Exp Brain Res (2006) 175: 353–362 DOI 10.1007/s00221-006-0557-9

RE SE AR CH AR TI C LE

Chia-Chin Tsai · Wen-Jui Kuo · Jung-Tai Jing Daisy L. Hung · Ovid J.-L. Tzeng

A common coding framework in self–other interaction: evidence from joint action task

Received: 29 November 2005 / Accepted: 10 May 2006 / Published online: 24 June 2006 © Springer-Verlag 2006

Abstract Many of our actions are inXuenced by the social context. Traditional approach attributes the inXuence of the social context to arousal state changes in a socially promotive way. The ideomotor approach, which postulates common coding between perceived events and intended actions, uses a conceptual scheme of ideomotor compatibility to explain self–other interaction. In this study, we recorded reaction times (RTs) and eventrelated potentials in a Go/NoGo task with stimulus– response (S–R) compatibility arrangement to examine how the social context aVects self–other interaction. Although the social facilitation theory predicted that RTs would be faster when acting together with audience rather than acting alone, the ideomotor theory predicted S–R compatibility eVects only for the joint condition. The results revealed S–R compatibility on the RTs, lateralized readiness potential of the Go trials, and P3 of the NoGo trials in the joint condition, which were in line with the predictions of the ideomotor theory. Owing to the anticipation of other’s actions, self and other’s actions are internally and unintentionally coded at the representational level and their functional equivalency can be realized through a common coding framework between perception and action systems. Social facilitation theory was not supported, because we found no signiWcant data diVerences depending on the setting.

C.-C. Tsai · W.-J. Kuo · J.-T. Jing · D. L. Hung · O. J.-L. Tzeng (&) Laboratory for Cognitive Neuroscience, National Yang-Ming University, Taipei, Taiwan E-mail: [email protected] Tel.: +886-2-28267242 Fax: +886-2-28204903 C.-C. Tsai · J.-T. Jing · D. L. Hung Institute of Neuroscience, National Yang-Ming University, Taipei, Taiwan O. J.-L. Tzeng Academic Sinica, Taipei, Taiwan

Keywords Self–other interaction · Joint action · Ideomotor theory · Social facilitation · Event-related potentials

Introduction Many of our actions and decisions are inXuenced by social contexts, and even caved in social pressure (Asch 1956). Neural mechanisms of this social interaction have been related to the mirror neuron system, which suggested that our action system, which specializes in detecting and matching other’s goal-directed actions with our own repertoire, can be modulated by observing other’s actions (Gallese et al. 1996; Rizzolatti et al. 1996; Nishitani and Hari 2000; Grezes et al. 2003). However, being equipped with such a system only may not be suYcient to coordinate or cooperate with others (MeltzoV and Decety 2003). The reason is that evidence previously demonstrated was derived from experimental environments in isolation, instead of interactive settings. In this study, we employed a newly invented motor cognition paradigm, the joint action paradigm, to investigate self–other interaction with diVerent social contexts. By using this paradigm, processes of recognition, anticipation, prediction, and interpretation of others’ action realized by a direct link of perception and execution can be practically examined at a representational level (Sebanz et al. 2003, 2005, 2006, in press). Research on self–other interaction, especially for joint action, leads to diVerent conclusions regarding its underlying mechanisms. One approach emphasizes the social facilitation eVect (Zajonc 1965), which suggests that coacting with or the presence of others will inXuence task performance in a general way. Subjects performed better in simple tasks and worse in diYcult tasks in cooperation and being observed contexts than in an individual context. This eVect was attributed to the change of arousal states, social comparison, and cognitive distraction (Guerin 1993). Apart from this general description, it is not clear how self and other’s actions interact at the

354

representational level (Sebanz et al. 2003). Another approach, i.e., ideomotor theory, emphasizes a common coding or shared representations between the perceived events and planned actions and the role of internal (volitional) causes of actions (Prinz 1997; Hommel et al. 2001). Its central notion is that “the perception of an event that shares features with an event that has been learnt to accompany, or follow from, one’s own action, will tend to induce that action” (Greenwald 1970a, b, 1972). By this approach, interaction is taking place at a higher or cognitive level. Given methodological advancements provided by the motor cognition paradigm, characteristics in processing and theory of this interaction can be further addressed. As compared to a traditional case of joint action with language communication format (Clark 1996), the current motor paradigm referred to acting together in a social context in a more general sense (Sebanz et al. 2006). The motor cognition paradigm is to combine a target-detection task with stimulus–response (S–R) compatibility under diVerent social contexts, a group, or an individual social context (Sebanz et al. 2003, in press). In the detection task, one of the two targets (circles in red or green) was presented with equal possibility in three spatial positions (left, middle, and right) on a computer screen. Two response alternatives, a left and a right button press, were designated for red and green stimuli, respectively. Therefore, two types of experimental trials can be deWned by the spatial relationship of target and response button position: compatible trials, in which the spatial relationship between stimulus and response is correspondent (i.e., when a red circle appears on the left side and with a key response on the left side is performed), and incompatible trials, in which the spatial relationship between stimulus and response is non-correspondent (i.e., when a red circle appears at the right side and with a key response on the left side is performed). In traditional cases, only a single subject performs the task at a time with both their left and right hands for responding to red and green targets, and typically, reaction times (RTs) on incompatible trials are considerably slower than RTs on compatible trials. This has been dubbed spatial S–R compatibility or Simon eVect. With modiWcation of this S–R compatibility task, designation of diVerent social contexts was included in a motor cognition paradigm. In this study, three social contexts were combined with a motor cognition paradigm to investigate self– other interaction, especially to test the social facilitation and the ideomotor accounts. In a joint context, two subjects were paired and instructed to respond to a red and a green target complementarily. Because the subjects responded to one target at a time, it became a Go/NoGo task for each subject. Although the left-seated subject responded to red circles, another subject on the right seat responded to green circles. In an individual context, subjects performed the same Go/NoGo task individually. In a being observed context, two subjects were paired as well but only one subject performed the task with

