NeuroImage 101 (2014) 150–158

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

Fair play doesn't matter: MEP modulation as a neurophysiological signature of status quo bias in economic interactions Alberto Pisoni a,⁎,1,e, Emanuele Lo Gerfo a,b,1,e, Stefania Ottone b,e, Ferruccio Ponzano c,e, Luca Zarri d, Alessandra Vergallito a, Leonor Josefina Romero Lauro a,e a

University of Milano Bicocca, Department of Psychology, Italy University of Milano Bicocca, Department of Economics, Management and Statistics, Italy University of Eastern Piedmont, Department of Political Science, Italy d University of Verona, Department of Economics, Italy e NeuroMI - Milan Center for Neuroscience b c

a r t i c l e

i n f o

Article history: Accepted 23 June 2014 Available online 28 June 2014 Keywords: TMS MEP Motor facilitation Economic games Neuroeconomics Status quo bias

a b s t r a c t Transcranial magnetic stimulation (TMS) studies show that watching others' movements enhances motor evoked potential (MEPs) amplitude of the muscles involved in the observed action (motor facilitation, MF). MF has been attributed to a mirror neuron system mediated mechanism, causing an excitability increment of primary motor cortex. It is still unclear whether the meaning an action assumes when performed in an interpersonal exchange context could affect MF. This study aims at exploring this issue by measuring MF induced by the observation of the same action coupled with opposite reward values (gain vs loss) in an economic game. Moreover, the interaction frame was manipulated by showing the same actions within different economic games, the Dictator Game (DG) and the Theft Game (TG). Both games involved two players: a Dictator/Thief and a receiver. Experimental participants played the game always as receivers whereas the Dictator/Thief roles were played by our confederates. In each game Dictator/Thief's choices were expressed by showing a grasping action of one of two cylinders, previously associated with fair/unfair choices. In the DG the dictator decides whether to share (gain condition) or not (no-gain condition) a sum of money with the receiver, while in TGs the thief decides whether to steal (loss condition) or not to steal (no-loss condition) it from the participants. While the experimental subjects watched the videos showing these movements, a single TMS pulse was delivered to their motor hand area and a MEP was recorded from the right FDI muscle. Results show that, in the DG, MF was enhanced by the status quo modification, i.e. MEP amplitude increased when the dictator decided to change the receivers' status quo and share his/her money, and this was true when the status quo was more salient. The same was true for the TG, where the reverse happened: MF was higher for trials in which the thief decided to steal the participants' money, thus changing the status quo, in the block in which the status quo maintenance occurred more often. Data support the hypothesis that the economic meaning of the observed actions differently modulates MEP amplitude, pointing at an influence on MF exerted by a peculiar interaction between economic outcomes and variation of the subjects' initial status quo. © 2014 Elsevier Inc. All rights reserved.

Introduction Action understanding is one of the primary skills that animals have developed throughout evolution (Blakemore and Decety, 2001). The relevance of this ability relies on its role in predicting others' future actions in order to adjust our own, as for example in a prey–predator fight. Actually, action interpretation is important not only for survival, ⁎ Corresponding author at: Piazza dell'Ateneo Nuovo, 1, 20126, Milano, Italy. E-mail address: [email protected] (A. Pisoni). 1 These authors contributed equally.

http://dx.doi.org/10.1016/j.neuroimage.2014.06.056 1053-8119/© 2014 Elsevier Inc. All rights reserved.

but also for successful social interactions. Many social animals indeed, such as dolphins (Pack and Herman, 2007), corvids (Emery and Clayton, 2004) and of course non-human primates (for a review see Call & Tomasello, 2003) are able to understand others' symbolic actions and to respond accordingly. Such an important skill has been thought to have a specific neural network that could explain its mechanisms of action. As a system matching action observation and execution, the mirror neuron system has been considered the basis of action understanding (MNs, Ferrari et al., 2003; Gallese et al., 1996; Rizzolatti et al., 1996). In the last decades, the MNs has been studied in humans (HMNs) with neuro-imaging techniques

