811–813. http://doi.org/10.1126/science.1169896
540
RI PT
538
Engbert, K., Wohlschläger, A., Thomas, R., & Haggard, P. (2007). Agency, subjective time, and other minds. Journal of Experimental Psychology. Human Perception and Performance, 33(6),
542
1261–1268. http://doi.org/10.1037/0096-1523.33.6.1261
544 545
Farrer, C., Franck, N., Georgieff, N., Frith, C. D., Decety, J., & Jeannerod, M. (2003). Modulating the experience of agency: a positron emission tomography study. NeuroImage, 18(2), 324–333.
M AN U
543
SC
541
Farrer, C., Frey, S. H., Horn, J. D. V., Tunik, E., Turk, D., Inati, S., & Grafton, S. T. (2008). The
546
Angular Gyrus Computes Action Awareness Representations. Cerebral Cortex, 18(2), 254–
547
261. http://doi.org/10.1093/cercor/bhm050
Farrer, C., & Frith, C. D. (2002). Experiencing oneself vs another person as being the cause of an
TE D
548 549
action: the neural correlates of the experience of agency. NeuroImage, 15(3), 596–603.
550
http://doi.org/10.1006/nimg.2001.1009
Fink, G. R., Marshall, J. C., Halligan, P. W., Frith, C. D., Driver, J., Frackowiak, R. S., & Dolan, R. J.
EP
551
(1999). The neural consequences of conflict between intention and the senses. Brain: A
553
Journal of Neurology, 122 ( Pt 3), 497–512.
AC C
552
554
Fletcher, P. C., & Frith, C. D. (2009). Perceiving is believing: a Bayesian approach to explaining the
555
positive symptoms of schizophrenia. Nature Reviews. Neuroscience, 10(1), 48–58.
556 557 558 559
http://doi.org/10.1038/nrn2536
Frith, C. D. (2014). Action, agency and responsibility. Neuropsychologia, 55, 137–142. http://doi.org/10.1016/j.neuropsychologia.2013.09.007 Frith, C. D., Blakemore, & Wolpert, D. M. (2000). Abnormalities in the awareness and control of
560
action. Philosophical Transactions of the Royal Society of London. Series B: Biological
561
Sciences, 355(1404), 1771–1788. http://doi.org/10.1098/rstb.2000.0734 21
ACCEPTED MANUSCRIPT 562
Haggard, P., Aschersleben, G., Gehrke, J., & Prinz, W. (2002). Action, binding, and awareness. In W.
563
Prinz & B. Hommel (Eds.), Common Mechanisms in Perception and Action B. (Vol. 19, pp.
564
266–285). OXFORD UNIVERSITY PRESS. Retrieved from http://eprints.ucl.ac.uk/4192/
566 567
Haggard, P., Clark, S., & Kalogeras, J. (2002). Voluntary action and conscious awareness. Nature Neuroscience, 5(4), 382–385. http://doi.org/10.1038/nn827 Haggard, P., Martin, F., Taylor-Clarke, M., Jeannerod, M., & Franck, N. (2003). Awareness of action
568
in schizophrenia. Neuroreport, 14(7), 1081–1085.
569
http://doi.org/10.1097/01.wnr.0000073684.00308.c0
Hecht, D., Walsh, V., & Lavidor, M. (2010). Transcranial Direct Current Stimulation Facilitates
SC
570
RI PT
565
Decision Making in a Probabilistic Guessing Task. The Journal of Neuroscience, 30(12),
572
4241–4245. http://doi.org/10.1523/JNEUROSCI.2924-09.2010
M AN U
571
Horvath, J. C., Forte, J. D., & Carter, O. (2014). Evidence that transcranial direct current stimulation
574
(tDCS) generates little-to-no reliable neurophysiologic effect beyond MEP amplitude
575
modulation in healthy human subjects: A systematic review. Neuropsychologia, 66C, 213–
576
236. http://doi.org/10.1016/j.neuropsychologia.2014.11.021
577
TE D
573
Jacobson, L., Goren, N., Lavidor, M., & Levy, D. A. (2012). Oppositional transcranial direct current stimulation (tDCS) of parietal substrates of attention during encoding modulates episodic
579
memory. Brain Research, 1439, 66–72. http://doi.org/10.1016/j.brainres.2011.12.036
581 582 583 584
Jacobson, L., Koslowsky, M., & Lavidor, M. (2012). tDCS polarity effects in motor and cognitive domains: a meta-analytical review. Experimental Brain Research, 216(1), 1–10.
AC C
580
EP
578
http://doi.org/10.1007/s00221-011-2891-9
Jahanshahi, M., Jenkins, I. H., Brown, R. G., Marsden, C. D., Passingham, R. E., & Brooks, D. J. (1995). Self-initiated versus externally triggered movements. I. An investigation using
585
measurement of regional cerebral blood flow with PET and movement-related potentials in
586
normal and Parkinson’s disease subjects. Brain: A Journal of Neurology, 118 ( Pt 4), 913–
587
933.
22
ACCEPTED MANUSCRIPT 588
Kloesel, B., Czarnecki, K., Muir, J. J., & Keller, A. S. (2010). Sequelae of a left-sided parietal stroke:
589
posterior alien hand syndrome. Neurocase, 16(6), 488–493.
590
http://doi.org/10.1080/13554794.2010.497154
592 593 594 595
Kranick, S. M., & Hallett, M. (2013). Neurology of volition. Experimental Brain Research, 229(3), 313–327. http://doi.org/10.1007/s00221-013-3399-2
RI PT
591
Krzanowski, W. J. (2000). Principles of Multivariate Analysis: A User’s Perspective (2 edition). Oxford Oxfordshire ; New York: OUP Oxford.
Kühn, S., Brass, M., & Haggard, P. (2013). Feeling in control: Neural correlates of experience of agency. Cortex, 49(7), 1935–1942. http://doi.org/10.1016/j.cortex.2012.09.002
597
Lee, W., & Reeve, J. (2013). Self-determined, but not non-self-determined, motivation predicts
SC
596
activations in the anterior insular cortex: an fMRI study of personal agency. Social Cognitive
599
and Affective Neuroscience, 8(5), 538–545. http://doi.org/10.1093/scan/nss029
M AN U
598
600
Meier, U. (2006). A note on the power of Fisher’s least significant difference procedure.
601
Pharmaceutical Statistics, 5(4), 253–263. http://doi.org/10.1002/pst.210 Moore, J. W., Middleton, D., Haggard, P., & Fletcher, P. C. (2012). Exploring implicit and explicit
TE D
602 603
aspects of sense of agency. Consciousness and Cognition, 21(4), 1748–1753.
604
http://doi.org/10.1016/j.concog.2012.10.005
607 608 609 610 611
EP
606
Moore, J. W., & Obhi, S. S. (2012). Intentional binding and the sense of agency: a review. Consciousness and Cognition, 21(1), 546–561. http://doi.org/10.1016/j.concog.2011.12.002 Moore, J. W., Ruge, D., Wenke, D., Rothwell, J., & Haggard, P. (2010). Disrupting the experience of
AC C
605
control in the human brain: pre-supplementary motor area contributes to the sense of agency. Proceedings. Biological Sciences / The Royal Society, 277(1693), 2503–2509. http://doi.org/10.1098/rspb.2010.0404
Moretto, G., Walsh, E., & Haggard, P. (2011). Experience of agency and sense of responsibility.
612
Consciousness and Cognition, 20(4), 1847–1854.
613
http://doi.org/10.1016/j.concog.2011.08.014
614 615
Nathan, S. S., Sinha, S. R., Gordon, B., Lesser, R. P., & Thakor, N. V. (1993). Determination of current density distributions generated by electrical stimulation of the human cerebral cortex. 23
ACCEPTED MANUSCRIPT 616
Electroencephalography and Clinical Neurophysiology, 86(3), 183–192.
617
http://doi.org/10.1016/0013-4694(93)90006-H Nitsche, M. A., Cohen, L. G., Wassermann, E. M., Priori, A., Lang, N., Antal, A., Pascual-Leone, A.
619
(2008). Transcranial direct current stimulation: State of the art 2008. Brain Stimulation, 1(3),
620
206–223. http://doi.org/10.1016/j.brs.2008.06.004
621
RI PT
618
Nitsche, M. A., Liebetanz, D., Lang, N., Antal, A., Tergau, F., & Paulus, W. (2003). Safety criteria for transcranial direct current stimulation (tDCS) in humans. Clinical Neurophysiology:
623
Official Journal of the International Federation of Clinical Neurophysiology, 114(11), 2220–
624
2222; author reply 2222–2223.
626 627
Nitsche, M. A., & Paulus, W. (2000). Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. The Journal of Physiology, 527 Pt 3, 633–639.
M AN U
625
SC
622
Okamoto, M., Dan, H., Sakamoto, K., Takeo, K., Shimizu, K., Kohno, S., Dan, I. (2004). Threedimensional probabilistic anatomical cranio-cerebral correlation via the international 10–20
629
system oriented for transcranial functional brain mapping. NeuroImage, 21(1), 99–111.
630
http://doi.org/10.1016/j.neuroimage.2003.08.026
631
TE D
628
Poreisz, C., Boros, K., Antal, A., & Paulus, W. (2007). Safety aspects of transcranial direct current stimulation concerning healthy subjects and patients. Brain Research Bulletin, 72(4-6), 208–
633
214. http://doi.org/10.1016/j.brainresbull.2007.01.004
EP
632
Rowe, J. B., Hughes, L., & Nimmo-Smith, I. (2010). Action selection: a race model for selected and
635
non-selected actions distinguishes the contribution of premotor and prefrontal areas.
636 637 638 639
AC C
634
NeuroImage, 51(2), 888–896. http://doi.org/10.1016/j.neuroimage.2010.02.045
Sirigu, A., Daprati, E., Ciancia, S., Giraux, P., Nighoghossian, N., Posada, A., & Haggard, P. (2004). Altered awareness of voluntary action after damage to the parietal cortex. Nature Neuroscience, 7(1), 80–84. http://doi.org/10.1038/nn1160
640
Sirigu, A., Daprati, E., Pradat-Diehl, P., Franck, N., & Jeannerod, M. (1999). Perception of self-
641
generated movement following left parietal lesion. Brain, 122(10), 1867–1874.
642
http://doi.org/10.1093/brain/122.10.1867
24
ACCEPTED MANUSCRIPT 643
Sperduti, M., Delaveau, P., Fossati, P., & Nadel, J. (2011). Different brain structures related to self-
644
and external-agency attribution: a brief review and meta-analysis. Brain Structure &
645
Function, 216(2), 151–157. http://doi.org/10.1007/s00429-010-0298-1
646
Spitoni, G. F., Pireddu, G., Cimmino, R. L., Galati, G., Priori, A., Lavidor, M., Pizzamiglio, L. (2013). Right but not left angular gyrus modulates the metric component of the mental body
648
representation: a tDCS study. Experimental Brain Research. Experimentelle Hirnforschung.
649
Expérimentation Cérébrale, 228(1), 63–72. http://doi.org/10.1007/s00221-013-3538-9
650
RI PT
647
Synofzik, M., Vosgerau, G., & Newen, A. (2008). Beyond the comparator model: a multifactorial two-step account of agency. Consciousness and Cognition, 17(1), 219–239.
652
http://doi.org/10.1016/j.concog.2007.03.010
Wolpe, N., Haggard, P., Siebner, H. R., & Rowe, J. B. (2013). Cue integration and the perception of
M AN U
653
SC
651
654
action in intentional binding. Experimental Brain Research, 229(3), 467–474.
655
http://doi.org/10.1007/s00221-013-3419-2
657 658
Wolpert, D. M., & Ghahramani, Z. (2000). Computational principles of movement neuroscience. Nature Neuroscience, 3 Suppl, 1212–1217. http://doi.org/10.1038/81497
TE D
656
Woods, A. J., Hamilton, R. H., Kranjec, A., Minhaus, P., Bikson, M., Yu, J., & Chatterjee, A. (2014). Space, time, and causality in the human brain. NeuroImage, 92, 285–297.
660
http://doi.org/10.1016/j.neuroimage.2014.02.015
662 663 664
665
Zwissler, B., Sperber, C., Aigeldinger, S., Schindler, S., Kissler, J., & Plewnia, C. (2014). Shaping Memory Accuracy by Left Prefrontal Transcranial Direct Current Stimulation. The Journal of
AC C
661
EP
659
Neuroscience, 34(11), 4022–4026. http://doi.org/10.1523/JNEUROSCI.5407-13.2014
666
667
668 25
ACCEPTED MANUSCRIPT Captions
670
Fig. 1. Schematic of intentional binding. Action and tone shifts are measured by subtracting each
671
participant’s mean judgement error in baseline conditions from judgement error in operant conditions.
672
These shifts serve as measures of intentional binding. Vertical bars and thin arrows represent mean
673
judgement errors in each condition. Thick arrows represent binding effects. See text for full
674
explanation.
675
Fig. 2. A. tDCS montage and study design (anode +, cathode -). See text for explanation. In
676
Experiment 1, to control for the cutaneous sensation of all three locations, all sponges were kept in
677
place across the three sessions. However, only two of them were actually functioning in each session.
678
B. Stimulation protocol. The order of conditions was randomised within each session.
679
Fig. 3. Intentional binding in Experiment 1. A) The dashed line indicates the perceived time of either
680
action or tone in the corresponding baseline condition. A separate baseline condition was used for
681
each session, and differences in baseline values across sessions have been removed for display
682
purposes. Binding effects are drawn to scale, and values are in milliseconds. B) Mean binding effects
683
in ms. The sign of tone binding effects has been inverted to allow for comparison with action binding.
684
Error bars show standard error of the mean. * p<0.05.
685
Fig. 4. Intentional binding in Experiment 2. Format as in Fig. 3.
686
Fig. 5. Intentional binding in Experiment 3. Format as in Fig. 3.
AC C
EP
TE D
M AN U
SC
RI PT
669
26
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT Baseline Action
Baseline Tone
Operant Action
Operant Tone
1
-9
1
-8
-59
3
-68
73
-24
14
4
-14
177
-25
60
A.1.
5
-194
-39
-78
134
Mea
6
52
130
86
29
n (in
7
-54
24
-24
8
22
69
75
9
-84
14
11
-56
-61
12
89
13
-92
14
-80
15
-18
16
-8
17
16
18
RI PT
Participant
33
Tabl e
ms)
SC
30
of
-48
-83
-161
all
M AN U
-86
152
90
con
-13
-48
-192
ditio
-6
-52
-150
-31
-21
-94
22
31
-88
-86
-48
-166
-95
-48
-41
-90
dal
2
27
24
144
22
stim
10
-74
-36
-6
-241
ulati
EP
TE D
18
AC C
on of left dLPFC from subjects in Experiment 1. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ns for ano
ACCEPTED MANUSCRIPT 1
113
88
140
91
3
60
98
73
62
4
207
151
97
183
5
133
95
191
164
6
121
128
89
100
7
91
124
70
8
70
78
62
9
37
86
47
11
73
89
12
74
83
13
96
91
14
33
15
39
16
40
17
70
18
84
2
134
10
105
SC
RI PT
Tab le A.2 .