another subject served as an observer. In the experiment, both behavioral and electrophysiological data were simultaneously recorded for the subjects. Event-related potential (ERP) was important because it helps to investigate the formation of self–other interaction because it provides information about the NoGo trials where no overt response can be recorded at behavioral level. Two ERP components, the N2 and the P3, are of direct relevance to the current task. The N2 is a negative ERP component that occurs approximately 200 ms after stimulus onset, is maximal at the frontal sites (Falkenstein et al. 1999; Bruin and Wijers 2002), and is typically larger on the NoGo than on the Go trials (PfeVerbaum et al. 1985; Falkenstein et al. 1995, 1999; Kopp et al. 1996), reXecting inhibition eVects (Kok 1986; Falkenstein et al. 1999; Lavric et al. 2004). The P3 is generally reported to have a centro-parietal maximum for the Go trials and a fronto-centrally maximum for the NoGo trials, with latency ranging from 300 to 500 ms (PfeVerbaum et al. 1985; Falkenstein et al. 1995; Bruin and Wijers 2002). The P3 is assumed to reXect the processing of stimulus evaluation (Kok 2001). Previous studies indicated a consensus that the Go P3 occurs after the stimulus has been evaluated, and can occur even on the basis of partial information. In contrast, the NoGo P3 is suggested to reXect an inhibition mechanism for action control (Falkenstein et al. 2002). Although there are some controversies regarding the functional attributes of the NoGo N2 and P3 components in inhibition (Bruin et al. 2001), evidence suggests that these two components are related to diVerent aspects of inhibitory processes (Falkenstein et al. 1999, 2002). In a task similar to S–R compatibility task, in addition to N2 and P3, there is another ERP component of interest, i.e., lateralized readiness potential (LRP). LRP directly relates to the preparation of motor responses in a real-time scale (Coles 1989) and its activation changes can provide information of response selection and preparation for comparison, especially for compatibility diVerence under diVerent social contexts. In a study with similar arrangement, Sebanz et al. exploited a Go/NoGo task with individual and joint contexts. They demonstrated a self–other interaction by revealing a compatibility eVect in the joint context but not in the individual context (Sebanz et al. 2003, in press). The joint compatibility eVect, an interaction of compatibility and social context, suggested that the subjects took others’ action into consideration, i.e., action representation of self and others’ was considered as functionally equivalents. As compared to previous studies (Sebanz et al. 2003, in press), our study has extended issues of current interest in several ways. First, we can directly test social facilitation against ideomotor account for social interaction with the introduction of the being observed condition. Second, electrophysiological indicators, i.e., N2 and P3 of the NoGo trial, help clarify how social context can inXuence response inhibition. For example, in the joint setting, the NoGo trials for one of the paired subjects are the Go trials for another. Subjects might

355

generate two kinds of action representation, i.e., action inhibition for themselves and action anticipation for others, and which might regulate N2 and P3 components. Moreover, the third, whether social symbols are necessary for the joint compatibility eVects can be tested using nonsocial stimuli as the response cues. In contrast to Sebanz et al. (2003) who used social symbols, i.e., pictures of a pointing Wnger wearing a color ring, we used color circles as experimental stimuli to explore this question. Taken together, current interest and arrangements led to several predictions. First, RTs of three social contexts would diVer. If the social facilitation account holds, RTs in the joint and being observed conditions would be similar and faster than that of the individual condition. In contrast, if the ideomotor theory holds, we should observe compatibility eVects of RT only in the joint condition. Second, electrophysiological components of N2 and P3 in the NoGo trials would behave diVerently in diVerent social contexts. If the social facilitation account holds, because it predicts that in a simple task people would perform better when there is audience, behaving proWle of N2 which relates to monitoring or inhibition process might diVer among the three social contexts. However, if the ideomotor account holds, NoGo P3 in joint action condition might diVer from the other two conditions because action anticipation of other’s is engendered only in the joint context. Third, as onset of P3 component in the Go trials has been regarded as emergence of evaluation processing, the Go P3 latency in the joint and being observed conditions should be faster than that in the individual condition if the social facilitation account holds. Moreover, as LRP component is sensitive to brief preparation for manual response transformation (Kutas and Donchin, 1980), for ideomotor approach, which suggests that conXict might occur after the stage of stimulus evaluation, the inXuence of social contexts would aVect LRP rather than Go P3. Social contexts would be possible to modulate the response activation, especially aVected the manual transformation on the non-corresponding side (incompatible trials) in the joint condition if the ideomotor account holds. Fig. 1 Illustration of the formation of three social contexts in our experimental setting: left joint Go/NoGo condition; middle being observed Go/NoGo condition; right individual Go/ NoGo condition