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

(PET, fMRI), neurophysiologic measurements (EEG, MEG) and noninvasive brain stimulation techniques (TMS) (Caspers et al., 2010; Grèzes and Decety, 2001; Rizzolatti and Craighero, 2004). Among direct neurophysiological measurements, the most extensively studied phenomenon attributed to the human mirror system, is motor facilitation (MF, Fadiga et al., 1995): a modulation of primary motor cortex excitability due to the observation of an action. Namely, motor evoked potentials (MEPs), which are a measure of motor cortex excitability, increase while observing an action involving the target muscle rather than during a resting state condition (Fadiga et al., 1995). Further studies unveiled several features of MF, showing that it is spatially and temporally strictly coupled with the dynamics of the observed action (Gangitano et al., 2001, 2004), largely dependent on the perspective from which the action is viewed, greater for natural gestures (Maeda et al., 2001, 2002), and affected by previous experience. Moreover, MF can be also induced by the mere listening to action-related sound or speech (see Fadiga et al., 2005 for a review). The reason why it has been argued that MF might be mediated by the HMNs (Fadiga et al., 2005; Rizzolatti and Craighero, 2004) is the strict neural connection between the pre-motor cortex, one of the main locations of mirror neurons in humans (according to neuroimaging studies, Buccino et al., 2001; Vogt et al., 2007), and M1 (Fadiga et al., 2005), connections that presumably can induce cortico-cortical excitation (Strafella and Paus, 2000) during action observation. This hypothesis was supported by a study in which paired pulse TMS was used to assess whether specific cortical areas exerted task-related inhibition or facilitation over the primary motor cortex (see Koch et al., 2006) during hand movement. Koch and colleagues (2010), indeed, found that the excitability of cortico–cortical pathways originating from key areas of the mirror system, such as the anterior intra-parietal cortex and ventral pre-motor cortex, changed in response to the observation of a specific hand posture. Since canonical pre-motor neurons should not respond to action observation, the authors attributed these changes to neurons with “mirror” properties acting as canonical mirror neurons found in monkey's F5 area (Koch et al., 2010). HMNs involvement in action understanding implies action, goals and agent's intention recognition (see Grafton, 2009 for a review but also Hickok et al., 2009 for a critical point of view). A growing body of evidence, indeed, indicates that the HMNs could be targeted by top down regulations, modifying its response as a result of the exposure to the same action with different meaning or goals. For instance, in a TMS study, Fecteau et al. (2010) showed a modulation of MEP amplitude by the vision of symbolic values mimicking hand positions and otherwise ascribed to a neutral connotation. Moreover, greater HMNs activations have been recorded after meaningful rather than meaningless actions (Newman-Norlunda et al., 2010). Similarly, Iacoboni et al. (2005) proved that the same grasping action embedded in different contexts modulated activity in pre-motor areas, thus accounting for a role of the HMNs in coding agent's intentions. However, other studies failed to probe a direct involvement of the HMNs in coding action goals. For instance, Hesse et al. (2008), in two fMRI and TMS studies, showed that the HMNs is mainly involved in coding the means of a seen action, whereas other areas, such as superior frontal, angular and middle temporal gyri, differently responded to particular action ends. Recently, researches focused also on the social value assumed by an action during interpersonal interaction, which is one of the most ecological sets in which the HMNs is supposed to play a relevant role (Hogeveen and Obhi, 2012). In particular, HMNs involvement seems greater for actions performed in interaction contexts as compared to the same actions performed in isolation, suggesting a sensitivity to social situations. Accordingly, an increase in contingent negative variation ERP component (Kourtis et al., 2010), as well as an increment in MF (Sartori et al., 2012), occurred when the observed action could elicit a complementary response rather than when it did not imply any active response. Similarly, the observation of grasping movements performed

151

with social intent (as to build a tower together with a partner) engenders a stronger activation within the HMNs as compared to isolated single movements (Becchio et al., 2012). In conclusion, the HMNs seems sensitive to a wide range of action meanings, including those with social connotation. What has not yet been investigated is whether the reward value attributable to others' actions, in terms of gains and losses, might modulate HMNs activity. Exploring this issue may have interesting implications for a relatively new branch of neuroscience, neuroeconomics, a research area aiming at integrating different disciplines such as economics, psychology and neuroscience, to better understand human choices and decision making in economically relevant interactive contexts. Previous neuroimaging studies have shown that fair and unfair actions may drive different brain responses according to context and interaction meaning (see Sanfey, 2007), as do unfair actions vs omissions (Cushman et al., 2012). In this study we aim at providing a direct neurophysiological measure of HMNs modulation by the reward value of an action, by recording MEPs during the observation of grasping action embedded in two economic game frameworks that imply fair and unfair interactions between two persons. To do so, we used two modified versions of the Dictator Game (Kahneman et al., 1986; Camerer, 2003) — a Mini-Dictator Game (DG) and a Mini-Theft Game (TG). In the DG, one player, the dictator (DC), is endowed with a sum of money and she/he may give half or none of this endowment to a second player (the receiver), who can only accept the dictator's proposal. In the TG, the two players are endowed with the same sum of money and the dictator (in this game named Thief, TF) has the opportunity to steal or not the receiver's endowment. In our study, participants always played as receivers, whereas the DC and TF were created by the experimental setting. The choices of the dictator/thief were communicated through a video showing his/her right hand grasping one of two objects, each associated with opposite reward values (fair or unfair). In the DG, the fair choice was associated with an active change of the initial economic balance between the players, while the unfair one maintained the starting status quo. In the TG, the reverse was true, fair choices being the result of the maintenance of the status quo and the unfair ones implying its modification (Cox et al., 2013). Our hypothesis was that if MF was modulated by action meaning and by the social interaction context in which the action was performed, MEP values should have been modified according to the different experimental conditions, thus tracing a direct relationship between the HMNs and action meaning processing. Experiment 1 Material and methods Participants Twenty-two (11 males, mean age = 25.59 years, SD = 3.72 years) subjects recruited at the University of Milano-Bicocca, took part in Experiment 1. They were naïve as to experimental procedure, and the aim of the study. All subjects were right-handed, as assessed by the Edinburgh handedness inventory (EHI, Oldfield, 1971), and with normal or corrected to normal vision. None of them had a history of chronic or acute neurologic, psychiatric, or medical disease or had any contraindication to TMS (Wassermann, 1998). Written informed consent was obtained from all participants, who were paid for their participation. The experimental protocol was approved by the ethical committee of the University of Milano-Bicocca and was carried out in accordance with the ethical standards of the revised Helsinki Declaration (World Medical Association General Assembly, 2013). TMS and EMG recordings TMS was applied using a Magstim Rapid Transcranial Stimulator (Magstim, Whitland, Dyfed, UK) and a 70-mm figure-of-eight coil. The coil was positioned tangentially to the scalp with the handle pointing backward and laterally at a 45° angle away from the mid-sagittal axis of the subject's head. This placement induces an electric current flow