104
Sta
89
nda
91
rd
141
76
82
78
111
69
44
73
114
49
62
of
113
50
78
jud
153
64
113
ge
106
68
88
118
213
190
296
240
93
dev iati on
me nt err
EP
TE D
M AN U
116
or across trials (in ms) of all conditions for anodal stimulation of left dLPFC from subjects in
AC C
Experiment 1. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ACCEPTED MANUSCRIPT 26
10
91
-55
3
-52
27
-70
-30
4
-6
126
15
-14
5
-21
-21
-37
83
6
-41
95
49
48
7
-114
-1
-108
-26
8
19
83
77
112
9
-51
33
-93
-41
11
-93
-45
-71
-154
12
117
46
199
39
13
-83
-27
-81
-236
14
-77
-14
-80
-168
15
-73
-95
-40
-41
16
20
64
35
-58
17
-1
-112
-63
-108
2
SC
M AN U
-105
-44
-103
-110
176
42
143
33
-87
-74
-20
-117
EP
10
TE D
18
RI PT
1
AC C
Table A.3. Mean (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Baseline Action
Baseline Tone
Operant Action
Operant Tone
1
86
79
112
93
3
59
92
74
72
4
65
146
72
120
5
139
99
154
88
6
86
161
76
7
105
120
87
8
60
83
80
9
46
72
44
11
107
64
12
60
96
13
114
14
48
15
60
16
41
17
93
18
82
2
284
10
55
SC
RI PT
Participant
125 86
70 83
120
72
87
112
100
104
47
54
56
69
37
42
75
72
72
130
57
211
136
76
136
117
181
106
91
49
118
EP
TE D
M AN U
129
AC C
Table A.4. Standard deviation of judgement error across trials (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
Tab -25
-14
33
-175
3
-59
42
230
13
4
9
208
-2
-147
5
1
0
-42
6
59
45
97
7
-21
-34
-68
8
17
82
9
-113
37
11
-86
-39
12
11
13
-172
14
-71
15
-35
16
-11
17
RI PT
1
88
le A.5 . Me
-161
an
115
48
(in
-54
-112
-40
-190
16
183
13
-62
-58
-227
25
-37
-71
con
-32
-44
-46
diti
72
32
-96
ons
-58
-82
-72
-83
for
18
-85
7
-74
-122
2
28
39
113
13
-57
6
-29
-127
M AN U
TE D
EP
10
SC
-44
of
AC C Baseline Action
all
sha m
from subjects in Experiment 1. Shaded rows are data from excluded subjects.
Participant
ms)
Baseline Tone
Operant Action
Operant Tone
91
121
77
98
3
68
64
70
53
4
78
161
80
195
5
137
115
99
83
6
137
115
94
101
7
90
113
78
8
80
47
102
9
69
132
78
11
82
92
12
80
111
13
101
14
47
15
48
16
41
17
70
18
89
2
155
10
62
SC
RI PT
1
M AN U
ACCEPTED MANUSCRIPT
95
70
92
77
172
62
120
117
122
47
51
73
77
42
38
80
45
69
105
68
83
131
83
116
88
271
136
85
65
77
AC C
EP
TE D
73
Table A.6. Standard deviation of judgement error across trials (in ms) of all conditions for sham from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
-61
-61
-87
-99
-86
-75
-112
-341
111
79
84
97
-72
-116
-107
-173
-38
44
25
4
-167
-70
-130
-169
-40
-51
-29
-146
-82
-59
-16
-17
-9
27
84
30
1 3
5 6 7
9
11 44
57
45
-24
20
-74
-35
-115
-14
-49
-64
-54
14
59
78
95
3
-77
-28
-75
-167
101
161
65
101
83
371
146
-158
-58
-10
-244
12 -68 14 -83 15 -53
2 13
EP
18
TE D
16 17
M AN U
10
SC
8
RI PT
4
AC C
-151
Table A.7. Mean (in ms) of all conditions for anodal stimulation of the left AG from subjects in Experiment 2. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action 1
Baseline Tone mn2
Operant Action n3
Operant Tone mn4
52
57
51
63
74
93
61
68
93
88
68
92
69
86
61
116
67
76
69
74
86
95
74
112
63
87
58
99
1 3
6 7
AC C
EP
TE D
M AN U
8
SC
5
RI PT
4
ACCEPTED MANUSCRIPT 9
4 5 6
83 Operant Action 94
-87
-24
-42
11 12 14 15 16
7 17 8 18 9 2 10 13
198 -111 83 77 53 -54 47 -12 62 -147 40 -55 55 -54 211 -7 123
-62
71 77
37
57 -135 96 95 51 -127 54 -56 50 -11 166 44 83
65 -121 81 49 92 -89 66 -28 77 -172 48 -58 60 -15 209 20 121
76 Operant Tone 99 -212 90 -362 62 62 78 -186 62 -20 63 -142 111 -143 71 -2 214 -81 140
RI PT
3
69 Baseline Tone 57
SC
1
68 Baseline Action 65
M AN U
Participant 10
Table A.8. Standard deviation of judgement error across trials (in ms) of all conditions for
AC C
EP
excluded subjects.
TE D
anodal stimulation of the left AG from subjects in Experiment 2. Shaded rows are data from
ACCEPTED MANUSCRIPT 11 53
58
92
-29
-30
-17
-60
63
-106
-105
-65
0
-69
33
-56
-20
83
66
88
37
-105
-57
-101
84
137
97
-177
-16
-168
-93
-104
-119
12 14 15
17 18 2
-223 73
-111
SC
13
RI PT
16
-229
M AN U
Table A.9. Mean (in ms) of all conditions for cathodal stimulation of the left AG from
Baseline Action 1
Baseline Tone mn2
Operant Action n3
Operant Tone mn4
55
83
59
78
57
99
45
86
87
120
59
87
66
76
65
80
56
60
58
64
89
109
154
150
56
58
67
73
159
130
84
87
AC C
Participant
EP
TE D
subjects in Experiment 2. Shaded rows are data from excluded subjects.
1 3 4 5 6 7 8 9
ACCEPTED MANUSCRIPT 10 74
64
76
84
63
93
54
73
68
73
66
64
66
65
83
66
65
63
77
57
103
94
60
53
55
35
59
60
44
92
141
155
407
193
11 12 14
16 17 18
84 46
193
270 147
M AN U
13
95
SC
2
RI PT
15
Table A.10. Standard deviation of judgement error across trials (in ms) of all conditions for cathodal stimulation of the left AG from subjects in Experiment 2. Shaded rows are data from
AC C
EP
TE D
excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
-108
-9
-83
-73
-150
-117
-136
-258
44
-21
25
-101
-60
-100
-77
-111
-85
87
-8
6
-94
-51
-116
-144
-29
-2
-38
-176
1 3 4 5 6 7 8
ACCEPTED MANUSCRIPT 9 -59
-1
-14
-34
-55
-20
-16
-175
43
36
-3
-55
-134
-21
-106
-21
-108
-113
-38
-36
-66
-56
-11
24
88
67
-88
0
-116
85
124
0
322
10 11 12
15 16 17
SC
18 2 -92
-117
-5 5
-98
49
70
98
180
M AN U
13
RI PT
14
-100
-252
Table A.11. Mean (in ms) of all conditions for sham from subjects in Experiment 2. Shaded
Baseline Action 1
Baseline Tone mn2
Operant Action n3
Operant Tone mn4
57
76
42
55
81
82
68
95
120
68
64
80
54
84
62
71
49
86
62
89
72
82
78
92
56
67
53
80
AC C
Participant
EP
TE D
rows are excluded subjects.
1 3 4 5 6 7 8
ACCEPTED MANUSCRIPT 9 70
131
81
80
66
239
91
79
69
99
62
96
67
54
58
66
64
60
76
72
62
92
71
78
70
103
46
94
46
45
48
96
46
61
246
183
10 11 12
16 17
41
56
123
259
M AN U
18
126
2 13 266
SC
15
RI PT
14
120
129
Table A.12. Standard deviation of judgement error across trials (in ms) of all conditions for
AC C
EP
TE D
sham from subjects in Experiment 2. Shaded rows are excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-9
-43
13
-2
-22
27
36
-66
-22
28
-96
-47
-200
-186
-50
6
61
-19
-11
-13
-14
-94
-63
-14
-30
6
48
62
-134
-106
-50
-146
-60
-37
-13
30
38
23
45
36
-71
-33
89
37
66
-22
-52
77
-50
-138
-89
-115
-130
-127
-25
-31
-86
-86
-60
-28
-37
-39
-75
-118
-85
2 3
7 8 9
12
-48
-10
-18
-72
-87
-52
-5
-9
-65
-65
-20
-39
-40
-2
-4
-26
-127
-143
-37
-29
-59
-2
-100
-85
-92
-108
-126
-61
-44
-22
-87
-79
-60
16
-32
-30
-3
-12
48
18
-30
-45
35
-19
-4
-50
87
-22
166
-27
-16
-19
72
51
11
-14
173
110
13 14 15
17 -108 1 -51 6
AC C
18
EP
-5
TE D
16
10
M AN U
11
SC
5
RI PT
4
Table A.13. Mean (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action Left 1
Baseline Action Right 1
Baseline Tone2
Operant Action Leftn3
Operant Action Rightn3
Operant Tone Leftmn4
Operant Tone Rightn4
66
83
94
60
73
46
47
86
126
111
74
119
95
110
106
122
87
77
61
82
62
66
81
85
99
86
146
79
118
78
83
109
85
125
128
84
70
123
77
124
144
89
102
80
84
67
73
178
136
78
87
106
88
71
79
115
78
51
49
75
50
83
62
74
45
94
60
61
84
61
57
50
87
60
48
70
53
137
78
102
64
88
108
91
56
101
57
51
54
77
85
2 3
7 8 9
12 13 14 15
52 1 33 6 65 10 111 18
54
63
64
90
77
42
76
114
185
88
83
243
108
76
70
131
115
116
132
114
298
130
126
232
213
219
64
234
115
AC C
110
51
EP
17
TE D
16
M AN U
11
SC
5
RI PT
4
Table A.14. Standard deviation of judgement error across trials (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-10
-33
20
-33
20
20
29
-39
-49
10
-24
-7
-151
-155
-30
-11
98
-17
-57
-48
-27
-42
-142
-93
-28
-76
85
70
63
38
-6
5
-1
-155
-139
-73
-69
12
9
3
9
-80
-79
2
4
-32
-33
-8
-62
-85
-79
-81
-22
-49
26
-17
-20
108
102
-85
-8
-91
-157
-185
-100
-129
-30
-18
-40
-107
-103
-27
-36
-83
-98
-88
-25
10
4
-10
-11
-29
-141
-141
-64
-99
5 7
9
13 14
16 -67 17 -95 1
10
-41
-27
-42
-38
-141
-154
-105
-39
-78
-67
-146
-131
-54
34
-3
-46
24
31
EP
-77 6
TE D
15
M AN U
11
SC
8
12
RI PT
3
56
65
-14
-39
5
26
10
-78
-58
44
67
13
31
2
0
74
12
67
-85
-210
AC C
18
40
Table A.15. Mean (in ms) of all conditions for anodal stimulation of right AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action Left 1
Baseline Action Right 1
Baseline Tone2
Operant Action Leftn3
Operant Action Rightn3
Operant Tone Leftmn4
Operant Tone Rightn4
84
70
62
33
75
73
60
62
100
90
83
86
87
63
69
56
84
58
63
93
64
36
66
94
44
97
61
58
87
113
58
109
85
103
105
111
101
139
91
151
99
98
73
105
59
67
106
115
75
57
79
95
79
59
85
102
43
63
44
44
47
42
63
60
48
2 3
5 7
9
12
46
44
70
65
55
84
39
67
63
57
77
73
46
52
64
99
81
47
72
60
58
43
55
42
53
47
42
59
63
35
137
91
38
92
275
58
52
55
269
81
113
64
181
205
96
67
78
230
258
117
178
106
66
109
97
235
135
274
13 14 15
17 45 1 39 6
AC C
18
EP
116
TE D
16
10
M AN U
11
SC
8
RI PT
4
Table A.16. Standard deviation of judgement error across trials (in ms) of all conditions for anodal stimulation of right AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-2
-8
15
-21
-26
23
37
-80
-36
-26
-92
-24
2
AC C
EP
TE D
M AN U
SC
RI PT
3 -274
-283
ACCEPTED MANUSCRIPT 4 -5
-46
103
-44
-43
-24
15
-102
-76
-24
-72
-64
-100
-99
-97
-77
-13
-56
42
-13
-20
-10
20
40
-79
-82
-45
-37
-39
-2
20
40
31
26
20
-64
-110
-56
-103
-92
-20
-50
-52
-43
-29
-28
-29
-76
-86
-59
-40
-3
62
48
-48
-60
-7
7
-9
-20
-17
1
12
-72
-92
-36
-109
-96
-60
-60
-92
-85
-18
-70
-49
-154
-170
-72
-74
-56
-48
-67
-136
-87
-55
-50
21
-70
-6
29
10
19
30
22
27
-2
-86
-47
5 7 8
RI PT
9 11 12
14
16 17 1 6
98 18 -37
TE D
10
M AN U
15
SC
13
103
40
-15
18
58
38
57
-5
68
100
48
29
EP
Table A.17. Mean (in ms) of all conditions for sham from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded
AC C
subjects.
Participant
Baseline Action Left 1
Baseline Action Right 1
Baseline Tone
Operant Action Leftn3
Operant Action Rightn3
Operant Tone Leftmn4
Operant Tone Rightn2
54
76
80
92
63
74
46
115
124
152
237
113
86
63
2 3
ACCEPTED MANUSCRIPT 4 86
92
95
67
67
79
96
40
41
87
53
48
53
106
115
115
92
57
150
89
85
82
92
122
114
88
85
81
62
96
98
93
89
74
184
49
81
92
83
83
63
122
74
64
36
58
76
53
61
68
66
76
64
68
47
56
77
43
87
71
72
75
49
61
46
70
79
57
66
64
58
54
48
62
32
72
72
171
270
5 7 8
12 13 14
16
83
61
63
113
72
183
27
51
87
68
56
205
51
54
74
173
54
69
126
119
209
98
107
83
176
181
98
156
133
90
223
166
17 1 6
80
TE D
10 18
M AN U
15
SC
11
RI PT
9
EP
Table A.18. Standard deviation of judgement error across trials (in ms) of all conditions for sham from subjects in Experiment 3. Values are shown separately for left and right hand
AC C
actions. Shaded rows are data from excluded subjects.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
TE D
Fig. A.1. Scores of the individual participants on the first canonical variate from a discriminant analysis of Experiment 1. Each dot indicates one participant, and a fitted Gaussian frequency distribution is also shown. The raw coefficients of the canonical variate
EP
were 0.015 for action binding, and 0.02 for tone binding. Both parietal and frontal stimulation
AC C
conditions were significantly different from sham.
ACCEPTED MANUSCRIPT Baseline Action
Baseline Tone
Operant Action
Operant Tone
1
-25
-5
-15
-42
3
-63
51
-38
21
4
14
153
-43
61
B.1.