Materials and methods Subjects Twenty-six right-handed volunteers, all of whom were students from National Yang-Ming University, participated in this experiment (19–24 years old). Each subject was paid for $14.28 for participation. Four data sets had to be discarded because of serious eye blink artifacts. Twenty-two subjects’ data (ten males and 12 females) were analyzed. Experimental setting and stimuli Figure 1 displays the three social contexts and task settings. Subjects performed a Go/NoGo task. (1) In the joint condition, paired subjects sat side by side and performed the task in a complementary way. Although the left-seated subject responded to red circles, another subject on the right seat responded to green circles. (2) In the individual condition, subjects performed the task alone. An empty chair remained beside for each participant. (3) In the being observed condition, two subjects were paired but only one subject performed the task with another subject serving as an observer. The two paired subjects were instructed together and that the observer was told to observe and learn the procedure and would covertly count the errors committed. By doing this, we expected to make an audience eVect for the subject performing the task. The order of conditions was counterbalanced. A rectangle (9 £ 3.5 cm2 in width and height) with three discs horizontally arranged inside (1 cm in radius and 0.5 cm between the discs) was presented centrally on a PC monitor. Targets were either red or green dots presented on one of the three discs at a time. The Wxation and target extended approximately 1.17° and 3° in height and width. Procedure and design Each trial started with a cue for 1,500 ms for eye blinking. In the following, a Wxation was presented for 500 ms, and

356

then a target was presented either on the right, central, or left disc for 100 ms. There was a time period of about 1,500 ms for responding before the next trial started. Subjects were instructed to respond to red or green target by pressing the “shift” keys on the left or right side of the keyboard with equal emphasis on the response speed and accuracy. In consideration of S–R compatibility, one-third of the trials were compatible arrangement, and incompatible for another one-third of the trials. In all conditions, subjects completed four blocks of 48 trials (including 24 Go and 24 NoGo trials) and switched seats before the third block. ERP recording EEG data were recorded from 64 scalp electrodes, and vertical and horizontal EOGs were recorded for eye movements. All channels were referenced to the linked mastoids. EEG and EOG were recorded with a sampling rate of 500 Hz. EEG epochs set at the range of ¡100 to 800 ms. Data were digitally Wltered with a band-pass Wlter (1–30 Hz, 12 dB/oct). Only artifact-free trials were averaged to create ERP. Trials containing eye movement artifact, A/D saturation, or with a baseline drift exceeding 60
V in any channel, were excluded. Stimulus-locked ERPs of electrodes C3 and C4 were used for LRP calculation. For response on the left side, ERP of C3 was subtracted from that of C4; for response on the right side, ERP of C4 was subtracted from that of C3. After subtraction, the results were averaged and digitally Wltered (lowpass cut-oV frequency of 12 Hz). LRPs of compatible and incompatible conditions were calculated separately.

being observed, and individual) and compatibility (compatible and incompatible) factors was conducted. SigniWcance was found for compatibility (F(1,21) = 22.99, p < .05) and interaction (F(2,42) = 15.52, p < .05), but not for social context (F(2,42) = 0.391, p > .05). As our prediction, this compatibility was only signiWcant in the joint condition. Subjects responded slower to incompatible than to compatible trials (joint: F(1,63) = 53.538, p < .05; being observed: F(1,63) = 1.274, p > .05; individual: F(1,63) = 2.89, p > .05). In order to diVerentiate between the facilitation and inhibition of S–R compatibility, one procedure was further exploited. In the procedure, RT performance of the trials with targets appearing on the central disc was taken as a baseline to be subtracted from the RTs of compatible and incompatible trials. The results were divided by the baseline afterward for indication of facilitation and inhibition (Spieler et al. 1996). For the convenience of calculation, we modiWed this index by multiplying it with 104. Table 1 also reports the indication of facilitation and inhibition in the three social contexts. A repeated-measures two-way ANOVA with the factors social context and the facilitation/inhibition showed that there was no signiWcant main eVect for the facilitation/inhibition (F(1,21) = 2.69, p > .05), but there was a signiWcant eVect for social context (F(2,42) = 15.512, p < .05). The interaction was signiWcant (F(2,42) = 10.897, p < .05). Post hoc analysis revealed a signiWcant diVerence in inhibition and showed a maximum in the joint condition rather than in the other two conditions. Electrophysiological data

Results Repeated-measures ANOVA was used to analyze the behavioral and ERP data. The Greenhouse–Geisser adjusted p values were reported when necessary, but original degrees of freedom are given. Behavioral results Table 1 presents RT data of the Go trial in three social contexts. A two-way ANOVA with social context (joint,

Averaged ERPs for compatibility on the NoGo and Go trials in three social contexts at three electrodes are shown in Figs. 2, 3, and 4. The P3 component was quantiWed by measuring the mean amplitude and peak latency in the range 310–360 ms for the Go trial and 350–400 ms for the NoGo trial (see Tables 3 and 4). The NoGo N2 was quantiWed by measuring mean amplitudes from 220 to 270 ms for the analysis of social context and compatibility (see Table 2). Time windows for compatibility eVect on the Go and NoGo trials and the electrodes of interest, i.e., Fz, Cz, and Pz, were selected in line with the

Table 1 Mean RT (Go trials) on compatible and incompatible trials in three social contexts RTs (ms)

Joint COM

Mean RT1 Mean RT2 Compatibility EVect size

434.4 (31.9)

Being observed IMC

449.6 (33.4) 442.0 (33.5) 15.2 Facilitation Interference 97.785 248.453

COM 437.5 (36.1)

Individual IMC

441.0 (33.0) 439.3 (34.6) 3.5 Facilitation Interference 66.995 17.892

COM

IMC

437.5 (43.7)