152

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

in the brain in a posterior–anterior direction, perpendicular to the central sulcus, which has been shown to be optimal for trans-synaptic activation of the cortico-spinal pathways (Brasil-Neto et al., 1992; Mills et al., 1992). The coil was moved over the left M1 in order to identify the TMS hotspot, defined as the point where stimulation evoked the largest MEP from the contralateral first dorsal interosseus (FDI) muscle. The site of stimulation was marked on the surface of a tightly fitting Lycra swimming cap that subjects wore during the experimental session. TMS was delivered at 110% of motor threshold (Loporto et al., 2013), assessed as the lowest intensity of the stimulator output that evoked MEPs with an amplitude of at last 50 μV in the contralateral FDI muscle with a 50% probability when the subject kept the muscle relaxed (Rossini et al., 1994). MEPs were recorded from the FDI muscle of the right hand using 9-mm diameter Ag–AgCl surface cup electrodes. The active electrode was placed over the muscle belly and the reference electrode over the metacarpo-phalangeal joint of the index finger. Responses were amplified with a Digitimer D360 amplifier (Digitimer Ltd, Welwyn Garden City, Hertfordshire, UK) through filters set at 20 Hz and 2 kHz, with a sampling rate of 5 kHz, digitized using an analog-digital converter (Power 1401, Cambridge Electronic Design (CED), Cambridge, UK) and recorded and stored on a personal computer using SIGNAL software (Cambridge Electronic Devices, Cambridge, UK, Version 3.9). Visual stimuli During the experimental procedure, digital video clips were presented on a 19-in. LCD screen placed approximately 80 cm from the subjects' head. Videos showed, in egocentric view, the hands of an actor, grasping one of two metal cylinders (diameter 2 cm; height 1.5 cm) placed in front of his/her right and left hands, each characterized by the presence of a different symbol on the upper side (a triangle or a square). 4 different actors (2 males, 2 females) performed the grasping action for a total of 120 videos. Grasping was a precision grip movement executed with the thumb and the index fingers of the right hand; action timing was strictly synchronized across videos, so that TMS pulse could be delivered around the same stage of action execution in each trial. Specifically, each video lasted 18.5 s, and the grasping action happened after 6 s from the beginning of the trial. TMS was delivered in a jittering interval randomly ranging from 5800 to 6000 ms from the beginning of the video, in order to match stimulation with the maximum index finger–thumb opening amplitude, the moment of M1 FDI area maximum excitability for MEP recording (Gangitano et al., 2001, 2004). Dictator Game structure Subjects were seated in front of a computer screen and were instructed that the computer was connected via internet with another experimental room. The experimenter explained that they would have played online in the role of receivers (RS) in four DGs with four other subjects, one per game, playing the role of the dictator (DC). This experimental setup induced participants to believe that they were actually playing with another real person rather than with a computer, a condition which has been shown affecting participants' reactions to DC's choices (Hertwig and Ortmann, 2001). However, whether participants actually believed the cover story was not explicitly inquired. Each DG consisted of 25 trials. At the beginning of each trial a sum of 100 tokens was awarded to the DC who could choose whether to share the sum with the subject (Fair choice: 50 tokens to the DC and 50 to the RS) or to keep it all for himself/herself (Unfair choice: 100 tokens for the DC and 0 for the RS). The choice was expressed by grasping one of the two small cylinders, which had been coupled with the two possible offer conditions (Fair vs Unfair) at the beginning of the experiment. This association remained the same throughout the whole experiment. The side of the hand performing the action was kept constant, since it is known that it might influence MEP measurements (Aziz-Zadeh et al., 2002). Videos, thus, showed actions performed by right hands only (i.e. subjects' dominant hand), while the association between economic