5
-162
-34
-84
152
Med
6
45
143
104
36
ian
7
-48
47
-20
8
27
71
76
9
-83
23
11
-43
-54
12
89
13
-87
14
-84
15
-14
16
-6
17
7
18
SC
RI PT
Participant
29
Tabl e
(in
22
ms)
-52
-105
-147
of
M AN U
-86
149
87
all
17
-67
-202
con
5
-62
-134
-12
-21
-86
19
13
-86
-96
-45
-194
-96
-33
-53
-99
ano
2
22
21
130
-16
dal
10
-75
-87
-59
-233
stim
EP
TE D
32
ditio ns for
AC C
ulation of left dLPFC from subjects in Experiment 1. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ACCEPTED MANUSCRIPT 1 Participant
31 Baseline Action
0 Baseline Tone
101 Operant Action
-50 Operant Tone
-59
6
-66
-19
4
-10
92
16
-7
5
-10
-4
-52
88
6
-48
62
50
44
7
-126
-13
-116
-21
8
16
86
95
106
9
-53
32
-95
-28
11
-89
-35
-99
-155
12
112
24
209
48
13
-91
-38
-82
-244
14
-73
-13
-74
-163
15
-60
-98
-35
-32
16
12
76
21
-66
17
0
-99
-60
-135
2
SC
M AN U
-100
-23
-109
-89
149
48
106
28
-101
-65
-27
-120
EP
10
TE D
18
RI PT
3
AC C
Table B.2. Median (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT 1 Participant 3
-26 Baseline Action -59
-31 Baseline Tone 36
28 Operant Action 234
-168 Operant Tone 10
4
15
193
-7
-174
5
-21
-31
-46
74
6
43
42
101
-41
7
-22
-24
-56
8
12
82
118
9
-104
11
-68
11
-83
-57
-52
12
9
2
13
-190
-57
14
-77
15
-31
16
-15
17
-55
18
SC
RI PT
Tabl
-149
e B.3. Med
31
ian
-124
(in
-212
ms) of
-11
-49
-207
18
-46
-79
-15
-36
-52
ns
65
30
-90
for
-77
-64
-69
sha
-100
12
-77
-113
m
2
18
35
51
21
10
-53
17
-32
-119
TE D
M AN U
178
all con ditio
fro m subj
EP
ects in Experiment 1. Shaded rows are data from excluded subjects.
AC C
Action binding: No significant effect of stimulation (F(2, 30)=3.11, p=0.06). Tone binding: Significant effect of stimulation (F(2, 30)=3.41, p=0.04). Tone binding was significantly reduced by anodal stimulation of the putative left AG compared to sham (t(15)=2.40, p=0.03).
ACCEPTED MANUSCRIPT 1 Participant 3 1 4
-63 Baseline Action -82 -74 112
-66 Baseline Tone -67 -6 92
-77 Operant Action -101
-102 Operant Tone -331
-50 70
-230 76
5 -63
-118
-98
-193
-39
34
40
4
-162
-90
-125
-171
-31
-37
-26
-173
-94
-53
-15
-15
-13
17
88
42
-2
44
-69
18
7 8 9
11
37
-2
-90
-33
-113
-6
-58
-46
-62
19
71
103
-4
-35
-66
-154
104
153
63
99
93
338
122
-143
-152
-39
-14
-269
14 -91 15 -56 16 58 17 -75
EP
13
TE D
18
M AN U
12
2
SC
10
RI PT
6
Table B.4. Median (in ms) of all conditions for anodal stimulation of the left AG from
AC C
subjects in Experiment 2. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT 3 -96
-51
-141
-354
67
6
43
57
-40
-122
-91
-196
-12
105
-29
-19
-135
-122
-184
-136
-53
-62
-55
-79
-5
-3
-8
34
15
59
61
81
-33
-19
-109
-90
4 5 6
8 9 10
12
15 -58 16 83 17 -110 18 85 2 13
-123
-89
-41 72
-72
7
34
-77
-19
85
78
25
-56
-104
-224
130
92
78
-23
-157
-163
-87
-238
TE D
-168
-3
-55
M AN U
14
-139
SC
11
RI PT
7
-95
EP
Table B.5. Median (in ms) of all conditions for cathodal stimulation of the left AG from
AC C
subjects in Experiment 2. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ACCEPTED MANUSCRIPT 1 -108
-16
-83
-69
-153
-120
-131
-247
34
-30
32
-123
-72
-110
-75
-129
-85
89
-6
15
-102
-52
-128
-32
-15
-26
-61
3
-2
-74
16
38
26
3 4 5
7 8
10
12 -151 14 -108 15 -75 16 20 17 18 80 2 28 13
-33
-160
3
-45
-107
-11
-106
-51
-23
-52
-2
-4
68
66
-10
-7
-104
-87
114
44
77
285
82
190
-112
-95
-247
EP
-119
-175
-35
TE D
-85
-145
-33
M AN U
11
SC
9
RI PT
6
Table B.6. Median (in ms) of all conditions for sham from subjects in Experiment 2. Shaded
AC C
rows are excluded subjects.
Action binding: No significant effect of stimulation (F(2, 30)=2.70, p=0.08). Tone binding: No significant effect of stimulation (F(2, 30)=0.48, p=0.62).
ACCEPTED MANUSCRIPT Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-20
-37
11
-3
-21
17
42
-64
-31
31
-96
-43
-192
-190
-50
-19
59
-23
-25
0
-9
-93
-67
-6
-62
-20
9
64
-124
-105
-70
-133
-56
-38
-22
35
42
4
38
10
-97
-37
83
17
66
-21
-51
62
-15
-142
-82
-123
-116
-118
-24
-18
-96
-86
-42
-32
-20
-44
-101
-106
-74
2 3
7 8 9
12
-50
-13
-25
-73
-79
-56
5
-8
-63
-60
-10
-46
-45
-17
5
-14
-102
-113
-57
-19
-69
-1
-82
-77
-82
-96
-125
-68
-68
-28
-98
-84
-62
12
-47
-65
2
-23
-18
10
-46
-62
35
-12
-29
-64
60
-34
147
-38
-13
-40
7
50
-37
-36
140
138
13 14 15
17 -99 1 -48 6
AC C
18
EP
-2
TE D
16
10
M AN U
11
SC
5
RI PT
4
Table B.7. Median (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Baseline Operant Operant Operant Operant Tone Action Action Tone Left Tone Baseline Operant Operant Operant Operant Left Right Right Tone Action Action Tone Left Tone Left Right Right -38
37
-3
31
19
-27
-6
-143
-160
93
-24
-23
-8
-2
-44
-39
-62
-102
-73
-39
-95
-95
-20
-50
93
23
38
-15
-35
-12
125
107
-109
-17
-80
-153
-188
-101
-115
EP
AC C
SC
RI PT
14
M AN U
Baseline Baseline Action Action Participant Baseline Baseline Left Right Action Action Left Right 2 -33 -28 3 -40 -41 4 -35 -3 5 -49 -60 7 -145 -111 8 88 119 9 -31 -6 11 -157 -145 12 -70 -62 13 5 1 14 4 8 15 -82 -66 16 -73 -35 17 -89 -106 1 -79 -68 6 52 8 10 -88 -44 18 41 -11
-28
-24
-51
-112
-87
-24
-36
-68
-89
-93
-34
4
-8
-19
-9
-31
-142
-137
-57
-93
-29
-41
-43
-140
-153
-38
-69
-71
-161
-163
26
-67
-60
19
41
-6
-20
8
25
-56
50
-3
-25
51
11
61
20
3
-95
-187
TE D
Participant
Table B.8. Median (in ms) of all conditions for anodal stimulation of right AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
7 8 9 11 12 13 14 15 16 17 1 6 10 18
-20
-16
39
36
-111
-53
-59
-25
-31
-276
-278
-6
-30
97
-51
-38
-6
-27
-92
-76
-17
-71
-69
-103
-108
-94
-104
-36
-59
18
-17
-31
-12
5
31
-88
-63
-36
-41
-48
-10
23
48
21
32
-27
-71
-98
-64
-101
-94
-14
-71
-58
-46
-25
-14
-49
-76
-78
-44
-34
-2
72
53
-38
-74
-3
15
-5
-22
-29
-4
14
-76
-79
-92
-84
-74
-88
-61
-47
-1
1
55 -14
RI PT
5
6
SC
4
15
M AN U
3
-14
-29
-121
-94
-68
-54
-20
-61
-66
-135
-160
-52
-42
-71
-144
-110
24
-33
3
21
10
-13
34
3
-97
-41
58
16
-38
2
19
28
44
-3
33
105
76
64
TE D
2
EP
Table B.9. Median (in ms) of all conditions for sham from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded
AC C
subjects.
Action binding: No significant main effect of stimulation (F(2, 26)=0.14, p=0.87), no significant main effect of acting hand (F(1, 13)=1.40, p=0.26), no significant interaction (F(2, 26)=0.75, p=0.48).
Tone binding: Significant main effect of stimulation (F(2, 26)=7.24, p<0.01). Tone binding was significantly reduced by anodal stimulation of the putative left AG relative to both sham (t(13)=2.67, p=0.02) and anodal stimulation of putative right AG (t(13)=3.19, p=0.01). No significant main effect of acting hand (F(1, 13)=0.04, p=0.85), no significant interaction (F(2, 26)=1.57, p=0.23).
S0010-9452(15)00139-2
DOI:
10.1016/j.cortex.2015.04.015
Reference:
CORTEX 1456
To appear in:
Cortex
Received Date: 22 January 2015 Revised Date:
10 April 2015
Accepted Date: 18 April 2015
Please cite this article as: Khalighinejad N, Haggard P, Modulating human sense of agency with noninvasive brain stimulation, CORTEX (2015), doi: 10.1016/j.cortex.2015.04.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
Modulating human sense of agency with non-invasive brain stimulation
2
Nima Khalighinejad, Patrick Haggard
4
Institute of Cognitive Neuroscience, University College London, London WC1N 3AR, UK
RI PT
3
5
Correspondence should be addressed to Nima Khalighinejad, Institute of Cognitive Neuroscience,
7
University College London, Alexandra House, 17 Queen Square, London WC1N 3AR, UK. Tel: +44
8
(0)20 7679 5434, E-mail: [email protected]
10
Running title: Human sense of agency
14
15
16
17
EP
13
AC C
12
TE D
11
M AN U
9
SC
6
18
19
20
0
ACCEPTED MANUSCRIPT Abstract
22
Human voluntary actions are accompanied by a distinctive subjective experience termed “sense of
23
agency”. We performed three experiments using transcranial direct current stimulation (tDCS) to
24
modulate brain circuits involved in control of action, while measuring stimulation-induced changes in
25
one implicit measure of sense of agency, namely the perceived temporal relationship between a
26
voluntary action and tone triggered by the action. Participants perceived such tones as shifted towards
27
the action that caused them, relative to baseline conditions with tones but no actions. Actions that
28
caused tones were perceived as shifted towards the tone, relative to baseline actions without tones.
29
This ‘intentional binding’ was diminished by anodal stimulation of the left parietal cortex (targeting
30
the angular gyrus (AG)), and, to a lesser extent, by stimulation targeting the left dorsolateral
31
prefrontal cortex (DLPFC), (Experiment 1). Cathodal AG stimulation had no effect (Experiment 2).
32
Experiment 3 replicated the effect of left anodal AG stimulation for actions made with either the left
33
or the right hand, and showed no effect of right anodal AG stimulation. The angular gyrus has been
34
identified as a key area for explicit agency judgements in previous neuroimaging and lesion studies.
35
Our study provides new causal evidence that the left angular gyrus plays a key role in the perceptual
36
experience of agency.
37
Keywords: sense of agency; angular gyrus; dorsolateral prefrontal cortex; intentional binding; tDCS
39
40
41
SC
M AN U
TE D
EP AC C
38
RI PT
21
42
43
44 1
ACCEPTED MANUSCRIPT 1. Introduction
46
Healthy human adults have the feeling that they are able to control their own actions, and, through
47
them, external events. This is referred to as the sense of agency. Sense of agency is central to
48
individual goal-directed action, and also to social responsibility and punishment (Frith, 2014; Moretto,
49
Walsh, & Haggard, 2011). Moreover, many neurological and psychiatric disorders involve
50
abnormalities of agency (de Jong, 2011; Fletcher & Frith, 2009; Kranick & Hallett, 2013).
51
Despite extensive theoretical work on agency, its neural correlates are not fully understood.
52
Neuroimaging studies found activation of AG (Farrer et al., 2003; Farrer & Frith, 2002; Farrer et al.,
53
2008) and DLPFC (Fink et al., 1999) associated with agency tasks, but the activation of these areas
54
was always greater in the conflicting, non-agency condition than in the agency condition. In a recent
55
meta-analysis of sense of agency, the single most consistent result was activation of a broadly-defined
56
temporoparietal junction area conditions associated with reduced or absent sense of agency (Sperduti,
57
Delaveau, Fossati, & Nadel, 2011). This broad ‘non-agency’ area includes AG. Computational
58
models of predictive motor control offer an important theoretical framework for understanding
59
agency. An internal forward model uses efference copies of the motor command to predict outcomes
60
(Wolpert & Ghahramani, 2000). According to these models, sense of agency arises when there is a
61
match between the predicted and actual sensory outcome of the generated action. Conversely, if
62
current sensory information does not match the model’s prediction, then the corresponding sensory
63
event cannot be self-generated, and no sense of agency is experienced (Frith, Blakemore, & Wolpert,
64
2000).
65
Farrer et al., (2008) used this framework to interpret fMRI activations of AG in particular, suggesting
66
that AG processes discrepancies between intended action and its actual consequences. Her data
67
showed increased activations of AG when a detectable temporal discrepancy was inserted between an
68
action and visual feedback of the outcome, and also when participants explicitly rejected agency over
69
the viewed outcome.
AC C
EP
TE D
M AN U
SC
RI PT
45
2
ACCEPTED MANUSCRIPT Most of these studies used explicit agency attribution tasks, in which participants judge whether they
71
did or did not cause a specific sensory event. Synofzik, Vosgerau, & Newen, (2008) noted that one
72
feels a sense of agency when acting, even without making any explicit judgements. One suitable
73
measure of this pre-reflective, sensorimotor feeling of agency is the perceived temporal relationship
74
between a voluntary action and its sensory outcome (Moore & Obhi, 2012). The perceived time of
75
voluntary actions and their sensory consequences are attracted towards each other. This ‘intentional
76
binding’ is absent, or less prominent, for involuntary movements, and for associations between
77
external events not involving voluntary actions (Cravo, Claessens, & Baldo, 2009).
78
The neural bases of such feelings of agency are poorly understood. One neuroimaging study found a
79
neural correlate of intentional binding in the medial frontal cortex (Kühn, Brass, & Haggard, 2013). A
80
‘virtual lesion’ study showed that theta-burst stimulation over a slightly more anterior medial frontal
81
location reduced the intentional binding effect (Moore, Ruge, Wenke, Rothwell, & Haggard, 2010).
82
On the other hand, other lesion (Sirigu et al., 2004) and stimulation (Desmurget et al., 2009) studies
83
suggested an important role of parietal cortex in intentional action and agency, though these studies
84
did not use binding. To our knowledge, no previous causal study has investigated the influence of
85
both frontal and parietal areas on sense of agency using implicit measures. We therefore performed
86
three transcranial direct current stimulation (tDCS) experiments, to modulate excitability of key brain
87
circuits underlying the control of action, while measuring the effects on sense of agency, using
88
intentional binding. Our experiments investigated the respective contributions of parietal and frontal
89
areas to intentional binding as a proxy measure of agency (Experiment 1), their susceptibility to both
90
up- and down-regulation (Experiment 2), and their hemispheric specialisation (Experiment 3).