439.9 (41.0) 438.7 (42.4) 2.4 Facilitation Interference 37.602 20.046

COM compatible trial; IMC incompatible trial. In parentheses are the standard deviations of the mean RT on Go trials. The compatibility eVect = RTincompatible trial ¡ RTcompatible trial. EVect size (corrected) for facilitation = [(RTcompatible ¡ RT neutral)/RTneutral] £ 104. EVect size (corrected) for inhibition = [(RTincompatible ¡ RTneutral)/RTneutral] £ 104. The RTneutral: 438.7 ms (joint); 439.1 ms (individual); 440.4 ms (being observed)

357

previous studies with typical Go/NoGo paradigms (PfeVerbaum et al. 1985; Falkenstein et al. 1995, 1999). For LRP analysis, it was quantiWed by measuring mean amplitudes from 200 to 300 ms for the Go trials and 100 to 200 ms for the NoGo trials, locked by the onset time of stimulus (see Table 5). NoGo N2 component Table 2 shows the NoGo N2 amplitude (see also Fig. 2), and a 3 £ 2 £ 3 ANOVA of amplitude with factors of social context, compatibility, and electrode was conducted. For the electrode factor, it had a signiWcant eVect (F(2,42) = 8.328, p < .05), and showed an anterior minimum. Compatibility eVect was signiWcant (F(1,21) = 7.124, p < .05). No interaction was observed. NoGo P3 component Figure 2 also displays the grand averages for the NoGo P3 (see also Table 3). A 3 £ 2 £ 3 ANOVA of amplitude with factors of social context, compatibility, and electrode was conducted. First, for the electrode factor, it had a signiWcant eVect (F(2,42) = 31.973, p < .05) and

showed an anterior-central maximum (Fz: 8.281
V, Cz: 8.811
V, Pz: 5.075
V). This was in line with the wellestablished Wnding of an anteriorization of topography of the scalp sites on the NoGo trials (PfeVerbaum et al. 1985; Falkenstein et al. 1995, 1999; Fallgatter and Strik 1999; Bokura et al. 2001). Second, the main eVect of social context (F(2,42) = 7.373, p < .05) was signiWcant. Post hoc analysis showed that NoGo P3 amplitude was larger in the joint condition rather than other two conditions. Third, the three-way interaction (F(4,84) = 3.663, p < .05) and two-way interaction between social context and compatibility at Fz electrode (F(2,126) = 4.023, p < .05) were also signiWcant. At Fz electrode, a signiWcant compatibility eVect was found only in the joint condition rather than other two conditions (joint: F(1,189) = 14.934, p < .05; being observed: F(1,189) = 0.482, p > .05; individual: F(1,189) = 0.283, p > .05). In the joint condition, the NoGo P3 amplitude was larger on incompatible than on compatible trials. For the analysis of peak latency, there was only a main eVect for electrode (F(2,42) = 6.203, p < .05). Post hoc analysis showed that the latency of NoGo P3 is faster at anterior-central electrodes (Fz: 361.2 ms, Cz: 360.3 ms, Pz: 366.4 ms).

Table 2 Electrophysiological results of the compatibility eVect (compatible vs. incompatible trial) of NoGo N2 on mean amplitude (
V) and peak latencies (ms) at three electrode sites (Fz, Cz, Pz) in three social contexts Compatibility eVect of mean amplitudes of NoGo N2 (time window: 220–270 ms) Joint

Fz Cz Pz

Being observed

Individual

Compatible

Incompatible

Compatible

Incompatible

Compatible

Incompatible

1.546 (2.94) 3.081 (3.22) 2.042 (3.16)

3.009 (2.64) 4.672 (3.10) 3.064 (3.13)

2.496 (2.77) 4.307 (3.74) 3.139 (4.05)

2.801 (3.17) 4.492 (3.79) 3.126 (3.75)

2.117 (2.40) 3.670 (3.20) 2.574 (3.36)

2.367 (2.90) 4.130 (3.61) 3.186 (3.62)

In parentheses are the standard deviations of the mean on NoGo N2 Fig. 2 ERP waveforms associated with compatibility eVect in three diVerent social contexts for NoGo stimulus at three electrodes: left joint NoGo, middle being observed NoGo, right individual NoGo; upper Fz, middle Cz, lower Pz; black line compatible trial, grey line incompatible trial

358 Table 3 Electrophysiological results of the compatibility eVect (compatible vs. incompatible trial) of NoGo P3 on mean amplitude (
V) and peak latencies (ms) at three electrode sites (Fz, Cz, Pz) in three social contexts Joint Compatible

Being observed Incompatible

Compatible

Individual Incompatible

Compatible

Incompatible

7.770 (4.18) 8.188 (3.96) 4.306 (2.95)

7.991 (4.78) 8.478 (4.43) 4.603 (2.83)

Compatibility eVect of mean amplitudes of NoGo P3 (time window: 350–400 ms) Fz Cz Pz

8.336 (3.48) 9.051 (3.41) 5.820 (2.69)

9.934 (4.38) 10.335 (3.87) 5.967 (2.80)

7.686 (3.63) 8.296 (3.78) 4.945 (2.70)

7.973 (3.55) 8.517 (3.35) 4.812 (2.33)

Compatibility eVect of peak latencies of NoGo P3 (time window: 350–400 ms) Fz Cz Pz

360.2 (14.3) 359.2 (12.1) 364.6 (13.8)