meaning of left and right token has been balanced across participants: for half of them the left token was coupled with the fair choice and the right one with the unfair choice, while the other half was presented with the reverse association. Before seeing the DC's choice, the RS was prompted to declare which token they predicted that the DC would have grasped by pressing one of two mouse keys with her/his left hand. The experimenter instructed the subjects to answer knowing that their prevision wouldn't have influenced the DC's choice. This prevision could be congruent or noncongruent with the actual DC's choice thus defining each trial as fair/ congruent (FC), fair/incongruent (FI), unfair/congruent (UC), and unfair/incongruent (UI) according to the DC's choice and RS's prevision. After subjects' prediction, the video with the DC's choice was displayed and, concurrently with the DC's grasping action, the TMS pulse was delivered and the MEP was recorded. Immediately after the end of the video, a screen showed how much the subject had won up to that trial (see Fig. 1 for a timeline of the experiment), knowing that earning 50 tokens corresponded to a real payment of 0.20€. Four different DGs (each one corresponding to one experimental block) were created to expose the RS to different types of interaction: a Gain block (80% of Fair and 20% of Unfair choices); a Loss block (20% of Fair and 80% of unfair choices); a Mixed block 1 (40% of Fair and 60% of Unfair choices) and a Mixed block 2 (60% of Fair and 40% of Unfair choices). Mixed blocks were added in order to convey a more realistic DC interaction style. This experimental structure led to a total gain of 2500 tokens, corresponding to 10€ which the subject received at the end of the experimental procedure. Order of blocks, of actor–block association as well as cylinders' value have been balanced across subjects. Experimental procedure Before starting the experimental session, electrodes were placed on the subject's right hand and the TMS hotspot and motor threshold were assessed. Subsequently, subjects were instructed on the experimental procedures and each cylinder was associated with its reward value (fair vs unfair). Two baseline sessions of 25 MEPs each were recorded, at the beginning and at the end of the experimental procedure. During the baseline condition, TMS pulses were delivered at a random ISI ranging from 3800 to 4000 ms while the subject looked at a fixation point on the computer screen. After the first baseline, subjects played four separate blocks of DG, each one made up of 25 trials and corresponding to a different experimental condition (Gain, Loss, Mix1 and Mix2). At the end of the experiment, the total sum of 10€ was provided to the participant. Analysis Trials with EMG activity greater than 100 μV in the 100 ms before TMS were excluded from the analysis to avoid MEP measurements contamination by background activity, as well as MEPs smaller than 50 μV (Sartori et al., 2013). The peak-to-peak amplitude of the remaining MEPs was measured off-line and averaged to derive individual mean values for each trial condition (FC, FI, UC, UI). Individual mean values were normalized to the individual baseline and the value was transformed in z-scores to correct for data normality. Outlier trials (± 2 SD from the subject's mean value) were identified and removed from subsequent analysis. Baseline MEP value was assessed averaging mean MEP amplitude of the first and second baseline sessions, while Mixed block values were derived averaging Mix 1 and Mix 2 blocks. Wilcoxon tests were used to assess differences between baseline 1 and baseline 2, as well as between raw MEP amplitudes of Mix 1 and Mix 2 blocks, since the variables distribution proved to be non-normal (Kolmogorov–Smirnoff's ps all b.05). Moreover, to assess whether MF was present or not, MEPs during baseline condition were compared, by means of a Wilcoxon test, by averaging the mean values of the video blocks, i.e. the four experimental conditions (Gain, Loss, Mix1 and Mix2).

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

153

Fig. 1. Timeline of an experimental trial in the DG.