91
Based on the existing neuroimaging data investigating explicit agency judgement (Farrer et al., 2003;
92
Farrer & Frith, 2002; Farrer et al., 2008), we predicted that anodal stimulation of putative AG should
93
also influence the sense of agency, as measured by intentional binding. Importantly, such a result
94
would identify a causal role for AG in sense of agency, but would not conclusively identify how AG
95
computes agency. We also investigated the role of prefrontal areas in sense of agency. Studies of
96
frontal contributions to sense of agency are more equivocal. Neurostimulation (Moore et al., 2010)
AC C
EP
TE D
M AN U
SC
RI PT
70
3
ACCEPTED MANUSCRIPT 97
and neuroimaging (Kühn et al., 2013) studies of intentional binding found evidence for medial
98
prefrontal involvement, but studies of explicit agency judgements in tasks requiring a choice between
99
alternative actions (Chambon, Wenke, Fleming, Prinz, & Haggard, 2013) identified a more lateral prefrontal focus. DLPFC has also been identified as a key area for initiation (Jahanshahi et al., 1995)
101
and monitoring of voluntary action (Rowe, Hughes, & Nimmo-Smith, 2010). Given the relative
102
inaccessibility of medial prefrontal cortex to neurostimulation, we focussed here on the lateral
103
prefrontal cortex. The stimulations targeted primarily the left hemisphere, and participants made
104
actions with their right hand (experiments 1,2), or with either hand (experiment 3).
105
2. Materials and Methods
106
2.1. Participants
107
In total 55 healthy volunteers, 18-35 years of age (25 females) were recruited from the Institute of
108
Cognitive Neuroscience subject data pool for three separate experiments. All participants were right
109
handed, had normal or corrected to normal vision, had no history or family history of seizure, epilepsy
110
or any neurologic or psychiatric disorder and did not have any metallic or electronic object in the
111
head. Participants affirmed that they had not participated in any other brain stimulation experiment in
112
the last 48 hours, nor had consumed alcohol in the last 24 hours. The sample for Experiment 1
113
consisted of 18 participants (8 females), Experiment 2 consisted of 19 participants (10 females) and
114
Experiment 3 consisted of 18 participants (7 females). One participant failed to finish Experiment 2
115
due to lack of concentration, and was therefore excluded. Experimental design and procedure were
116
approved by the UCL research ethics committee, and followed the principles of the Declaration of
117
Helsinki. Transcranial stimulation followed established safety procedures (Nitsche et al., 2003;
118
Poreisz, Boros, Antal, & Paulus, 2007). Participants were paid a minimum amount for participating in
119
each session of the experiment. Participants were paid a small additional bonus at the end of the last
120
session.
121
2.2. Behavioural task
AC C
EP
TE D
M AN U
SC
RI PT
100
4
ACCEPTED MANUSCRIPT We used intentional binding paradigm as an implicit measure of agency (Fig. 1). The task was based
123
on previous studies (Haggard, Clark, & Kalogeras, 2002), and was programmed in LabVIEW 2012
124
(Austin, Texas). Participants viewed a clock hand rotating on a computer screen which was located
125
60cm in front of the participants in a quiet room. The initial clock position was random. Clock
126
rotation was initiated by participants pressing the return key on a keyboard. Each full rotation lasted
127
2560ms. Participants were instructed to look at the centre of the clock. They made voluntary actions,
128
when instructed, by pressing the return key with their right index finger (Experiments 1, 2), or by
129
pressing F9 or F4 with their right or left index finger, respectively (Experiment 3). Participants chose
130
for themselves when to make these voluntary actions. After each key press, the clock hand stopped at
131
a random location, participants made a time judgement according to condition (see later). Each
132
experimental session consisted of four types of trials, presented in separate blocked and randomised
133
conditions. At the beginning of each block, brief instructions for the relevant condition were displayed
134
on the screen. In the baseline action condition, participants had to press the key at a time of their own
135
free choice. The clock hand stopped after 1500-2500ms (at random), and participants then judged the
136
clock hand position at the time of their key press, entering their response on the keyboard. In this
137
condition, the participant’s actions produced no sensory outcome. In the baseline tone condition,
138
participants were instructed to look at the clock but not to press any key. While the clock was rotating,
139
a pure tone (1000Hz, 100ms duration) was played over a loudspeaker, 1750-4000ms (at random) after
140
the onset of the trial. Participants were then asked to judge the clock hand position at the time of the
141
tone. In the operant action condition, participants pressed a key at a time of their own choosing, and
142
each keypress produced a tone after 250ms. Participants judged the clock hand position at the moment
143
of pressing the key. Finally, the operant tone condition was similar to the operant action condition,
144
with the difference that participants had to judge the clock hand position at the time of the tone. Each
145
condition was tested in a separate block of 30 trials. The order of the blocks was randomised and there
146
was a 1 minute break between each block.
147
This common basic design was slightly changed according to the demands of each specific
148
experiment. In Experiment 3, text below and above the clock instructed participants to reply with
AC C
EP
TE D
M AN U
SC
RI PT
122
5
ACCEPTED MANUSCRIPT either their left or right index fingers. The order of hands was randomised in each block. The block
150
length was increased by 40 trials to allow sufficient trials for analysis of each hand’s data.
151
Before each experiment, participants were trained and familiarised with the task. They were reminded
152
to look at the centre of the clock, to avoid following the clock hand with their eyes, to be spontaneous
153
in their key presses and to be as precise as possible in their judgements, in particular not confining
154
themselves to those numbers 5,10,15… marked on the clock face. Each experimental session
155
consisted of four blocks and took approximately 20 minutes. The short duration of each individual
156
session was planned to coincide with the known effective period of tDCS.
157
2.3. tDCS
158
Direct current stimulation was delivered by StarStim noninvasive wireless neurostimulator
159
(Neuroelectrics, Barcelona, Spain). Circular rubber electrodes (25cm2) were covered in saline-soaked
160
sponges, installed in a 27 channel neoprene cap, and connected to a wireless current generator. tDCS
161
was then controlled by Neuroelectrics Instrument Controller (NIC v1.2) through a separate computer.
162
Current strength was set at 1mA in all experiments, generating a current density of 0.04mA/cm2 at the
163
scalp surface. For each experiment, all participants underwent three separate sessions of tDCS, two
164
effective stimulations and one sham session. The order of the sessions was randomised and
165
counterbalanced across participants. There was a minimum of 48 hours (and a maximum of 1 week)
166
between each stimulation session to minimise any potential carry over effects of tDCS (Nitsche et al.,
167
2008). The duration of stimulation in each session was set at 25 minutes, including 30s to ramp-up
168
and down the stimulating current. For the sham condition, electrical current was only applied during
169
the first and last 30 seconds of the stimulation, so as to induce the same cutaneous sensation as real
170
stimulation, and thus blind the participants as to stimulation condition. During the first 5min of each
171
stimulation, participants were asked to relax on their seats and close their eyes. This delay was
172
designed to allow potential neuro-modulatory effects to build up (Zwissler et al., 2014). Next,
173
participants began the behavioural task while stimulation continued (Fig. 2). All participants finished
174
the behavioural task approximately after 20min, the same time as the end of stimulation. In case
175
participants finished the task prior to the end of stimulation they were asked to remain seated until the
AC C
EP
TE D
M AN U
SC
RI PT
149
6
ACCEPTED MANUSCRIPT end of the stimulation. In case the task outlasted the stimulation, they continued to perform the task
177
without further stimulation. The task period never exceeded the stimulation period by more than 2
178
minutes.
179
Fig. 2 shows tDCS montages of the three experiments. In Experiment 1, the anodal electrode was
180
placed on the left DLPFC (F3 according to the 10/20 international EEG electrode placement) or
181
putative left AG (position P3) (Okamoto et al., 2004; Spitoni et al., 2013) in separate sessions. During
182
the sham session, the position of the stimulating electrode was counterbalanced between F3 and P3. In
183
all three sessions, the return electrode (cathodal) was placed on the right supraorbital area. For
184
Experiment 2, anode and cathode were placed on the putative left AG in separate sessions while the
185
return electrode was placed on the right supraorbital area. This arrangement was retained during the
186
sham session. Experiment 3 used a biparietal montage (Cohen Kadosh, Soskic, Iuculano, Kanai, &
187
Walsh, 2010; Hecht, Walsh, & Lavidor, 2010). For anodal stimulation of the putative left AG, the
188
anode was placed over P3 and cathode was placed over P4. This arrangement was reversed for anodal
189
stimulation of the putative right AG. For sham stimulation, the anode was pseudorandomly placed
190
either at P3 or P4. After each session participants were asked as part of debriefing if they had
191
experienced any notable effects of stimulation. No effects were reported other than mild tingling
192
sensations localised to the electrodes.
193
2.4. Data analysis
194
The difference between the judged clock hand position and the actual onset of the corresponding
195
event was calculated, giving a judgement error for each trial. A perceptual delay was represented by a
196
positive judgement error, and an anticipation by a negative judgement error. The mean and standard
197
deviation of the judgement errors across trials were then measured for each condition. Action binding
198
was defined as the shift of action toward its outcome, and was calculated by subtracting each
199
participant’s mean judgement error in the baseline action from that in the operant action condition.
200
Likewise, tone binding was defined as a shift in the perceived time of a tone towards the action that
201
caused it. Tone binding was calculated by subtracting each participant’s mean judgement error in
AC C
EP
TE D
M AN U
SC
RI PT
176
7
ACCEPTED MANUSCRIPT baseline tone condition from that in the operant tone condition. Thus, perceptual association of an
203
action with a subsequent tone produced a positive value for action binding, and a negative value for
204
tone binding. We analysed action and tone binding separately, since there is evidence from both
205
cognitive studies (Wolpe, Haggard, Siebner, & Rowe, 2013), and previous neurostimulation studies
206
(Moore et al., 2010), that they are driven by distinct mechanisms.
207
Some participants were excluded because of highly variable time judgement. A standard deviation of
208
judgement error across trials of over 250ms in any condition was used as a marker of poor time
209
perception. As in previous intentional binding experiments (Haggard, Aschersleben, Gehrke, & Prinz,
210
2002), these participants were excluded. On this basis, two participants were excluded from
211
Experiment 1, two from Experiment 2, and four from Experiment 3. Importantly, these exclusion
212
criteria are orthogonal to the mean judgement errors used for statistical inference.
213
In Experiments 1 and 2, inferential statistics were based on one-way repeated measures ANOVA,
214
with paired-sample t-tests for follow-up testing. Because our ANOVA had only 3 levels, Bonferroni
215
correction was not required for follow-up testing after a significant result (Cardinal & Aitken, 2013;
216
Meier, 2006). Additionally, in Experiment 1, we used linear discriminant analysis to determine which
217
percepts were most strongly affected by stimulation condition. Experiment 3 used repeated measures
218
ANOVA, with the additional factor of acting hand (right hand responses vs. left hand responses). A
219
final pooled analysis was performed to compare effects of stimulation common to all conditions.
220
3. Results
221
3.1. Experiment 1: frontal vs parietal anodal stimulation
222
This experiment compared the effects of frontal (targeting left DLPFC) and parietal (targeting left
223
AG) cortex stimulation on intentional binding for actions and tones. One-way repeated measures-
224
ANOVA with the factor of stimulation type (anodal frontal vs. anodal parietal vs. sham) showed that
225
action binding was not significantly affected by the type of stimulation (F(2, 30)=1.90, p=0.17,
226
η2=0.11). However, an identical ANOVA on tone binding showed significant differences (F(2,
227
30)=4.30, p=0.02, η2=0.22). Follow-up testing showed that tone binding was significantly reduced by
AC C
EP
TE D
M AN U
SC
RI PT
202
8
ACCEPTED MANUSCRIPT anodal stimulation of the putative left AG compared to sham (t(15)=2.67, p=0.02, d=0.43). Anodal
229
stimulation of the left DLPFC showed a clear trend to reduce tone binding, which approached the
230
border of conventional significance (t(15)=2.07, p=0.06, d=0.42). There was no significant difference
231
between the frontal and parietal stimulation (t(15)=-0.10, p=0.92, d=0.02) (Fig. 3; see also
232
supplementary table A.1-6).
233
We additionally performed the same analysis using median rather than mean judgement error in each
234
condition (see supplementary table B.1-3), since median measures are more robust than means to the
235
influence of outliers. The patterns of statistical significance were unchanged.
236
Finally, to confirm that anodal stimulation of putative AG affected primarily the operant tone
237
condition and not the baseline tone, the effect of stimulation type on participants’ judgement error in
238
baseline tone conditions was assessed using repeated-measure one-way ANOVA. There was no
239
significant main effect of stimulation type on baseline tone condition (F(2, 30)=0.58, p=0.56,
240
η2=0.04). This suggests that stimulation influenced a neurocognitive process that is present primarily
241
in the operant condition.
242
We additionally applied multivariate linear discriminant analysis (Krzanowski, 2000) to identify the
243
linear combination of action binding and tone binding variables that optimally discriminates the
244
different stimulation conditions. Linear discriminant analysis significantly differentiated the three
245
stimulation conditions (Wilks' Lambda=0.59, approx. F(4,58)=4.36, p<0.01). Inspection of canonical
246
coefficients showed that this difference was primarily due to tone binding (standardized canonical
247
coefficient 1.86) rather than action binding (-0.93) (The scores of the individual participants on the
248
first discriminant variate are shown in supplementary Fig. A.1). Post-hoc comparisons between
249
conditions showed a highly significant difference between parietal and sham stimulation (p<0.01;
250
standardised coefficients -1.03 for action binding, 2.19 for tone binding), and also a significant
251
difference between frontal and sham stimulation (p=0.04; standardised coefficients -0.82 for action
252
binding, 1.55 for tone binding). Interestingly, the frontal effect thus involved a slightly larger action
253
binding coefficient, considered relative to the tone binding coefficient, than did the parietal effect,
AC C
EP
TE D
M AN U
SC
RI PT
228
9
ACCEPTED MANUSCRIPT though no inferential statistics can be applied to this ratio. Frontal and parietal stimulation did not
255
differ significantly (p=0.74).
256
3.2. Experiment 2: anodal vs cathodal parietal stimulation
257
If anodal stimulation boosts activity in the left AG then cathodal stimulation of the same area should
258
lead to its suppression. Thus, if anodal stimulation decreases intentional binding, cathodal stimulation
259
should increase it. To test this hypothesis, putative left AG was exposed to anodal, cathodal or sham
260
stimulation in different sessions, and effects on intentional binding were evaluated.
261
One-way repeated measures-ANOVA with the factor of stimulation type (anodal parietal vs. cathodal
262
parietal vs. sham) was used for analysis. Action binding was not significantly affected by the type of
263
stimulation (F(2, 30)=0.50, p=0.61, η2=0.03). Tone binding was also unaffected by type of stimulation
264
(F(2, 30)=0.45, p=0.64, η2=0.03), contrary to our predictions from Experiment 1. Nevertheless, the
265
numerical effect of anodal stimulation of the putative left AG was in the same direction as Experiment
266
1, namely a decreased tone binding compared to sham and cathodal stimulation (Fig. 4; see also
267
supplementary table A.7-12).
268
We additionally performed the same analysis using median rather than mean judgement error in each
269
condition (see supplementary table B.4-6). The patterns of statistical significance were unchanged.