361.5 (9.7) 360.6 (10.0) 366.8 (14.5)

360.9 (13.0) 358.4 (12.6) 362.1 (15.6)

363.1 (14.2) 358.1 (12.2) 364.6 (17.1)

361.5 (13.6) 363.5 (14.2) 372.5 (17.1)

359.9 (13.1) 362.3 (15.2) 368.0 (16.0)

In parentheses are the standard deviations of the mean on NoGo P3

Go P3 component Figure 3 presents the Go P3 proWles at three representative electrodes (Fz, Cz, Pz) in the three social contexts (see also Table 4). A 3 £ 2 £ 3 ANOVA of amplitude with factors of social context, compatibility, and electrode was conducted. P3 amplitudes diVered across electrodes (F(2,42) = 20.609, p < .05) and showed an anterior minimum. Interaction between social context and compatibility was signiWcant (F(2, 42) = 3.472, p < .05). For the analysis of peak latencies, the main eVect for compatibility (F(1,21) = 8.434, p < .05) and electrode (F(2,42) = 9.837, p < .05) was signiWcant. Post hoc analysis showed that Go P3 peak latency was faster at Pz than at Fz (Fz: 334.7 ms, Cz: 330.1 ms, Pz: 326.1 ms). No signiWcance was found further. LRP component for the Go and NoGo trials Figure 4 presents the LRP data in the three social contexts separately on the Go and NoGo, compatible and incompatible trials (see also Table 5). For the Go Fig. 3 ERP waveforms associated with compatibility eVect in three diVerent social contexts for Go stimulus at three electrode sites: left joint Go, middle being observed Go, right Individual Go; upper Fz, middle Cz, lower Pz; black line compatible trial, grey line incompatible trial

response of the incompatible trials, LRPs had a positive deXection Wrst and a delayed negative deXection later due to a conXict between spatial and response dimensions (Gratton et al. 1988). Thus, a one-way ANOVA with factor of social context was conducted separately for compatible and incompatible trials. For compatible trials, the eVect of social context (F(2,42) = 0.037, p > .05) was not signiWcant during the negative deXection (200– 300 ms). For incompatible trials, the negative deXection (200–300 ms) was modulated by social context (F(2,42) = 5.425, p < .05). Post hoc analysis showed a more negative potential in the joint condition rather than other two conditions. Positive deXection (125–175 ms) was not aVected by social context (F(2,42) = 0.024, p > .05). For the NoGo response, incompatible trials also demonstrated a similar positive deXection. A oneway ANOVA with factor of social context was conducted separately for compatible and incompatible trials but only the positive deXection was tested on incompatible trials. For compatible trials, the negative deXection (100–200 ms) was modulated by social context

359 Table 4 Electrophysiological results of the compatibility eVect (compatible vs. incompatible trial) of Go P3 on mean amplitude (
V) and peak latencies (ms) at three electrode sites (Fz, Cz, Pz) in three social contexts Joint Compatible

Being observed Incompatible

Compatible

Individual Incompatible

Compatible

Incompatible

4.582 (2.83) 8.342 (3.32) 6.863 (2.69)

5.432 (3.23) 9.115 (3.88) 6.828 (2.71)

5.618 (3.65) 9.441 (4.03) 6.978 (2.73)

Compatibility eVect of mean amplitudes of Go P3 (time window: 310–360 ms) Fz Cz Pz

5.720 (2.83) 8.574 (3.37) 6.964 (2.80)

6.024 (2.64) 9.133 (3.12) 7.750 (2.58)

5.471 (3.01) 8.875 (3.92) 6.932 (3.11)

Compatibility eVect of peak latencies of Go P3 (time window: 310–360 ms) Fz Cz Pz

328.6 (19.2) 326.7 (17.0) 323.3 (15.4)

334.2 (14.2) 331.6 (15.8) 328.9 (15.2)

333.7 (17.2) 330.0 (15.6) 325.0 (13.7)

337.0 (17.8) 333.6 (16.4) 328.9 (15.2)

336.2 (17.9) 327.8 (17.1) 324.6 (15.2)

338.7 (17.4) 330.9 (18.5) 326.1 (15.6)

In parentheses are the standard deviations of the mean on Go-P3

Behavioral evidence

Fig. 4 LRP waveforms associated with compatibility eVect in three diVerent social contexts for Go and NoGo stimulus: left Go trial, right NoGo trial; upper compatible trial, lower incompatible trial

(F(2,42) = 5.411, p < .05). Post hoc analysis showed a more negative potential in the joint condition rather than other two conditions. On incompatible trial, there is no social context diVerence (F(2,42) = 0.697, p > .05) for the positive reXection (100–200 ms).

Discussion By using a motor cognition paradigm with diVerent social contexts, while behavioral S–R compatibility eVect was shown only in the joint context, electrophysiological changes in response to the S–R correspondence was also found only in the joint context. Both behavioral and electrophysiological evidence beneWt the idea that the individual will represent and anticipate others’ actions when acting together. In general, the ideomotor theory but not social facilitation theory gains support from our behavioral and electrophysiological Wndings.