MEP z-scores were analyzed by means of a repeated measures ANOVA (RM Anova) with Block (3 levels: Gain, Loss and Mixed) Fairness (2 levels: Fair vs Unfair) and Congruency (2 levels: Congruent vs Incongruent) as within subject factors. When a mean MEP value for a condition in a subject was missing (2.6% of cases) it was replaced with linear trend at point method in SPSS version 20 (Rinaldi et al., 2006; Vercoulen et al., 1994). For all statistical analyses, a p value of b0.05 was considered to be significant. All post-hoc analyses have been Bonferroni–Holm corrected. To test whether participants learned the DC's playing style throughout the block, we computed the percentage of “Fair” prediction made by the subjects in three different parts of each block: T1 (trials 1–8), T2 (trials 9–17), T3 (trials 18–25). We ran a repeated measure ANOVA on the arc sine square root transformation of mean number of “Fair” predictions with Block (3 levels: Gain, Loss, Mixed) and Time (3 levels: T1, T2, T3) as independent variables. Greenhouse Geisser correction was applied when sphericity assumption was violated, post hoc tests have been Bonferroni–Holm corrected. The same analyses used for MEPs were applied for baselined root mean square transformation of 100 ms pre-TMS EMG activity to rule out a potential effect given by pre-contraction of the FDI muscle. Results The Wilcoxon tests revealed no significant difference between the two baseline sessions (Z = − .58; p = .57) and between Mix1 and Mix2 blocks (Z = −.89; p = .37). The Wilcoxon test performed on videos vs baseline MEP values was significant (Z = 2.68; p = .007), highlighting a greater MEP amplitude during experimental blocks than during baseline (.74 vs .58 mV respectively), thus confirming the presence of the MF effect. The repeated measures ANOVA revealed a significant main effect of Block [F(2,42) = 4.39; p = .019], values of Gain block being smaller than in the Loss block (MEP z-scores − .19 vs .11 respectively), while both Fairness [F(1,21) = .013; p = .91] and Congruency [F(1,21) = 2.88; p = .104] main effects did not reach significance. The Block × Fairness interaction was significant [F(2,42) = 3.73; p = .032]. Post-hoc analyses showed a greater amplitude for MEPs for Fair choices during Loss block as compared to the same condition in both Gain block (z-scores .22 vs − .27 respectively; p b .001) and Mixed block (z-score −.10; p = .039, see Fig. 2). No other interaction reached significance. The RM Anova ran on pre-TMS EMG activity highlighted no significant main effect or interaction. Crucially, the Block by Fairness interaction was not significant [F(1.38, 28.95) = .67; p = .52]. Concerning participants' predictions, as Fig. 3 shows, subjects did learn DC's playing style, since in Gain blocks they increasingly produced more “Fair” predictions, in Loss block they predicted less “Fair” trials in the final part of the block as compared to the first one, while in Mixed blocks

the percentage of “Fair” predictions remained stable around the chance level (50%). The analysis showed no main effect of Block [F(1.33, 27.94) = .997; p = .35] or Time [F(2, 42) = .50; p = .61]. The interaction, instead, was significant [F(4,84) = 10.85; p b .001, n2 = .34]. Post-hoc comparisons confirmed the increasing trend in producing “Fair” predictions in the Gain block, these being fewer in part 1 (41.4%) as compared to part 2 (58.08%, p = .009) and part 3 (65.78%, p b .0001). Similarly, a decreasing trend in producing “Fair” predictions was present in the Loss block, since the number of these predictions was higher in part 1 (60.55%) as compared to part 3 (48.61, p = .039). On the other hand, no trend was highlighted for Mixed condition, in which parts 1, 2 and 3 did not differ one from the other (all ps N .9). Moreover, while in part 1 “Fair” predictions were more in the Loss block as compared to the Gain one (p b .001), the reverse was true in part 3, where interaction style learning led to a higher percentage of “Fair” predictions in the Gain block as compared to both Loss (p = .002) and Mixed (p = .017) conditions.

Comment Results from Experiment 1 confirmed the MF occurrence, since MEPs were greater during grasping action observation as compared to the baseline condition. The main effect of block and the significant interaction block × fairness suggest that MEPs may be modulated by action

Fig. 2. MEP Z-scores for fair and unfair trials in the Gain, Loss and Mixed DGs. Error bars represent +/- 1 MSE. Asterisks indicate significant differences at p b .05.Non z-transformed MEPs values are provided in the supplementary material section.

154

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

Fig. 3. Participants' percentage of “Fair” predictions over time for the three experimental conditions in the DG. Error bars represent ±/-1 MSE.

meaning, defined as fair or unfair proposals, in an economic game interaction. In particular, Fair offers in the Loss block resulted in a greatly increased cortico-spinal response as compared to the same condition in Mixed and Gain blocks. It would seem reasonable to attribute the observed MEP increment to event frequency, since in the Loss block only 20% of trials represented a DC's fair choice. However, the unfair trials in the Gain block had the same occurrence ratio as fair trials in the Loss block, but did not affect MEP amplitude as compared to the other experimental conditions, thus ruling out that event frequency per se could account for the observed results. In the same way, the negative valence of the interaction style alone (balanced, more fair or more unfair) could not explain the results, since in the main effect of block, with Loss resulting in greater MEPs as compared to Gain block, only fair rather than unfair trials seem responsible for the MEP increment, as revealed by the significant fairness × block interaction. Another possible explanation would implicate participants' expectations. Despite the fact that the congruency main factor was not significant, it is possible that other theories, such as the Pearce Hall unsigned prediction error, might explain our results. The Pearce Hall unsigned prediction error, indeed, posits that during learning, response to a surprising event engages more resources to enhance their processing, rather than expected ones, thus driving a faster learning (Courville et al., 2006; Hayden et al., 2011; Kaye and Pearce, 1984; Swan and Pearce, 1988). Since participants showed a higher response to unusual stimuli that deviated from the norm, we cannot entirely rule out a possible influence of unsigned prediction error in our results. If that was the case, however, two conditions should be present in our data: first, participants should “learn” the playing style throughout the block; second, learning the block condition should somehow influence MEP amplitude. The analysis on participants' predictions showed that subjects did learn the block status, and that this happened quite fast, since the number of “Fair” predictions started to rise from T2 on in the Gain block and tended to decrease in the Loss one. We then expect responses to be higher while subjects are learning the block condition as compared to when no learning is taking place, and this should be generalized through both the Gain and the Loss conditions. To test this prediction, we analyzed MEP amplitude over time performing a repeated measure Anova with Block (3 levels: Gain, Loss and Mixed), Fairness (2 levels: Fair vs Unfair) and Time (3 levels: T1, trials 1–8; T2, trials 9–17; T3, trials 18–25) as independent within subjects variables. Results highlighted a main effect of Time [F(2,42) = 3.46; p = .041] since overall, MEPs tended to decrease with trials. MEP amplitude in the first part of the block, indeed, was higher than the one recorded in the last third (p = .045). Nevertheless, Time did not interact with Block [F(4,84) = .62; p = .65] or with the Block ∗ Fairness