270
3.3. Experiment 3: left vs right parietal stimulation
271
Experiment 3 aimed to replicate the effects of Experiment 1, and additionally investigated the
272
lateralisation of intentional binding using a biparietal montage. The biparietal montage may provide a
273
higher local current density, because of the relatively short path between anode and cathode. In
274
addition, this montage controls for any possible effect of cathodal stimulation of prefrontal areas that
275
may occur with the conventional supraorbital placement of the cathode. Therefore, putative left and
276
right AG were exposed to anodal stimulation in separate sessions, and a third session involved sham
277
stimulation. We also investigated whether the putative AG involvement in intentional binding is hand-
278
specific or hemisphere-specific, by asking participants to make actions with either the left or right
AC C
EP
TE D
M AN U
SC
RI PT
254
10
ACCEPTED MANUSCRIPT hand, choosing randomly on each trial. Analysis of action binding showed no significant main effect
280
of stimulation type (F(2, 26)=0.06, p=0.94, η2=0.01) or acting hand F(1, 13)=0.10, p=0.76, η2=0.01)
281
and no significant interaction (F(2, 26)=0.89, p=0.42, η2=0.06). Analysis of tone binding showed a
282
highly significant main effect of stimulation (F(2, 26)=5.93, p<0.01, η2=0.31). Follow-up testing
283
showed that anodal stimulation of the putative left AG significantly decreased tone binding relative to
284
both sham stimulation (t(13)=2.55, p=0.02, d=0.40), and relative to anodal stimulation of the putative
285
right AG (t(13)=2.90, p=0.01, d=0.56). No significant difference was observed between anodal
286
stimulation of the putative right AG and sham (t(13)=-0.75, p=0.47, d=0.10) (Fig. 5; see also
287
supplementary table A.13-18). Acting hand had no significant main effect on tone binding (F(1,
288
13)=0.01, p=0.94, η2<0.01) and no interaction was observed between the stimulation and acting hand
289
(F(2, 26)=0.15, p=0.86, η2=0.01).
290
We additionally performed the same analysis using median rather than mean judgement error in each
291
condition (see supplementary table B.7-9). The patterns of statistical significance were unchanged.
292
To check whether the decrease in tone binding was primarily due to shifts in the operant tone, or in
293
the baseline tone condition, participants’ judgement errors in the baseline tone condition were
294
compared across the stimulation groups. Analysis showed no significant main effect of stimulation
295
type on baseline tone condition (F(2, 26)=0.50, p=0.60, η2=0.04).
296
3.4. Pooled data: anodal parietal vs sham stimulation
297
Anodal stimulation of the putative left AG was common to all three experiments reported here, as was
298
a sham stimulation condition, although the experiments differed in other respects. Therefore, we
299
pooled the data in these specific conditions across the 46 participants (21 female) from the three
300
experiments, in a single analysis. We found that anodal stimulation of putative left AG significantly
301
reduced the perceptual shift of tone toward action compared to sham (t(45)= 3.28, p<0.01, d=0.35),
302
but had no effect on action binding (t(45)=-1.37, p=0.18, d=0.22).
303
The anodal montage in the third session used a different return (cathode) location compared to the
304
first and second experiments. Our decision to pool data of the three studies was based on the common
AC C
EP
TE D
M AN U
SC
RI PT
279
11
ACCEPTED MANUSCRIPT placement of the anode across these three experiments, and the general observation that cathodal
306
effects on cognitive function are rare (Jacobson, Koslowsky, & Lavidor, 2012). In that case,
307
differences in cathode location may be relatively unimportant, and need not prevent pooling across
308
studies. However, because we cannot entirely exclude some contribution of cathodal location to our
309
main results, we ran a further pooled analysis using the left anodal stimulation conditions of
310
experiments 1 and 2 only, which share a supraorbital cathode location, but excluding experiment 3.
311
This analysis again found that anodal stimulation of putative left AG significantly reduced the
312
perceptual shift of tone toward action compared to sham (t(31)= 2.45, p=0.02, d=0.35), but had no
313
effect on action binding (t(31)=-1.48, p=0.15, d=0.32).
314
To investigate the generality of the anodal AG effect across experiments, we performed a mixed
315
ANOVA with a between-subject factor of experiments (1, 2 or 3), and a repeated-measures factor of
316
stimulation type (anodal left AG vs. sham). The main effect of stimulation type (F(1, 43)=10.7,
317
p<0.01, η2=0.20) recapitulating the pooled t-test reported above. There was no significant main effect
318
of experiment (F(2, 43)=0.80, p=0.45, η2=0.04). Importantly, there was no hint of interaction between
319
experiment and stimulation (F(2,43)=0.96, p=0.39, η2=0.04).
320
Our main inferences above are based on comparing experimental stimulation with sham. We therefore
321
additionally investigated whether sham stimulation had different effects in the three experiments. We
322
found no significant difference among the sham conditions for the 3 experiments for action binding
323
(F(2,43)=2.20, p=0.12), or tone binding (F(2,43)=1.10, p=0.33).
324
4. Discussion
325
4.1. Stimulation-induced modulation of the left parietal cortex
326
We performed a series of three tDCS experiments to investigate the neural circuits responsible for the
327
sense of agency, as measured by the perceptual association between the time of a voluntary action and
328
the time of a resulting auditory tone. We found a significant decrease in the binding of outcomes
329
towards actions after anodal stimulation of the putative left AG (Experiment 1). Anodal stimulation of
330
the left DLPFC also decreased action and tone binding compared to sham. DLPFC affected tone
AC C
EP
TE D
M AN U
SC
RI PT
305
12
ACCEPTED MANUSCRIPT binding as much as AG stimulation, but its effects were less consistent across participants.
332
Nevertheless, our discriminant analysis showed a significant effect of DLPFC stimulation compared
333
to sham, when action and tone binding were considered together.
334
Our tDCS stimulation of putative AG could have widespread effects across the inferior parietal cortex
335
(IPC), since tDCS has quite low spatial specificity. For example, anterior parts of the IPC may also be
336
affected. IPC is routinely activated in neuroimaging studies when participants judge whether their
337
own action, or some other cause, is responsible for a specific sensory event. In a study by Farrer and
338
Frith (2002), the IPC area was more active when participants attributed a visual event to another
339
person, rather than to themselves. Similarly, a PET study (Farrer et al., 2003) observed that neural
340
activity in IPC increased with the level of discrepancy between the executed and the observed action
341
on the screen. In an fMRI study (Farrer et al., 2008), the subjective feeling of loss of control
342
correlated with BOLD response in the AG, as did the awareness of temporal discrepancy between
343
action and feedback. The authors of those studies suggested that AG houses the comparison between
344
the efference copy of the intended action and the actual sensory outcome. Any mismatch between
345
these signals will then give rise to the explicit awareness of non-agency, or an external source of
346
action.
347
Our overall results are consistent with this view. We found that anodal stimulation of the putative AG
348
decreases intentional binding, our proxy measure for agency. Anodal stimulation is generally thought
349
to increase the activity of the cortical region immediately under the electrode. However, AG
350
activation is routinely associated with lack of agency, rather than with experience of positive agency
351
(Farrer & Frith, 2002; Farrer et al., 2008; Sperduti et al., 2011). Therefore, excitation of a neural
352
substrate of non-agency might be expected to decrease intentional binding. The conventional polarity-
353
specific (anode-boosting, cathode-suppressing) framework of tDCS was developed on the basis of
354
effects in primary motor cortex stimulation (Nitsche & Paulus, 2000). Its applicability to non-primary
355
areas and cognitive processing has recently been questioned (Horvath, Forte, & Carter, 2014).
356
Nevertheless, our findings are broadly compatible with the conventional polarity-specific view.
AC C
EP
TE D
M AN U
SC
RI PT
331
13
ACCEPTED MANUSCRIPT 4.2. Stimulation-induced modulation of the left frontal cortex
358
Since DLPFC is normally thought to facilitate intentional action, one may question why prefrontal
359
anodal tDCS did not increase intentional binding. Rowe et al., (2010) questioned whether DLPFC
360
played any important role in initiation of simple voluntary actions, such as those tested here, and
361
suggested a role in monitoring sequential action patterns instead. Fink et al., (1999) observed
362
activation of DLPFC using PET when an intentional action and its sensory outcome were
363
incompatible. Anodal tDCS over DLPFC might correspond to an increased coding for action-outcome
364
conflict, even though our task did not explicitly manipulate action-outcome compatibility. Our
365
discriminant analysis found some evidence consistent with this interpretation. However, this effect
366
was investigated in only one experiment, and achieved statistical significance in multivariate analysis,
367
but not in univariate analyses of action binding and tone binding separately. Therefore, further
368
research is required before a strong statement about frontal tDCS effects on sense of agency can be
369
made.
370
4.3. Polarity-specific effects of tDCS
371
Experiment 2 aimed to investigate whether parietal stimulation effects were polarity-specific. On one
372
model, tDCS would simply add neural noise, irrespective of polarity. On another model, anodal
373
stimulation would upregulate putative non-agency coding in AG, while cathodal stimulation should
374
down-regulate it. The result of anodal left AG stimulation in experiment 2 followed the expected
375
trend for tone binding, but did not reach statistical significance. Replication of statistically significant
376
results is an important and controversial issue in modern neuroscience (Cumming, 2005). All effects
377
measured in experiments represent a combination of the underlying ‘true’ effect, and noise.
378
Importantly, when a nonzero true effect indeed exists, but is modest in size, it is quite likely for the
379
effect to reach statistically significant levels in one study, but not in another. Thus, absence of a
380
significant anodal tDCS effect in experiment 2 does not prove that no true effect exists: we return to
381
this point later.
AC C
EP
TE D
M AN U
SC
RI PT
357
14
ACCEPTED MANUSCRIPT Experiment 2 found no significant difference between cathodal and sham stimulation, although we
383
had predicted that cathodal stimulation might enhance intentional binding. Although inhibitory
384
cathodal effects on motor function are well established, a recent review of 34 studies found that
385
cathodal inhibitory effects on cognitive function are rare (Jacobson, Koslowsky, et al., 2012). Another
386
possible reason for the absence of any significant cathodal AG effect in Experiment 2 could be the
387
placement of the anode electrode on the supraorbital area. This location is standard for tDCS studies
388
of action (Nitsche et al., 2008). However, it causes a strong current density close to the frontopolar
389
and prefrontal areas, where the anode is located. These areas may also contribute to intentional action
390
(Brass & Haggard, 2007). Thus, our montage for cathodal stimulation of AG in experiment 2 involved
391
anodal stimulation at a frontopolar site, which may not be strictly neutral for sense of agency. Future
392
studies could address this issue by using extracephalic cathode placement.
393
4.4. Hemispheric specialisation of the sense of agency
394
Experiment 3 avoided the potential confound of frontopolar stimulation using a biparietal montage.
395
This produces a higher current density in a small region surrounding the electrodes (Cohen Kadosh et
396
al., 2010; Nathan, Sinha, Gordon, Lesser, & Thakor, 1993), compared to the conventional supra-
397
orbital location. This might result in a more focal stimulation. More importantly, the biparietal
398
montage excludes the possibility that the significant effects of anodal AG stimulation in experiments
399
1 and 2 were in fact caused by cathodal frontopolar stimulation. Specifically, if the effects in
400
experiments 1 and 2 were merely due to cathodal frontopolar stimulation, then no effect of stimulation
401
should be found in experiment 3. The biparietal montage also allowed us to investigate lateralisation
402
of agency by varying both tDCS polarity and the hand used for action. Similar approaches have been
403
used previously in other studies (Bardi, Kanai, Mapelli, & Walsh, 2013; Jacobson, Goren, Lavidor, &
404
Levy, 2012).
405
The results of the third experiment replicated our previous findings. Anodal stimulation of putative
406
left AG significantly decreased tone binding compared to both sham and cathodal stimulation of the
AC C
EP
TE D
M AN U
SC
RI PT
382
15
ACCEPTED MANUSCRIPT same area. The tDCS effect was statistically equivalent whether the action was made with the left or
408
the right hand. No effects were observed with anodal stimulation of the putative right AG.
409
Experiment 3 does not support the alternative interpretation of experiments 1 and 2 based on a
410
putative cathodal frontopolar stimulation. In contrast, experiment 3 supports the interpretation of an
411
anodal left AG effect. We cannot conclusively rule out some contribution of frontopolar stimulation to
412
our results, but we can rule in a specific contribution of the left AG.
413
Experiment 3 adds several important elements to the previous studies. First, it demonstrates an
414
involvement of AG in a task involving randomised, stimulus-driven selection between alternative
415
actions, as opposed to mere repetition of a simple action. Second, it suggests that left, but not right
416
AG is responsible for action-outcome binding for actions made by either hand. We found no
417
interaction between stimulation and hand used for action. Previous neuroimaging studies have
418
reported activation corresponding to non-agency judgements in both left and right AG. Interestingly,
419
right AG activations appeared to dominate (Farrer et al., 2003; Farrer & Frith, 2002; Farrer et al.,
420
2008), in contrast to our finding. However, in a more recent fMRI study, Lee & Reeve, (2013)
421
reported higher activity in the left AG during non-self-determined behaviour, consistent with our
422
hypothesis in Experiment 1 that anodal AG stimulation activates a neural code for ‘non-agency’.
423
Finally, hemispheric specialisation of agency could plausibly depend on the task used, and the type of
424
agency judgement. Previous neuroimaging studies generally used explicit judgements of agency, and
425
often used complex manual actions with visual feedback (Farrer & Frith, 2002; Farrer et al., 2008;
426
Sperduti et al., 2011). We are not aware of any neuroimaging study investigating the hemispheric
427
lateralisation of low-level implicit measures of agency.
428
4.5. Limitations
429
The results of experiment 3 by themselves could not distinguish between an effect of anodal
430
stimulation of the putative left AG from an effect of cathodal stimulation of the putative right AG.
431
However, this result does allow us to exclude a model in which tDCS simply acts to increase neural
432
noise, irrespective of polarity. Moreover, our experiment 1 found some evidence of a left-hemisphere
AC C
EP
TE D
M AN U
SC
RI PT
407
16
ACCEPTED MANUSCRIPT anodal tDCS effect, while our experiment 2 found no evidence of any cathodal effect (though in the
434
left hemisphere, rather than the right). Cathodal stimulation effects in cognitive tasks are reported to
435
be weak (Jacobson, Koslowsky, et al., 2012). Therefore, we provisionally favour an interpretation of
436
experiment 3 based on a left parietal anodal effect, rather than a right-hemisphere cathodal effect.
437
Further research would be required to draw a definitive conclusion.
438
Our study is further limited because we did not control for cases of crossmodal binding in the absence
439
of active movement. Therefore, we cannot exclude the possibility that AG stimulation influenced
440
some general feature of time perception, as opposed to temporal processing specific to agency.
441
However, several studies have shown stronger binding between voluntary actions and outcomes than
442
between other, similarly paired, events, including involuntary movements and outcomes (Engbert,
443
Wohlschläger, Thomas, & Haggard, 2007; Haggard et al., 2002) or pairs of sensory stimuli (Haggard,
444
Aschersleben, Gehrke, & Prinz, 2002; Haggard, Martin, Taylor-Clarke, Jeannerod, & Franck, 2003).
445
Moreover, other studies have investigated effects of parietal tDCS on time perception in general, in
446
the absence of action: and these studies found no effect (Woods et al., 2014). Thus, the weight of
447
other studies suggests that the intentional binding phenomenon reflects a distortion of perceptual
448
timing that is, at least partly, specific to voluntary action.