RT compatibility eVect was shown in the joint condition rather than in the individual or being observed condition, which indicated that acting together or alone caused diVerence in action planning, i.e., stimulus evaluation or response selection of the task. This observation lent little support to the social facilitation account because the subjects behaved similarly regardless of the being observed or individual conditions. However, one might concerns, when taking a close look at the being observed condition, there seemed to be a facilitation eVect in the being observed condition, and which might favor the social facilitation account. Even so, the social facilitation eVect is very limited as compared to the joint compatibility eVect. In contrast, the ideomotor theory gains support from current Wndings. Compatibility eVects revealed in the joint condition indicated that other’ action representation was taken into account. That is, in the incompatible trials of joint condition, subjects responded slower due to a conXict between the two competing action codes from relevant response and irrelevant stimulus dimensions. This conXict did not exist in the other two conditions (Fig. 5). Especially, the audience in the being observed condition was not enough to activate the common coding system of self–other interaction. Our data were in line with the Wndings of Sebanz et al. (2003), and which suggested that when acting together, the self–other interaction can be realized through a common coding framework between perception and action systems. The joint compatibility eVect in a Go/NoGo task suggested that self and others’ actions can be represented as the functional equivalents (Sebanz et al. 2003). From the ideomotor point of view, the compatibility eVect can be explained in terms of dimensional overlap between irrelevant stimulus and response dimensions (Kornblum et al. 1990; Kornblum and Lee 1995; Prinz 1997). In this study, although the relevant stimulus (i.e., color of the stimulus) is processed via a controlled route to activate the correct response, the irrelevant stimulus

360 Table 5 Electrophysiological results of Go/NoGo LRP on compatible and incompatible trials in three social contexts Go LRP (200–300 ms)

Compatible Incompatible

NoGo LRP (100–200 ms)

Joint

Being observed

Individual

Joint

Being observed

Individual

¡2.510 (2.21) ¡3.220 (2.07)

¡2.104 (2.26) ¡2.186 (2.10)

¡2.071 (2.16) ¡2.105 (2.17)

¡2.097 (1.70) 1.192 (1.57)

¡1.352 (1.29) 0.890 (1.67)

¡1.280 (1.11) 0.863 (1.64)

In parentheses are the standard deviations of the LRPs on Go and NoGo trials

Fig. 5 Illustration of direct link and task demand associated with three social settings and compatibility situations: upper joint Go/ NoGo condition; middle being observed Go/NoGo condition; lower individual Go/NoGo condition

dimension (i.e., the location of the stimulus) can be activated via an automatic route. Thus, a slower RT on incompatible trials is due to a conXict between the irrelevant stimulus dimension and the response dimension. Therefore, this theory also proposed that the locus of interference might occur after identiWcation of relevant dimension, and possibly up to response selection stage (Kornblum et al. 1990; Masaki et al. 2000; Valle-Inclan 1996). Electrophysiological evidence in the NoGo trials In a NoGo trial, the Wnal action plan of a subject is to inhibit his motor response. However, when a NoGo trial

in a joint context, it will serve as a Go trial for another subject as well, which will lead one to anticipate other’s action. Therefore, in a joint NoGo trial, a control for inhibiting self and anticipating other’s actions will be necessitated. As there is no behavioral response in the NoGo trials, our electrophysiological data help to clarify these processing proWles. First, we found that NoGo P3 amplitude in the joint condition varied from the other two conditions, whereas the NoGo N2 did not change among the three social contexts. As suggested by the physiological meaning of NoGo N2 and P3 discussed aforementioned (Falkenstein et al. 1999, 2002), these results together indicated that action inhibition (NoGo N2) occurred in the three social contexts equally, but action anticipation (NoGo P3) revealed only in the joint situation and caused more control processing for action monitoring. Bokura et al. (2005) also suggested that NoGo P3 rather than NoGo N2 might reXect executive function including expectation and planning by observing decreased NoGo P3 amplitude in patients with Parkinson’s disease. Additional action control in the joint condition was perfectly in justiWcation of the ideomotor theory. When perceiving a stimulus requiring an action from the coactor, it would also awaken related action representation in the self. NoGo trials, therefore, not only induced response inhibition but also action anticipation. In order to suppress the increased activation following anticipation of the other’s action, additional action control in joint condition was engendered. We might anticipate the coactor’s action representation or plan in terms of the taskspeciWc relationship he faced (Sebanz et al. 2005). Accordingly, S–R compatibility on NoGo trials was reXected in the electrophysiological response, i.e., P3 amplitude. For example, when a green stimulus appeared on the right side (incompatible NoGo trial), not only irrelevant stimulus dimension (RIGHT) and related task demand (GREEN is the NoGo response for self) were processed, but also anticipation toward other’s actions activated automatically (now she/he has to respond to GREEN stimulus with a RIGHT key). Thus, larger NoGo P3 amplitude on incompatible trials might result from the reconciliation between action anticipation and response inhibition. As noted, in a joint action the individuals need to devote more eVort than acting alone. Given that the mirror system is ego-centered and does not imply another agent in one’s own action plan (Knoblich and Jordan 2002, 2003), one needs another group-centered coordination system to represent joint action and to modulate one’s action plan.