interaction [F(4,84) = .63, p = .64], and approached significance only in interaction with Fairness [F(2,42) = 3.1, p = .059]. This second analysis highlighted a general decreasing tendency of MEP amplitude throughout the blocks. MEP decrement might be due to a general decrease of arousal within each block, which could have affected MEP amplitude (De Gennaro et al., 2007; Sale et al., 2007). Nevertheless, it still may be argued that an unsigned prediction error might exert a role in enhancing MEP amplitude. In the experimental tasks adopted in the present study, we may consider as a surprising event a loss in the Gain block and a gain in the Loss block. As seen, from the participants' prediction analyses, subjects learned the block interaction style in the first third of the trials, since already from T2 they show clear trends toward a higher rate of “Fair” or “Unfair” predictions according to whether they are playing in the Gain or in the Loss block. It is not unusual, then, that MEP modulation decreases even for surprising events when the block's status quo has been learned (i.e. at T2 and T3). Surprising events, indeed, do not serve anymore to learn the interaction style, thus not eliciting higher neurophysiological responses as compared to other trials. However, the effect of Pearce Hall unsigned prediction error cannot account for all our results. The higher neurophysiological reaction to a surprising event, indeed should occur when any surprising event happens, but this is not true for surprising events that did not change the status quo (i.e. unfair trials in the DG Gain blocks and fair trials in the TG Loss block). Notably, in the DG fair choices were the outcome of a direct DC choice, while unfair ones resulted from the DC's decision to maintain the status quo, i.e. to act fairly, the DC had to actively operate to change the initial economic balance between the two players (from a situation where 100–0 tokens were assigned to a 50–50 token outcome). It has been proposed, in decision making and economic theory, that decisions affecting the initial status quo are perceived as less normal and considered less easily justifiable as compared to decisions toward its maintenance (“status quo bias”, Connolly and Zeelenberg, 2002; Samuelson and Zeckhauser, 1988). Similarly, fair and unfair actions resulting from an active action are perceived respectively as better or worse as compared to outcomes resulting from a lack of change in the status quo (Baron and Ritov, 1994; Cox et al., 2013; Ritov and Baron, 1992, 1995). Accordingly, in our experimental paradigm, given that every DC choice was expressed by a hand movement, trials that elicited the greatest MF effect were the ones deriving from an active choice of the DC changing the receiver's status quo defined at the beginning of the interaction, i.e. fair choices in the Loss block. Fair choices, indeed, implied a change of the initial endowment. Moreover, the Loss block was the one with a stronger status quo definition, since the initial status quo of no endowment was maintained in 80% of the trials. Therefore, a possible interpretation of this first experiment's data is that rather than a main effect of gain/loss or fairness/unfairness per se, what could be responsible for MEP modulation is how fairness is coupled with an active change of the status quo when this is more salient. To test whether a different combination of action/omission and fairness/unfairness conditions could differently modulate MEP amplitude, or if the pattern of the present results was specific for Experiment 1 design, we ran a second experiment (Experiment 2) mirroring the first one. Namely, instead of the DG we used the TG in which the dictator is replaced by a thief (TF). As in Experiment 1, the TF was created by the experimental setting, whereas study participants were enrolled as receivers (RS). In the TG both players had 50 tokens assigned at the beginning of each trial. The TF could decide to steal or not RS' initial sum, thus performing an active unfair action (loss) or a fair omission (noloss). Notably, in this case the coupling of fair/unfair action/omission is reversed. If status quo modification and active value of the action were the real factors influencing the results from Experiment 1, we expect to find a MEP modulation of trials in which an active choice is taken (unfair) in the condition in which the status quo (50 tokens per each player) is mostly maintained (Gain block).

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

155

Experiment 2 Materials and methods Participants Twenty-two (8 men, mean age = 25.36 years, SD = 2.55 years) participants, different from those enrolled in the first experiment, took part in Experiment 2. As for Experiment 1, they were Italian righthanded undergraduate students, and the same inclusion criteria and recruitment procedures were applied.