SC
M AN U
TE D
449
RI PT
433
4.6. Dissociation between action binding and tone binding
451
Anodal stimulation over putative left AG was a common condition in all 3 experiments. Accordingly,
452
we could perform a pooled analysis of intentional binding results to compare this to the sham
453
stimulations that were also included in each experiment. This analysis showed a highly significant
454
reduction in tone binding with anodal stimulation of the putative left AG. We found no overall effect
455
on action binding. Dissociations between action binding and tone binding have been reported
456
previously (Wolpe et al., 2013), so it is possible that left parietal cortex is concerned primarily with
457
tone binding, rather than with action binding. This conclusion would be consistent with previous
458
studies suggesting that the AG processes mismatches in action outcomes (Farrer et al., 2008). On the
459
other hand, recent studies of explicit agency judgement suggest that AG also processes prospective,
AC C
EP
450
17
ACCEPTED MANUSCRIPT premotor information arising during action selection (Chambon, Moore, & Haggard, 2014; Chambon
461
et al., 2013).
462
Both online prospective and retrospective processes contribute to the intentional binding phenomenon
463
(Moore and Haggard, 2008). The experimental design used here cannot identify the independent
464
contribution of each process. However, binding of action towards outcome may rely more on
465
prospective processes during action selection, while perceptual shift of outcome toward action may
466
depend on retrospective, more inferential processes, triggered by reafferent signals about action
467
outcome (Chambon, Moore, & Haggard, 2014). Future studies may address this issue by designing
468
new paradigms which dissociate prospective and retrospective components of agency and examine the
469
role of dLPFC and AG in each of these components.
470
4.7. Conclusion and clinical implications
471
Sense of agency is an important and distinctive feature of human voluntary action. We used a causal
472
intervention (tDCS) and an implicit perceptual measure of sense of agency (intentional binding) to
473
examine the role of different brain areas in sense of agency. Anodal stimulation of parietal cortex
474
consistently reduced the binding of tones towards actions. We hypothesised that the angular gyrus
475
might contribute to the sense of agency by monitoring the linkage of actions to outcomes, or,
476
alternatively and equivalently, failures of such linkage. Anodal stimulation of this area may
477
correspond to artificial boosting of a mismatch detection process.
478
Sense of agency is altered following several classes of psychiatric and neurological disorders. In
479
particular, patients with apraxia following lesions to the left parietal fail to recognise the source of a
480
viewed manual gesture (Sirigu, Daprati, Pradat-Diehl, Franck, & Jeannerod, 1999). This deficit is
481
formally equivalent to an overestimation of agency in an explicit judgement task, consistent with
482
damage to a neural centre detecting mismatches. The posterior form of ‘alien hand syndrome’ is also
483
associated with contralateral parietal lesions (e.g., (Kloesel, Czarnecki, Muir, & Keller, 2010)).
484
Interestingly, these patients show involuntary and spontaneous movements of the contralateral limb,
485
but may correctly perceive that they are not agents over these actions. The capacity for voluntary
AC C
EP
TE D
M AN U
SC
RI PT
460
18
ACCEPTED MANUSCRIPT 486
movement is often preserved. Quantitative assessment of the implicit sense of agency in parietal
487
patients would be of considerable value in understanding the neural basis of sense of agency.
488
Competing interests
490
The authors declare no competing financial interests.
491
Acknowledgements
492
This work was supported by the European Research Council Advanced Grant HUMVOL. PH was
493
additionally supported by a Professorial Research Fellowship from the ESRC. The funding sources
494
had no involvement, in study design; in the collection, analysis or interpretation of data; in the writing
495
of the report; or in the decision to submit the article for publication.
496
We thank Andrea Desantis for helpful comments during the preparation of the manuscript.
SC
M AN U
497
502
503
EP
501
AC C
500
TE D
498
499
RI PT
489
504
505
506
507 19
ACCEPTED MANUSCRIPT 508
References
509
Bardi, L., Kanai, R., Mapelli, D., & Walsh, V. (2013). Direct current stimulation (tDCS) reveals
510
parietal asymmetry in local/global and salience-based selection. Cortex, 49(3), 850–860.
511
http://doi.org/10.1016/j.cortex.2012.04.016 Brass, M., & Haggard, P. (2007). To do or not to do: the neural signature of self-control. The Journal
RI PT
512 513
of Neuroscience: The Official Journal of the Society for Neuroscience, 27(34), 9141–9145.
514
http://doi.org/10.1523/JNEUROSCI.0924-07.2007
517 518 519
SC
516
Buxbaum, L. J., Sirigu, A., Schwartz, M. F., & Klatzky, R. (2003). Cognitive representations of hand posture in ideomotor apraxia. Neuropsychologia, 41(8), 1091–1113.
Cardinal, R. N., & Aitken, M. R. F. (2013). ANOVA for the Behavioral Sciences Researcher.
M AN U
515
Psychology Press.
Chambon, V., Moore, J. W., & Haggard, P. (2014). TMS stimulation over the inferior parietal cortex
520
disrupts prospective sense of agency. Brain Structure & Function.
521
http://doi.org/10.1007/s00429-014-0878-6
Chambon, V., Wenke, D., Fleming, S. M., Prinz, W., & Haggard, P. (2013). An Online Neural
TE D
522 523
Substrate for Sense of Agency. Cerebral Cortex, 23(5), 1031–1037.
524
http://doi.org/10.1093/cercor/bhs059
Cohen Kadosh, R., Soskic, S., Iuculano, T., Kanai, R., & Walsh, V. (2010). Modulating neuronal
EP
525
activity produces specific and long-lasting changes in numerical competence. Current
527
Biology: CB, 20(22), 2016–2020. http://doi.org/10.1016/j.cub.2010.10.007
528 529 530
AC C
526
Cravo, A. M., Claessens, P. M. E., & Baldo, M. V. C. (2009). Voluntary action and causality in temporal binding. Experimental Brain Research, 199(1), 95–99. http://doi.org/10.1007/s00221-009-1969-0
531
Cumming, G. (2005). Understanding the Average Probability of Replication Comment on Killeen
532
(2005). Psychological Science, 16(12), 1002–1004. http://doi.org/10.1111/j.1467-
533
9280.2005.01650.x
20
ACCEPTED MANUSCRIPT 534
De Jong, B. M. (2011). Neurology of widely embedded free will. Cortex; a Journal Devoted to the
535
Study of the Nervous System and Behavior, 47(10), 1160–1165.
536
http://doi.org/10.1016/j.cortex.2011.06.011
537
Desmurget, M., Reilly, K. T., Richard, N., Szathmari, A., Mottolese, C., & Sirigu, A. (2009). Movement intention after parietal cortex stimulation in humans. Science (New York, N.Y.),
539
324(5928), 811–813. http://doi.org/10.1126/science.1169896
540
RI PT
538
Engbert, K., Wohlschläger, A., Thomas, R., & Haggard, P. (2007). Agency, subjective time, and other minds. Journal of Experimental Psychology. Human Perception and Performance, 33(6),
542
1261–1268. http://doi.org/10.1037/0096-1523.33.6.1261
544 545
Farrer, C., Franck, N., Georgieff, N., Frith, C. D., Decety, J., & Jeannerod, M. (2003). Modulating the experience of agency: a positron emission tomography study. NeuroImage, 18(2), 324–333.
M AN U
543
SC
541
Farrer, C., Frey, S. H., Horn, J. D. V., Tunik, E., Turk, D., Inati, S., & Grafton, S. T. (2008). The
546
Angular Gyrus Computes Action Awareness Representations. Cerebral Cortex, 18(2), 254–
547
261. http://doi.org/10.1093/cercor/bhm050
Farrer, C., & Frith, C. D. (2002). Experiencing oneself vs another person as being the cause of an
TE D
548 549
action: the neural correlates of the experience of agency. NeuroImage, 15(3), 596–603.
550
http://doi.org/10.1006/nimg.2001.1009
Fink, G. R., Marshall, J. C., Halligan, P. W., Frith, C. D., Driver, J., Frackowiak, R. S., & Dolan, R. J.
EP
551
(1999). The neural consequences of conflict between intention and the senses. Brain: A
553
Journal of Neurology, 122 ( Pt 3), 497–512.
AC C
552
554
Fletcher, P. C., & Frith, C. D. (2009). Perceiving is believing: a Bayesian approach to explaining the
555
positive symptoms of schizophrenia. Nature Reviews. Neuroscience, 10(1), 48–58.
556 557 558 559
http://doi.org/10.1038/nrn2536
Frith, C. D. (2014). Action, agency and responsibility. Neuropsychologia, 55, 137–142. http://doi.org/10.1016/j.neuropsychologia.2013.09.007 Frith, C. D., Blakemore, & Wolpert, D. M. (2000). Abnormalities in the awareness and control of
560
action. Philosophical Transactions of the Royal Society of London. Series B: Biological
561
Sciences, 355(1404), 1771–1788. http://doi.org/10.1098/rstb.2000.0734 21
ACCEPTED MANUSCRIPT 562
Haggard, P., Aschersleben, G., Gehrke, J., & Prinz, W. (2002). Action, binding, and awareness. In W.
563
Prinz & B. Hommel (Eds.), Common Mechanisms in Perception and Action B. (Vol. 19, pp.
564
266–285). OXFORD UNIVERSITY PRESS. Retrieved from http://eprints.ucl.ac.uk/4192/
566 567
Haggard, P., Clark, S., & Kalogeras, J. (2002). Voluntary action and conscious awareness. Nature Neuroscience, 5(4), 382–385. http://doi.org/10.1038/nn827 Haggard, P., Martin, F., Taylor-Clarke, M., Jeannerod, M., & Franck, N. (2003). Awareness of action
568
in schizophrenia. Neuroreport, 14(7), 1081–1085.
569
http://doi.org/10.1097/01.wnr.0000073684.00308.c0
Hecht, D., Walsh, V., & Lavidor, M. (2010). Transcranial Direct Current Stimulation Facilitates
SC
570
RI PT
565
Decision Making in a Probabilistic Guessing Task. The Journal of Neuroscience, 30(12),
572
4241–4245. http://doi.org/10.1523/JNEUROSCI.2924-09.2010
M AN U
571
Horvath, J. C., Forte, J. D., & Carter, O. (2014). Evidence that transcranial direct current stimulation
574
(tDCS) generates little-to-no reliable neurophysiologic effect beyond MEP amplitude
575
modulation in healthy human subjects: A systematic review. Neuropsychologia, 66C, 213–
576
236. http://doi.org/10.1016/j.neuropsychologia.2014.11.021
577
TE D
573
Jacobson, L., Goren, N., Lavidor, M., & Levy, D. A. (2012). Oppositional transcranial direct current stimulation (tDCS) of parietal substrates of attention during encoding modulates episodic
579
memory. Brain Research, 1439, 66–72. http://doi.org/10.1016/j.brainres.2011.12.036
581 582 583 584
Jacobson, L., Koslowsky, M., & Lavidor, M. (2012). tDCS polarity effects in motor and cognitive domains: a meta-analytical review. Experimental Brain Research, 216(1), 1–10.
AC C
580
EP
578
http://doi.org/10.1007/s00221-011-2891-9
Jahanshahi, M., Jenkins, I. H., Brown, R. G., Marsden, C. D., Passingham, R. E., & Brooks, D. J. (1995). Self-initiated versus externally triggered movements. I. An investigation using
585
measurement of regional cerebral blood flow with PET and movement-related potentials in
586
normal and Parkinson’s disease subjects. Brain: A Journal of Neurology, 118 ( Pt 4), 913–
587
933.
22
ACCEPTED MANUSCRIPT 588
Kloesel, B., Czarnecki, K., Muir, J. J., & Keller, A. S. (2010). Sequelae of a left-sided parietal stroke:
589
posterior alien hand syndrome. Neurocase, 16(6), 488–493.
590
http://doi.org/10.1080/13554794.2010.497154
592 593 594 595
Kranick, S. M., & Hallett, M. (2013). Neurology of volition. Experimental Brain Research, 229(3), 313–327. http://doi.org/10.1007/s00221-013-3399-2
RI PT
591
Krzanowski, W. J. (2000). Principles of Multivariate Analysis: A User’s Perspective (2 edition). Oxford Oxfordshire ; New York: OUP Oxford.
Kühn, S., Brass, M., & Haggard, P. (2013). Feeling in control: Neural correlates of experience of agency. Cortex, 49(7), 1935–1942. http://doi.org/10.1016/j.cortex.2012.09.002
597
Lee, W., & Reeve, J. (2013). Self-determined, but not non-self-determined, motivation predicts
SC
596
activations in the anterior insular cortex: an fMRI study of personal agency. Social Cognitive
599
and Affective Neuroscience, 8(5), 538–545. http://doi.org/10.1093/scan/nss029
M AN U
598
600
Meier, U. (2006). A note on the power of Fisher’s least significant difference procedure.
601
Pharmaceutical Statistics, 5(4), 253–263. http://doi.org/10.1002/pst.210 Moore, J. W., Middleton, D., Haggard, P., & Fletcher, P. C. (2012). Exploring implicit and explicit
TE D
602 603
aspects of sense of agency. Consciousness and Cognition, 21(4), 1748–1753.
604
http://doi.org/10.1016/j.concog.2012.10.005
607 608 609 610 611
EP
606
Moore, J. W., & Obhi, S. S. (2012). Intentional binding and the sense of agency: a review. Consciousness and Cognition, 21(1), 546–561. http://doi.org/10.1016/j.concog.2011.12.002 Moore, J. W., Ruge, D., Wenke, D., Rothwell, J., & Haggard, P. (2010). Disrupting the experience of
AC C
605
control in the human brain: pre-supplementary motor area contributes to the sense of agency. Proceedings. Biological Sciences / The Royal Society, 277(1693), 2503–2509. http://doi.org/10.1098/rspb.2010.0404
Moretto, G., Walsh, E., & Haggard, P. (2011). Experience of agency and sense of responsibility.
612
Consciousness and Cognition, 20(4), 1847–1854.
613
http://doi.org/10.1016/j.concog.2011.08.014
614 615
Nathan, S. S., Sinha, S. R., Gordon, B., Lesser, R. P., & Thakor, N. V. (1993). Determination of current density distributions generated by electrical stimulation of the human cerebral cortex. 23
ACCEPTED MANUSCRIPT 616
Electroencephalography and Clinical Neurophysiology, 86(3), 183–192.
617
http://doi.org/10.1016/0013-4694(93)90006-H Nitsche, M. A., Cohen, L. G., Wassermann, E. M., Priori, A., Lang, N., Antal, A., Pascual-Leone, A.
619
(2008). Transcranial direct current stimulation: State of the art 2008. Brain Stimulation, 1(3),
620
206–223. http://doi.org/10.1016/j.brs.2008.06.004
621
RI PT
618
Nitsche, M. A., Liebetanz, D., Lang, N., Antal, A., Tergau, F., & Paulus, W. (2003). Safety criteria for transcranial direct current stimulation (tDCS) in humans. Clinical Neurophysiology:
623
Official Journal of the International Federation of Clinical Neurophysiology, 114(11), 2220–
624
2222; author reply 2222–2223.
626 627
Nitsche, M. A., & Paulus, W. (2000). Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. The Journal of Physiology, 527 Pt 3, 633–639.
M AN U
625
SC
622
Okamoto, M., Dan, H., Sakamoto, K., Takeo, K., Shimizu, K., Kohno, S., Dan, I. (2004). Threedimensional probabilistic anatomical cranio-cerebral correlation via the international 10–20
629
system oriented for transcranial functional brain mapping. NeuroImage, 21(1), 99–111.