361

Electrophysiological evidence in the Go trials In general, no signiWcant compatibility and social context eVect were found in the Go-P3 amplitude, although the compatibility eVect in joint condition seemed to be larger than other two conditions (Table 4). In a further analysis suggested by interaction between social context and compatibility, we only found nearly signiWcant compatibility eVect in the joint condition (Joint: F(1,63) = 3.378, p = 0.07). Changes of Go P3 amplitude have been related to the loading of stimulus evaluation, and which would occur prior to action selection and preparation (Kok 2001). No diVerence was found in this study suggesting that the subjects were equally loaded with stimulus evaluation for the Go trials under diVerent social contexts. This Wts with the hypothesis that the interference eVect in an S–R compatibility task occurs at the response selection stage instead of evaluation (Kornblum et al. 1990; Masaki et al. 2000). Our results were partly not consistent with the Wndings of Sebanz et al. (in press). In their Wndings, they found reduced P3 amplitude on incompatible trials at the posterior electrodes in both the group and individual conditions, and a smaller positivity for incompatible Go trials at anterior electrode in the group condition than in the individual condition. This diVerence might result from the cuing stimuli of diVerent domains, i.e., social vs. non-social symbols. Information derived from P3 latency of the Go trial is more controversial. Some studies reported that P3 latency correlates with stimulus evaluation rather than response selection and is insensitive to S–R compatibility (McCarthy and Donchin 1981; Magliero et al. 1984; Smulders et al. 1995). Some studies reported longer latency for P3 component and slower RT to the incompatible stimuli, suggesting a locus of interference at response selection (Ragot and Renault 1981; Ragot 1984). Our results were consistent with the later cases. However, the present ERP data were not enough for claiming that the locus of interference on a joint action only occurs at the stage of response selection due to the lack of rule or other manipulation in the experiment. More research is necessitated, and one of our unpublished studies does support this possibility. Electrophysiological evidence of LRP LRP is an index for motor preparation and it provides another way to investigate self–other interaction. As ideomotor theory suggests that perceiving other’s action will evoke a corresponding motor activation at representation level, modulation of social context on LRP should be possible to reXect response priming and transformation. In this study, LRP from compatible trials of NoGo response and incompatible trials of Go responses were signiWcantly modulated by the social context. In a joint context, it behaved distinctly to other contexts, and it might indicate a priming eVect of cortical response in response to other’s action. The Wndings supported the ideomotor approach. When taking P3, LRP, and RTs for

the Go trials, compatibility eVects on RTs and LRP (sensitive to the stage of response selection) but not on P3 (sensitive to the stage of stimulus evaluation) suggested that conXict in a joint action in this case might occur at the stage after the stimulus evaluation, probably at the stage of response selection. A common coding framework for action perception and execution in a joint context Our Wndings support the idea that self–other interaction can be realized through a common coding framework between perception and action systems (Hommel et al. 2001), in which one’s own and others’ actions are represented in a functional equivalent way. Three current Wndings support this claim. First, behavioral compatibility eVects in the joint condition suggested that action perception and action execution share a common coding at the representation level and thus are commensurate (Hommel et al. 2001). Second, the NoGo P3 Wndings demonstrate that action anticipation causes a speciWc demand and play a crucial role in joint action. Finally, subjects might be forced to form a relevant response set for self and others’ actions. When there is a conXict between responses for relevant and irrelevant stimulus dimensions, subjects have to inhibit responses for irrelevant dimension and select responses for relevant information dimension. Accordingly, the Simon-like interference takes place at a response selection stage, or at least after the stage of identifying stimuli of relevance. It might be the reason that no compatibility eVect in the Go P3 was found in our study. As a concluding remark, we would like to emphasize that the ideomotor approach not only provides a new and appropriate framework to understand joint action and social perception theoretically, but also allows us to examine these phenomena empirically. In this study, evidence from behavioral and electrophysiological data suggested a common representation framework for self– other interaction. Many relevant issues are in need of being explored, e.g., the role of agency (doing joint action with an agent or with a computer program). Moreover, it is also useful to explore issues of joint action with neuroimaging tools, e.g., fMRI, to clarify what the neural networks underpinning this common framework. Acknowledgments This research was conducted in the Laboratory for Cognitive Neuroscience and was partly supported by Academic Sinica, National Science Council (NSC 94-2572-H-010-002-PAE), and the Tzong Jwo Jang Educational Foundation of Taiwan. We also thank Shin-Mai Sun for her help in collecting the ERP data.

References Asch SE (1956) Studies of independence and conformity: a minority of one against a unanimous majority. Psychol Monogr 70:416 Bokura H, Yamaguchi S, Kobayashi S (2001) Electrophysiological correlates for response inhibition in a Go/NoGo task. Clin Neurophysiol 112:2224–2232