Theft Game structure As for the DG, in the TG participants were told that they would have played online as RS with four other subjects, who had been assigned to the role of the thief. Visual stimuli were the same as in Experiment 1. At the beginning of each trial a sum of 50 tokens was awarded to both players and the TF could choose whether to give up the RS's tokens (Fair choice: 50 tokens to the TF and 50 to the RS) or to steal them (Unfair choice: 100 tokens for the TF and 0 for the RS). As in Experiment 1, the TF's choice was expressed by showing a video of his/her hand grasping one of two cylinders, coupled with the choice meaning at the beginning of the experimental session. TMS and EMG protocol were identical to Experiment 1 as well as experimental block design. Four different TGs, with different winning contexts, were then created: Gain block (80% of Fair and 20% of Unfair choices); Loss block (20% of Fair and 80% of Unfair choices); Mixed block 1 (40% of Fair and 60% of Unfair choices); Mixed block 2 (60% of Fair and 40% of Unfair choices). As for Experiment 1, actors, blocks, and TF's choice value coupling with the two symbols were balanced across participants. Subjects knew that 50 tokens were valued at 0.20€ and as for the DGs, the experimental procedure led to a fixed payment of 10€ per subject. Analysis The same analysis procedure performed on data from Experiment 1 has been ran on MEPs from Experiment 2. Missing mean MEP values (3.4% of cases) were replaced with linear trend at point method in SPSS version 20. Time course of participants' predictions and MEP modulation were also analyzed following procedures adopted in Experiment 1.

Results As for Experiment 1 the Wilcoxon tests revealed no significant difference between the two baseline sessions (Z = − .89; p = .37) and between Mix1 and Mix2 blocks (Z = − .54; p = .59), while the Wilcoxon test performed on videos vs baseline MEP values resulted significant (Z = 2.65; p = .008), confirming the MF effect (.67 vs .49 mV). The repeated measures ANOVA showed no significant main effect of Block [F(2,42) = 0.4; p = .67], Fairness [F(1,21) = .1.31; p = .26] or Congruency [F(1,21) = .014; p = .90]. The Block × Fairness [F(2,42) = 7.17; p = .002] and Block × Congruency [F(2,42) = 3.43; p = .04] interactions showed significant results. Post-hoc analyses on the Block × Fairness interaction highlighted a greater amplitude for MEPs recorded during Unfair choices as compared to Fair ones in the Gain block (zscores .18 vs −.19 respectively; p = .007) and for Unfair choices in Gain block as compared to Loss ones (z-scores .18 vs − .17; p = .013, see Fig. 4). Even if post hoc tests showed no significant difference, a visual inspection of the tendencies for congruent and incongruent trials in the three blocks seems to indicate a higher z MEP value for incongruent trials in the Gain block as compared to congruent ones in the same condition and incongruent in the Loss block (Fig. 5). No other interaction reached significance.

Fig. 4. MEP Z-scores for fair and unfair trials in the Gain, Loss and Mixed TGs. Error bars represent ±/-1 MSE. Asterisks indicate significant differences at p b .05. Non z-transformed MEPs values are provided in the supplementary material section.

Pre-TMS EMG analysis showed no significant main effect or interaction (all ps N .18). In particular the Block by Fairness interaction was not significant [F(2,40) = .018; p = .98]. Participants' prediction analysis highlighted a significant main effect of Block [F(1.27,25.34) = 8.39, p = .005; n2 = .29], “Fair” predictions being generally more present in the Gain block (64%) as compared to the Loss condition (46.3%, p b .001). Time main effect was not significant [F(2,40) = 2.7, p = .08; n2 = .12] while Block by Time interaction was [F(2.55, 51.11) = 6, p b .001; n2 = .23]. Post hoc comparisons confirmed the increasing trend in producing “Fair” predictions in the Gain block, the percentage being lower in part 1 (52.27%) as compared to both parts 2 (65.61%; p = .043) and 3 (74.15%; p b .001). The tendency to produce less “Fair” predictions over time in the Loss block, instead, did not show significant results. Even if part 1 included more “Fair” predictions (50.74%) as compared to part 3 (43.01%), indeed, the difference between the conditions was not significant (p = .8). Confirming the DG's analysis, no trend was highlighted for Mixed condition (all p's N .9). Moreover, while in part 1 no difference was highlighted between the three blocks (all p's N .9), they started to differ in part 2, with Gain block presenting more “Fair” predictions (65.6%) as compared to Loss block (45.18%, p b .001). Lastly, in part 3, Gain block showed more “Fair” predictions (74.15%) than both Loss (43.01%; p b .001) and Mixed (56.56%; p = .004) conditions (see Fig. 6). Concerning MEP changes over time, the analysis highlighted, as for the DG, a significant main effect of Time [F(2,40) = 7.67; p = .002], since MEPs recorded in the first part of the block showed higher values

Fig. 5. MEP modulation for congruent and incongruent trials over the Blocks in the TG. Error bars represent +/-1 MSE.