630
http://doi.org/10.1016/j.neuroimage.2003.08.026
631
TE D
628
Poreisz, C., Boros, K., Antal, A., & Paulus, W. (2007). Safety aspects of transcranial direct current stimulation concerning healthy subjects and patients. Brain Research Bulletin, 72(4-6), 208–
633
214. http://doi.org/10.1016/j.brainresbull.2007.01.004
EP
632
Rowe, J. B., Hughes, L., & Nimmo-Smith, I. (2010). Action selection: a race model for selected and
635
non-selected actions distinguishes the contribution of premotor and prefrontal areas.
636 637 638 639
AC C
634
NeuroImage, 51(2), 888–896. http://doi.org/10.1016/j.neuroimage.2010.02.045
Sirigu, A., Daprati, E., Ciancia, S., Giraux, P., Nighoghossian, N., Posada, A., & Haggard, P. (2004). Altered awareness of voluntary action after damage to the parietal cortex. Nature Neuroscience, 7(1), 80–84. http://doi.org/10.1038/nn1160
640
Sirigu, A., Daprati, E., Pradat-Diehl, P., Franck, N., & Jeannerod, M. (1999). Perception of self-
641
generated movement following left parietal lesion. Brain, 122(10), 1867–1874.
642
http://doi.org/10.1093/brain/122.10.1867
24
ACCEPTED MANUSCRIPT 643
Sperduti, M., Delaveau, P., Fossati, P., & Nadel, J. (2011). Different brain structures related to self-
644
and external-agency attribution: a brief review and meta-analysis. Brain Structure &
645
Function, 216(2), 151–157. http://doi.org/10.1007/s00429-010-0298-1
646
Spitoni, G. F., Pireddu, G., Cimmino, R. L., Galati, G., Priori, A., Lavidor, M., Pizzamiglio, L. (2013). Right but not left angular gyrus modulates the metric component of the mental body
648
representation: a tDCS study. Experimental Brain Research. Experimentelle Hirnforschung.
649
Expérimentation Cérébrale, 228(1), 63–72. http://doi.org/10.1007/s00221-013-3538-9
650
RI PT
647
Synofzik, M., Vosgerau, G., & Newen, A. (2008). Beyond the comparator model: a multifactorial two-step account of agency. Consciousness and Cognition, 17(1), 219–239.
652
http://doi.org/10.1016/j.concog.2007.03.010
Wolpe, N., Haggard, P., Siebner, H. R., & Rowe, J. B. (2013). Cue integration and the perception of
M AN U
653
SC
651
654
action in intentional binding. Experimental Brain Research, 229(3), 467–474.
655
http://doi.org/10.1007/s00221-013-3419-2
657 658
Wolpert, D. M., & Ghahramani, Z. (2000). Computational principles of movement neuroscience. Nature Neuroscience, 3 Suppl, 1212–1217. http://doi.org/10.1038/81497
TE D
656
Woods, A. J., Hamilton, R. H., Kranjec, A., Minhaus, P., Bikson, M., Yu, J., & Chatterjee, A. (2014). Space, time, and causality in the human brain. NeuroImage, 92, 285–297.
660
http://doi.org/10.1016/j.neuroimage.2014.02.015
662 663 664
665
Zwissler, B., Sperber, C., Aigeldinger, S., Schindler, S., Kissler, J., & Plewnia, C. (2014). Shaping Memory Accuracy by Left Prefrontal Transcranial Direct Current Stimulation. The Journal of
AC C
661
EP
659
Neuroscience, 34(11), 4022–4026. http://doi.org/10.1523/JNEUROSCI.5407-13.2014
666
667
668 25
ACCEPTED MANUSCRIPT Captions
670
Fig. 1. Schematic of intentional binding. Action and tone shifts are measured by subtracting each
671
participant’s mean judgement error in baseline conditions from judgement error in operant conditions.
672
These shifts serve as measures of intentional binding. Vertical bars and thin arrows represent mean
673
judgement errors in each condition. Thick arrows represent binding effects. See text for full
674
explanation.
675
Fig. 2. A. tDCS montage and study design (anode +, cathode -). See text for explanation. In
676
Experiment 1, to control for the cutaneous sensation of all three locations, all sponges were kept in
677
place across the three sessions. However, only two of them were actually functioning in each session.
678
B. Stimulation protocol. The order of conditions was randomised within each session.
679
Fig. 3. Intentional binding in Experiment 1. A) The dashed line indicates the perceived time of either
680
action or tone in the corresponding baseline condition. A separate baseline condition was used for
681
each session, and differences in baseline values across sessions have been removed for display
682
purposes. Binding effects are drawn to scale, and values are in milliseconds. B) Mean binding effects
683
in ms. The sign of tone binding effects has been inverted to allow for comparison with action binding.
684
Error bars show standard error of the mean. * p<0.05.
685
Fig. 4. Intentional binding in Experiment 2. Format as in Fig. 3.
686
Fig. 5. Intentional binding in Experiment 3. Format as in Fig. 3.
AC C
EP
TE D
M AN U
SC
RI PT
669
26
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT Baseline Action
Baseline Tone
Operant Action
Operant Tone
1
-9
1
-8
-59
3
-68
73
-24
14
4
-14
177
-25
60
A.1.
5
-194
-39
-78
134
Mea
6
52
130
86
29
n (in
7
-54
24
-24
8
22
69
75
9
-84
14
11
-56
-61
12
89
13
-92
14
-80
15
-18
16
-8
17
16
18
RI PT
Participant
33
Tabl e
ms)
SC
30
of
-48
-83
-161
all
M AN U
-86
152
90
con
-13
-48
-192
ditio
-6
-52
-150
-31
-21
-94
22
31
-88
-86
-48
-166
-95
-48
-41
-90
dal
2
27
24
144
22
stim
10
-74
-36
-6
-241
ulati
EP
TE D
18
AC C
on of left dLPFC from subjects in Experiment 1. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ns for ano
ACCEPTED MANUSCRIPT 1
113
88
140
91
3
60
98
73
62
4
207
151
97
183
5
133
95
191
164
6
121
128
89
100
7
91
124
70
8
70
78
62
9
37
86
47
11
73
89
12
74
83
13
96
91
14
33
15
39
16
40
17
70
18
84
2
134
10
105
SC
RI PT
Tab le A.2 .
104
Sta
89
nda
91
rd
141
76
82
78
111
69
44
73
114
49
62
of
113
50
78
jud
153
64
113
ge
106
68
88
118
213
190
296
240
93
dev iati on
me nt err
EP
TE D
M AN U
116
or across trials (in ms) of all conditions for anodal stimulation of left dLPFC from subjects in
AC C
Experiment 1. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ACCEPTED MANUSCRIPT 26
10
91
-55
3
-52
27
-70
-30
4
-6
126
15
-14
5
-21
-21
-37
83
6
-41
95
49
48
7
-114
-1
-108
-26
8
19
83
77
112
9
-51
33
-93
-41
11
-93
-45
-71
-154
12
117
46
199
39
13
-83
-27
-81
-236
14
-77
-14
-80
-168
15
-73
-95
-40
-41
16
20
64
35
-58
17
-1
-112
-63
-108
2
SC
M AN U
-105
-44
-103
-110
176
42
143
33
-87
-74
-20
-117
EP
10
TE D
18
RI PT
1
AC C
Table A.3. Mean (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Baseline Action
Baseline Tone
Operant Action
Operant Tone
1
86
79
112
93
3
59
92
74
72
4
65
146
72
120
5
139
99
154
88
6
86
161
76
7
105
120
87
8
60
83
80
9
46
72
44
11
107
64
12
60
96
13
114
14
48
15
60
16
41
17
93
18
82
2
284
10
55
SC
RI PT
Participant
125 86
70 83
120
72
87
112
100
104
47
54
56
69
37
42
75
72
72
130
57
211
136
76
136
117
181
106
91
49
118
EP
TE D
M AN U
129
AC C
Table A.4. Standard deviation of judgement error across trials (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
Tab -25
-14
33
-175
3
-59
42
230
13
4
9
208
-2
-147
5
1
0
-42
6
59
45
97
7
-21
-34
-68
8
17
82
9
-113
37
11
-86
-39
12
11
13
-172
14
-71
15
-35
16
-11
17
RI PT
1
88
le A.5 . Me
-161
an
115
48
(in
-54
-112
-40
-190
16
183
13
-62
-58
-227
25
-37
-71
con
-32
-44
-46
diti
72
32
-96
ons
-58
-82
-72
-83
for
18
-85
7
-74
-122
2
28
39
113
13
-57
6
-29
-127
M AN U
TE D
EP
10
SC
-44
of
AC C Baseline Action
all
sha m
from subjects in Experiment 1. Shaded rows are data from excluded subjects.
Participant
ms)
Baseline Tone
Operant Action
Operant Tone
91
121
77
98
3
68
64
70
53
4
78
161
80
195
5
137
115
99
83
6
137
115
94
101
7
90
113
78
8
80
47
102
9
69
132
78
11
82
92
12
80
111
13
101
14
47
15
48
16
41
17
70
18
89
2
155
10
62
SC
RI PT
1
M AN U
ACCEPTED MANUSCRIPT
95
70
92
77
172
62
120
117
122
47
51
73
77
42
38
80
45
69
105
68
83
131
83
116
88
271
136
85
65
77
AC C
EP
TE D
73
Table A.6. Standard deviation of judgement error across trials (in ms) of all conditions for sham from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
-61
-61
-87
-99
-86
-75
-112
-341
111
79
84
97
-72
-116
-107
-173
-38
44
25
4
-167
-70
-130
-169
-40
-51
-29
-146
-82
-59
-16
-17
-9
27
84
30
1 3
5 6 7
9
11 44
57
45
-24
20
-74
-35
-115
-14
-49
-64
-54
14
59
78
95
3
-77
-28
-75
-167
101
161
65
101
83
371
146
-158
-58
-10
-244
12 -68 14 -83 15 -53
2 13
EP
18
TE D
16 17
M AN U
10
SC
8
RI PT
4
AC C
-151
Table A.7. Mean (in ms) of all conditions for anodal stimulation of the left AG from subjects in Experiment 2. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action 1
Baseline Tone mn2
Operant Action n3
Operant Tone mn4
52
57
51
63
74
93
61
68
93
88
68
92
69
86
61
116
67
76
69
74
86
95
74
112
63
87
58
99
1 3
6 7
AC C
EP
TE D
M AN U
8
SC
5
RI PT
4
ACCEPTED MANUSCRIPT 9
4 5 6
83 Operant Action 94
-87
-24
-42
11 12 14 15 16
7 17 8 18 9 2 10 13
198 -111 83 77 53 -54 47 -12 62 -147 40 -55 55 -54 211 -7 123
-62
71 77
37
57 -135 96 95 51 -127 54 -56 50 -11 166 44 83
65 -121 81 49 92 -89 66 -28 77 -172 48 -58 60 -15 209 20 121
76 Operant Tone 99 -212 90 -362 62 62 78 -186 62 -20 63 -142 111 -143 71 -2 214 -81 140
RI PT
3
69 Baseline Tone 57
SC
1
68 Baseline Action 65
M AN U
Participant 10
Table A.8. Standard deviation of judgement error across trials (in ms) of all conditions for
AC C
EP
excluded subjects.
TE D
anodal stimulation of the left AG from subjects in Experiment 2. Shaded rows are data from
ACCEPTED MANUSCRIPT 11 53
58
92
-29
-30
-17
-60
63
-106
-105
-65
0
-69
33
-56
-20
83
66
88
37
-105
-57
-101
84
137
97
-177
-16
-168
-93
-104
-119
12 14 15
17 18 2
-223 73
-111
SC
13
RI PT
16
-229
M AN U
Table A.9. Mean (in ms) of all conditions for cathodal stimulation of the left AG from
Baseline Action 1
Baseline Tone mn2
Operant Action n3
Operant Tone mn4
55
83
59
78
57
99
45
86
87
120
59
87
66
76
65
80
56
60
58
64
89
109
154
150
56
58
67
73
159
130
84
87
AC C
Participant
EP
TE D
subjects in Experiment 2. Shaded rows are data from excluded subjects.
1 3 4 5 6 7 8 9
ACCEPTED MANUSCRIPT 10 74
64
76
84
63
93
54
73
68
73
66
64
66
65
83
66
65
63
77
57
103
94
60
53
55
35
59
60
44
92
141
155
407
193
11 12 14
16 17 18
84 46
193
270 147
M AN U
13
95
SC
2
RI PT
15
Table A.10. Standard deviation of judgement error across trials (in ms) of all conditions for cathodal stimulation of the left AG from subjects in Experiment 2. Shaded rows are data from
AC C
EP
TE D
excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
-108
-9
-83
-73
-150
-117
-136
-258
44
-21
25
-101
-60
-100
-77
-111
-85
87
-8
6
-94
-51
-116
-144
-29
-2
-38
-176
1 3 4 5 6 7 8
ACCEPTED MANUSCRIPT 9 -59
-1
-14
-34
-55
-20
-16
-175
43
36
-3
-55
-134
-21
-106
-21
-108
-113
-38
-36
-66
-56
-11
24
88
67
-88
0
-116
85
124
0
322
10 11 12
15 16 17
SC
18 2 -92
-117
-5 5
-98
49
70
98
180
M AN U
13
RI PT
14
-100
-252
Table A.11. Mean (in ms) of all conditions for sham from subjects in Experiment 2. Shaded
Baseline Action 1
Baseline Tone mn2
Operant Action n3
Operant Tone mn4
57
76
42
55
81
82
68
95
120
68
64
80
54
84
62
71
49
86
62
89
72
82
78
92
56
67
53
80
AC C
Participant
EP
TE D
rows are excluded subjects.