362 Bokura H, Yamaguchi S, Kobayashi S (2005) Event-related potentials for response inhibition in Parkinson’s disease. Neuropsychologia 43:967–975 Bruin KJ, Wijers AA (2002) Inhibition, response mode, and stimulus probability: a comparative event-related potential study. Clin Neurophysiol 112:1172–1182 Bruin KJ, Wijers AA, van Staveren AS (2001) Response priming in a Go/NoGo task: do we have to explain the Go/NoGo N2 eVect in terms of response activation instead of inhibition? Clin Neurophysiol 112:1660–1671 Clark H (1996) Using language. Cambridge University Press, UK Coles MG (1989) Modern mind-brain reading: psychophysiology, physiology, and cognition. Psychophysiology 26:251–269 Falkenstein M, Koshlykova NA, Kiroj VN, Hoormann J, Hohnsbein J (1995) Late ERP components in visual and auditory Go/ NoGo tasks. Electroen Clin Neurophysiol 96:36–43 Falkenstein M, Hoormann J, Hohnsbein J (1999) ERP components in Go/NoGo tasks and their relation to inhibition. Acta Psychol 101:267–291 Falkenstein M, Hoormann J, Hohnsbein J (2002) Inhibition-related ERP components: variation with modality, age and time-on task. J Psychophysiol 16:167–175 Fallgatter AJ, Strik WK (1999) The NoGo anteriorization as a neurophysiological standard index for cognitive response control. Int J Psychophysiol 32:233–238 Gallese V, Fadiga L, Fogassi L, Rizzolatti G (1996) Action recognition in the premotor cortex. Brain 119:593–609 Gratton G, Coles MG, Sirevaag EJ, Eriksen CW, Donchin E (1988) Pre- and poststimulus activation of response channels: a psychophysiological analysis. J Exp Psychol Hum 14:331–344 Greenwald AG (1970a) A choice reaction time test of ideomotor theory. J Exp Psychol 86:20–25 Greenwald AG (1970b) Sensory feedback mechanisms in performance control: with special reference to the ideomotor mechanism. Psychol Rev 77:73–99 Greenwald AG (1972) On doing two things at once: time sharing as a function of ideomotor compatibility. J Exp Psychol 94:52–57 Grezes J, Armony JL, Rowe J, Passingham RE (2003) Activations related to “mirror” and “canonical” neurons in the human brain: an fMRI study. Neuroimage 18:928–937 Guerin B (1993) Social facilitation. Cambridge University Press, Cambridge Hommel B, Musseler J, Aschersleben G, Prinz W (2001) The theory of event coding (TEC): a framework for perception and action planning. Behav Brain Sci 24:849–878 Knoblich G, Jordan JS (2002) The mirror system and joint action. In: Stamenov MI, Gallese V (eds) Mirror neurons and the evolution of brain and language. Benjamins, Amsterdam, pp 115–124 Knoblich G, Jordan JS (2003) Action coordination in groups and individuals: learning anticipatory control. J Exp Psychol Learning 29:1006–1016 Kok A (1986) EVects of degradation of visual stimulation on components of the event-related potential (ERP) in Go/NoGo reaction tasks. Biol Psychol 23:21–38 Kok A (2001) On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology 38:557–577

Kopp B, Mattler U, Goertz R, Rist F (1996) N2, P3 and the lateralized readiness potential in a NoGo task involving selective response priming. Electroen Clin Neurophysiol 99:19–27 Kornblum S, Lee JW (1995) Stimulus–response compatibility with relevant and irrelevant stimulus dimensions that do and do not overlap with the response. J Exp Psychol Hum 21:855–875 Kornblum S, Hasbroucq T, Osman A (1990) Dimension overlap: cognitive basis for stimulus–response compatibility—a model and taxonomy. Psychol Rev 97:253–270 Kutas M, Donchin E (1980) Preparation to respond as manifested by movement-related brain potentials. Brain Res 202:95–115 Lavric A, Pizzagalli DA, Forstmeier S (2004) When ‘Go’ and ‘NoGo’ are equally frequent: ERP components and cortical tomography. Eur J Neurosci 20:2483–2488 Magliero A, Bashore T, Coles MGH, Donchin E (1984) On the dependence of P300 latency on stimulus evaluation processes. Psychophysiology 21:171–186 Masaki H, Takasawa N, Yamazaki K (2000) An electrophysiological study of the locus of the interference eVect in a stimulus–response compatibility paradigm. Psychophysiology 37:464–472 MeltzoV AN, Decety J (2003) What imitation tells us about social cognition: a rapprochement between developmental psychology and cognitive neuroscience. Philos Trans R Soc Lond B Biol Sci 358:491–500 McCarthy G, Donchin E (1981) A metric of thought: a comparison of P300 latency and reaction time. Science 21:171–186 Nishitani N, Hari R (2000) Temporal dynamics of cortical representation for action. Proc Natl Acad Sci USA 97:913–918 PfeVerbaum A, Ford JM, Weller BJ, Kopell BS (1985) ERPs to response production and inhibition. Electroen Clin Neurophysiol 60:423–434 Prinz W (1997) Perception and action planning. Eur J Cogn Psychol 9:129–154 Ragot R (1984) Perceptual and motor space representation: an event-related potential study. Psychophysiology 21:159–170 Ragot R, Renault B (1981) P300, as the function of S–R compatibility and motor programming. Biol Psychol 13:289–294 Rizzolatti G, Fadiga L, Gallese V, Fogassi L (1996) Premotor cortex and the recognition of motor actions. Cogn Brain Res 3:131–141 Sebanz N, Knoblich G, Prinz W (2003) Representing others’ actions: just like one’s own? Cognition 88:B11–B21 Sebanz N, Knoblich G, Prinz W (2005) How to share a task: co-representing stimulus–response mappings. J Exp Psychol Hum 31:1234–1246 Sebanz N, Bekkering H, Knoblich G (2006) Joint action: bodies and minds moving together. Trends Cogn Sci 10:70–76 Sebanz N, Knoblich G, Prinz W, Wascher E (in press) Twin peaks: an ERP study of action planning and control in co-acting individuals. J Cogn Neurosci Smulders FT, Kok A, Kenemans JL, Bashore TR (1995) The temporal selectivity of additive factor eVects on the reaction process revealed in ERP component latencies. Acta Psychol 90:97–109 Spieler DH, Balota DA, Faust ME (1996) Stroop performance in healthy younger and older adults and in individuals with dementia of the Alzheimer’s type. J Exp Psychol 22:461–469 Valle-Inclan F (1996) The locus of interference in the Simon eVect: an ERP study. Biol Psychol 43:147–162 Zajonc RB (1965) Social facilitation. Science 149:269–274