1 3 4 5 6 7 8
ACCEPTED MANUSCRIPT 9 70
131
81
80
66
239
91
79
69
99
62
96
67
54
58
66
64
60
76
72
62
92
71
78
70
103
46
94
46
45
48
96
46
61
246
183
10 11 12
16 17
41
56
123
259
M AN U
18
126
2 13 266
SC
15
RI PT
14
120
129
Table A.12. Standard deviation of judgement error across trials (in ms) of all conditions for
AC C
EP
TE D
sham from subjects in Experiment 2. Shaded rows are excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-9
-43
13
-2
-22
27
36
-66
-22
28
-96
-47
-200
-186
-50
6
61
-19
-11
-13
-14
-94
-63
-14
-30
6
48
62
-134
-106
-50
-146
-60
-37
-13
30
38
23
45
36
-71
-33
89
37
66
-22
-52
77
-50
-138
-89
-115
-130
-127
-25
-31
-86
-86
-60
-28
-37
-39
-75
-118
-85
2 3
7 8 9
12
-48
-10
-18
-72
-87
-52
-5
-9
-65
-65
-20
-39
-40
-2
-4
-26
-127
-143
-37
-29
-59
-2
-100
-85
-92
-108
-126
-61
-44
-22
-87
-79
-60
16
-32
-30
-3
-12
48
18
-30
-45
35
-19
-4
-50
87
-22
166
-27
-16
-19
72
51
11
-14
173
110
13 14 15
17 -108 1 -51 6
AC C
18
EP
-5
TE D
16
10
M AN U
11
SC
5
RI PT
4
Table A.13. Mean (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action Left 1
Baseline Action Right 1
Baseline Tone2
Operant Action Leftn3
Operant Action Rightn3
Operant Tone Leftmn4
Operant Tone Rightn4
66
83
94
60
73
46
47
86
126
111
74
119
95
110
106
122
87
77
61
82
62
66
81
85
99
86
146
79
118
78
83
109
85
125
128
84
70
123
77
124
144
89
102
80
84
67
73
178
136
78
87
106
88
71
79
115
78
51
49
75
50
83
62
74
45
94
60
61
84
61
57
50
87
60
48
70
53
137
78
102
64
88
108
91
56
101
57
51
54
77
85
2 3
7 8 9
12 13 14 15
52 1 33 6 65 10 111 18
54
63
64
90
77
42
76
114
185
88
83
243
108
76
70
131
115
116
132
114
298
130
126
232
213
219
64
234
115
AC C
110
51
EP
17
TE D
16
M AN U
11
SC
5
RI PT
4
Table A.14. Standard deviation of judgement error across trials (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-10
-33
20
-33
20
20
29
-39
-49
10
-24
-7
-151
-155
-30
-11
98
-17
-57
-48
-27
-42
-142
-93
-28
-76
85
70
63
38
-6
5
-1
-155
-139
-73
-69
12
9
3
9
-80
-79
2
4
-32
-33
-8
-62
-85
-79
-81
-22
-49
26
-17
-20
108
102
-85
-8
-91
-157
-185
-100
-129
-30
-18
-40
-107
-103
-27
-36
-83
-98
-88
-25
10
4
-10
-11
-29
-141
-141
-64
-99
5 7
9
13 14
16 -67 17 -95 1
10
-41
-27
-42
-38
-141
-154
-105
-39
-78
-67
-146
-131
-54
34
-3
-46
24
31
EP
-77 6
TE D
15
M AN U
11
SC
8
12
RI PT
3
56
65
-14
-39
5
26
10
-78
-58
44
67
13
31
2
0
74
12
67
-85
-210
AC C
18
40
Table A.15. Mean (in ms) of all conditions for anodal stimulation of right AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT Participant
Baseline Action Left 1
Baseline Action Right 1
Baseline Tone2
Operant Action Leftn3
Operant Action Rightn3
Operant Tone Leftmn4
Operant Tone Rightn4
84
70
62
33
75
73
60
62
100
90
83
86
87
63
69
56
84
58
63
93
64
36
66
94
44
97
61
58
87
113
58
109
85
103
105
111
101
139
91
151
99
98
73
105
59
67
106
115
75
57
79
95
79
59
85
102
43
63
44
44
47
42
63
60
48
2 3
5 7
9
12
46
44
70
65
55
84
39
67
63
57
77
73
46
52
64
99
81
47
72
60
58
43
55
42
53
47
42
59
63
35
137
91
38
92
275
58
52
55
269
81
113
64
181
205
96
67
78
230
258
117
178
106
66
109
97
235
135
274
13 14 15
17 45 1 39 6
AC C
18
EP
116
TE D
16
10
M AN U
11
SC
8
RI PT
4
Table A.16. Standard deviation of judgement error across trials (in ms) of all conditions for anodal stimulation of right AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-2
-8
15
-21
-26
23
37
-80
-36
-26
-92
-24
2
AC C
EP
TE D
M AN U
SC
RI PT
3 -274
-283
ACCEPTED MANUSCRIPT 4 -5
-46
103
-44
-43
-24
15
-102
-76
-24
-72
-64
-100
-99
-97
-77
-13
-56
42
-13
-20
-10
20
40
-79
-82
-45
-37
-39
-2
20
40
31
26
20
-64
-110
-56
-103
-92
-20
-50
-52
-43
-29
-28
-29
-76
-86
-59
-40
-3
62
48
-48
-60
-7
7
-9
-20
-17
1
12
-72
-92
-36
-109
-96
-60
-60
-92
-85
-18
-70
-49
-154
-170
-72
-74
-56
-48
-67
-136
-87
-55
-50
21
-70
-6
29
10
19
30
22
27
-2
-86
-47
5 7 8
RI PT
9 11 12
14
16 17 1 6
98 18 -37
TE D
10
M AN U
15
SC
13
103
40
-15
18
58
38
57
-5
68
100
48
29
EP
Table A.17. Mean (in ms) of all conditions for sham from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded
AC C
subjects.
Participant
Baseline Action Left 1
Baseline Action Right 1
Baseline Tone
Operant Action Leftn3
Operant Action Rightn3
Operant Tone Leftmn4
Operant Tone Rightn2
54
76
80
92
63
74
46
115
124
152
237
113
86
63
2 3
ACCEPTED MANUSCRIPT 4 86
92
95
67
67
79
96
40
41
87
53
48
53
106
115
115
92
57
150
89
85
82
92
122
114
88
85
81
62
96
98
93
89
74
184
49
81
92
83
83
63
122
74
64
36
58
76
53
61
68
66
76
64
68
47
56
77
43
87
71
72
75
49
61
46
70
79
57
66
64
58
54
48
62
32
72
72
171
270
5 7 8
12 13 14
16
83
61
63
113
72
183
27
51
87
68
56
205
51
54
74
173
54
69
126
119
209
98
107
83
176
181
98
156
133
90
223
166
17 1 6
80
TE D
10 18
M AN U
15
SC
11
RI PT
9
EP
Table A.18. Standard deviation of judgement error across trials (in ms) of all conditions for sham from subjects in Experiment 3. Values are shown separately for left and right hand
AC C
actions. Shaded rows are data from excluded subjects.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
TE D
Fig. A.1. Scores of the individual participants on the first canonical variate from a discriminant analysis of Experiment 1. Each dot indicates one participant, and a fitted Gaussian frequency distribution is also shown. The raw coefficients of the canonical variate
EP
were 0.015 for action binding, and 0.02 for tone binding. Both parietal and frontal stimulation
AC C
conditions were significantly different from sham.
ACCEPTED MANUSCRIPT Baseline Action
Baseline Tone
Operant Action
Operant Tone
1
-25
-5
-15
-42
3
-63
51
-38
21
4
14
153
-43
61
B.1.
5
-162
-34
-84
152
Med
6
45
143
104
36
ian
7
-48
47
-20
8
27
71
76
9
-83
23
11
-43
-54
12
89
13
-87
14
-84
15
-14
16
-6
17
7
18
SC
RI PT
Participant
29
Tabl e
(in
22
ms)
-52
-105
-147
of
M AN U
-86
149
87
all
17
-67
-202
con
5
-62
-134
-12
-21
-86
19
13
-86
-96
-45
-194
-96
-33
-53
-99
ano
2
22
21
130
-16
dal
10
-75
-87
-59
-233
stim
EP
TE D
32
ditio ns for
AC C
ulation of left dLPFC from subjects in Experiment 1. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ACCEPTED MANUSCRIPT 1 Participant
31 Baseline Action
0 Baseline Tone
101 Operant Action
-50 Operant Tone
-59
6
-66
-19
4
-10
92
16
-7
5
-10
-4
-52
88
6
-48
62
50
44
7
-126
-13
-116
-21
8
16
86
95
106
9
-53
32
-95
-28
11
-89
-35
-99
-155
12
112
24
209
48
13
-91
-38
-82
-244
14
-73
-13
-74
-163
15
-60
-98
-35
-32
16
12
76
21
-66
17
0
-99
-60
-135
2
SC
M AN U
-100
-23
-109
-89
149
48
106
28
-101
-65
-27
-120
EP
10
TE D
18
RI PT
3
AC C
Table B.2. Median (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 1. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT 1 Participant 3
-26 Baseline Action -59
-31 Baseline Tone 36
28 Operant Action 234
-168 Operant Tone 10
4
15
193
-7
-174
5
-21
-31
-46
74
6
43
42
101
-41
7
-22
-24
-56
8
12
82
118
9
-104
11
-68
11
-83
-57
-52
12
9
2
13
-190
-57
14
-77
15
-31
16
-15
17
-55
18
SC
RI PT
Tabl
-149
e B.3. Med
31
ian
-124
(in
-212
ms) of
-11
-49
-207
18
-46
-79
-15
-36
-52
ns
65
30
-90
for
-77
-64
-69
sha
-100
12
-77
-113
m
2
18
35
51
21
10
-53
17
-32
-119
TE D
M AN U
178
all con ditio
fro m subj
EP
ects in Experiment 1. Shaded rows are data from excluded subjects.
AC C
Action binding: No significant effect of stimulation (F(2, 30)=3.11, p=0.06). Tone binding: Significant effect of stimulation (F(2, 30)=3.41, p=0.04). Tone binding was significantly reduced by anodal stimulation of the putative left AG compared to sham (t(15)=2.40, p=0.03).
ACCEPTED MANUSCRIPT 1 Participant 3 1 4
-63 Baseline Action -82 -74 112
-66 Baseline Tone -67 -6 92
-77 Operant Action -101
-102 Operant Tone -331
-50 70
-230 76
5 -63
-118
-98
-193
-39
34
40
4
-162
-90
-125
-171
-31
-37
-26
-173
-94
-53
-15
-15
-13
17
88
42
-2
44
-69
18
7 8 9
11
37
-2
-90
-33
-113
-6
-58
-46
-62
19
71
103
-4
-35
-66
-154
104
153
63
99
93
338
122
-143
-152
-39
-14
-269
14 -91 15 -56 16 58 17 -75
EP
13
TE D
18
M AN U
12
2
SC
10
RI PT
6
Table B.4. Median (in ms) of all conditions for anodal stimulation of the left AG from
AC C
subjects in Experiment 2. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT 3 -96
-51
-141
-354
67
6
43
57
-40
-122
-91
-196
-12
105
-29
-19
-135
-122
-184
-136
-53
-62
-55
-79
-5
-3
-8
34
15
59
61
81
-33
-19
-109
-90
4 5 6
8 9 10
12
15 -58 16 83 17 -110 18 85 2 13
-123
-89
-41 72
-72
7
34
-77
-19
85
78
25
-56
-104
-224
130
92
78
-23
-157
-163
-87
-238
TE D
-168
-3
-55
M AN U
14
-139
SC
11
RI PT
7
-95
EP
Table B.5. Median (in ms) of all conditions for cathodal stimulation of the left AG from
AC C
subjects in Experiment 2. Shaded rows are data from excluded subjects.
Participant
Baseline Action
Baseline Tone
Operant Action
Operant Tone
ACCEPTED MANUSCRIPT 1 -108
-16
-83
-69
-153
-120
-131
-247
34
-30
32
-123
-72
-110
-75
-129
-85
89
-6
15
-102
-52
-128
-32
-15
-26
-61
3
-2
-74
16
38
26
3 4 5
7 8
10
12 -151 14 -108 15 -75 16 20 17 18 80 2 28 13
-33
-160
3
-45
-107
-11
-106
-51
-23
-52
-2
-4
68
66
-10
-7
-104
-87
114
44
77
285
82
190
-112
-95
-247
EP
-119
-175
-35
TE D
-85
-145
-33
M AN U
11
SC
9
RI PT
6
Table B.6. Median (in ms) of all conditions for sham from subjects in Experiment 2. Shaded
AC C
rows are excluded subjects.
Action binding: No significant effect of stimulation (F(2, 30)=2.70, p=0.08). Tone binding: No significant effect of stimulation (F(2, 30)=0.48, p=0.62).
ACCEPTED MANUSCRIPT Participant
Baseline Action Left
Baseline Action Right
Baseline Tone
Operant Action Left
Operant Action Right
Operant Tone Left
Operant Tone Right
-20
-37
11
-3
-21
17
42
-64
-31
31
-96
-43
-192
-190
-50
-19
59
-23
-25
0
-9
-93
-67
-6
-62
-20
9
64
-124
-105
-70
-133
-56
-38
-22
35
42
4
38
10
-97
-37
83
17
66
-21
-51
62
-15
-142
-82
-123
-116
-118
-24
-18
-96
-86
-42
-32
-20
-44
-101
-106
-74
2 3
7 8 9
12
-50
-13
-25
-73
-79
-56
5
-8
-63
-60
-10
-46
-45
-17
5
-14
-102
-113
-57
-19
-69
-1
-82
-77
-82
-96
-125
-68
-68
-28
-98
-84
-62
12
-47
-65
2
-23
-18
10
-46
-62
35
-12
-29
-64
60
-34
147
-38
-13
-40
7
50
-37
-36
140
138
13 14 15
17 -99 1 -48 6
AC C
18
EP
-2
TE D
16
10
M AN U
11
SC
5
RI PT
4
Table B.7. Median (in ms) of all conditions for anodal stimulation of left AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
Baseline Operant Operant Operant Operant Tone Action Action Tone Left Tone Baseline Operant Operant Operant Operant Left Right Right Tone Action Action Tone Left Tone Left Right Right -38
37
-3
31
19
-27
-6
-143
-160
93
-24
-23
-8
-2
-44
-39
-62
-102
-73
-39
-95
-95
-20
-50
93
23
38
-15
-35
-12
125
107
-109
-17
-80
-153
-188
-101
-115
EP
AC C
SC
RI PT
14
M AN U
Baseline Baseline Action Action Participant Baseline Baseline Left Right Action Action Left Right 2 -33 -28 3 -40 -41 4 -35 -3 5 -49 -60 7 -145 -111 8 88 119 9 -31 -6 11 -157 -145 12 -70 -62 13 5 1 14 4 8 15 -82 -66 16 -73 -35 17 -89 -106 1 -79 -68 6 52 8 10 -88 -44 18 41 -11
-28
-24
-51
-112
-87
-24
-36
-68
-89
-93
-34
4
-8
-19
-9
-31
-142
-137
-57
-93
-29
-41
-43
-140
-153
-38
-69
-71
-161
-163
26
-67
-60
19
41
-6
-20
8
25
-56
50
-3
-25
51
11
61
20
3
-95
-187
TE D
Participant
Table B.8. Median (in ms) of all conditions for anodal stimulation of right AG from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded subjects.
ACCEPTED MANUSCRIPT
7 8 9 11 12 13 14 15 16 17 1 6 10 18
-20
-16
39
36
-111
-53
-59
-25
-31
-276
-278
-6
-30
97
-51
-38
-6
-27
-92
-76
-17
-71
-69
-103
-108
-94
-104
-36
-59
18
-17
-31
-12
5
31
-88
-63
-36
-41
-48
-10
23
48
21
32
-27
-71
-98
-64
-101
-94
-14
-71
-58
-46
-25
-14
-49
-76
-78
-44
-34
-2
72
53
-38
-74
-3
15
-5
-22
-29
-4
14
-76
-79
-92
-84
-74
-88
-61
-47
-1
1
55 -14
RI PT
5
6
SC
4
15
M AN U
3
-14
-29
-121
-94
-68
-54
-20
-61
-66
-135
-160
-52
-42
-71
-144
-110
24
-33
3
21
10
-13
34
3
-97
-41
58
16
-38
2
19
28
44
-3
33
105
76
64
TE D
2
EP
Table B.9. Median (in ms) of all conditions for sham from subjects in Experiment 3. Values are shown separately for left and right hand actions. Shaded rows are data from excluded
AC C
subjects.
Action binding: No significant main effect of stimulation (F(2, 26)=0.14, p=0.87), no significant main effect of acting hand (F(1, 13)=1.40, p=0.26), no significant interaction (F(2, 26)=0.75, p=0.48).
Tone binding: Significant main effect of stimulation (F(2, 26)=7.24, p<0.01). Tone binding was significantly reduced by anodal stimulation of the putative left AG relative to both sham (t(13)=2.67, p=0.02) and anodal stimulation of putative right AG (t(13)=3.19, p=0.01). No significant main effect of acting hand (F(1, 13)=0.04, p=0.85), no significant interaction (F(2, 26)=1.57, p=0.23).