156

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

Fig. 6. Participants' percentage of Fair predictions over time for the three experimental conditions in the TG. Error bars represent +/-1 MSE.

compared to both the second (p = .023) and the third parts of the block (p = .001). Moreover, the Block by Time by Fairness interaction showed significant results [F(4,80) = 2.84; p = .031]. Post hoc analysis revealed a greater MEP amplitude at T1 for Unfair trials in the Gain block as compared to fair trials in the same block (p b .001) and unfair trials in the Loss block, and lower amplitude for unfair trials at T3 in the Gain block as compared to the same type of trials in the same block at T1. As for the DG, in the TG participants learned the block condition during T1, showing tendencies to produce more or less “Fair” predictions in accordance with the majority of trials from T2 on. MEP amplitude decreased over time and this was more pronounced for unfair trials in the Gain block. In the TG, thus, learning did interact with the experimental design, supporting a possible unbiased prediction error-like mechanism influencing the highlighted MF modulation. Nevertheless, as described for the DG, this still did not account for the lack of such an effect for fair trials in the Loss block. General discussion To the best of our knowledge this is the first study which systematically tested the influence on MF of fair and unfair actions during social interactions in economic games. In line with previous studies (Clark et al., 2003; Fadiga et al., 1995; Gangitano et al., 2004; Maeda et al., 2002; see Fadiga et al., 2005 for review) our results confirm the MF effect: cortico-spinal excitability increases during the observation of a movement in which the muscles, whose M1 area is targeted by TMS, are involved. In both experiments, indeed, MEP amplitude was greater while subjects watched videos showing grasping actions as compared to the baseline condition. In the DG, during Loss blocks, in which the unfair actions were the majority of trials, the view of a movement matched with a fair choice elicited a larger MEP than the same action performed in Gain blocks, in which there was a preponderance of fair trials, and in Mixed blocks, where the fair and unfair choice ratio was balanced. In the TG, instead, MEP amplitude was significantly greater during Gain blocks for unfair choices than for fair ones and for the same type of action in Loss blocks. In both experiments no difference in the modulation of MEP amplitude was highlighted for Mixed blocks. Environmental and social context in which an action is framed play an important role in understanding action meaning, and the HMNs likely underlies this ability. The role of the context has already been investigated in fMRI studies (Iacoboni et al., 2005; Kaplan and Iacoboni, 2006) showing a modulation of HMNs responses to observed actions differently framed. Moreover, in an EEG study, Oberman et al. (2007) demonstrated that mu wave suppression, used as an index of MN activity, was greater for experimental conditions where subjects watched videos showing social interactions (three individuals tossed a ball to

each other and occasionally the ball would be thrown off the screen toward the viewer), as compared to videos showing no interaction (three individuals tossed a ball up in the air to themselves). These results suggest that the HMNs is not only engaged in action observation and simulation but also that it is specialized in stimuli with social relevance (Oberman et al., 2007). The results of the present experiments do not highlight a modulatory effect on MF by action meaning (fair vs unfair) per se, since in both experiments the main effect of fairness was not significant. The significant Block × Fairness interaction rather suggests a role of the specific valence an action assumes according to the interaction context, in our case defined by the DC/TF's interaction styles. In particular, game rules differently denoted the meaning of the two possible choices: fair actions appeared as a gain in DGs and as a no-loss in TGs, while unfair actions as a no-gain trial in DGs and as a loss in TGs. In the DG an increase of MEP amplitude occurred when the RS received fair proposals in a no-gain context, whereas in the TG that was true when the RS received unfair proposals in a no-loss context. It seems, then, that just manipulating the game-frame in which responders are playing might be enough to change their point of view and their reactions to fair and unfair offers, even if the purely economic outcome of the two possible choices was identical across games (+50 tokens for fair trials and −50 tokens for unfair ones). This is in line with previous behavioral economics studies, which already suggested that the evaluation of loss or gain during a choice is not based on absolute values, but rather depends on a subjective point of view that can be modified according to the way the choice is framed (Tversky and Kahneman, 1981). Although it would seem reasonable to ascribe a role in the highlighted MEP amplitude modulation to event frequency or to participants' prediction error, these alternative explanations cannot account for all the present results. Participants effectively learned whether in a certain block interactions tended toward a majority of gain or loss trials, accordingly changing their predictions, and MEP size was higher at the beginning of the blocks, thus highlighting a potential role of prediction error in maximizing learning for unexpected events. However this explanation cannot account for the different effect of the Block by Fairness interaction on MEP size. Since in both games Gain and Loss blocks had the same percentage of infrequent events, the same number of surprising effects boosting learning was present. Nevertheless, only trials modifying the status quo in a context where this was mostly confirmed, showed a differential MEP amplitude modulation, and this was true for both experimental games. The results obtained in the two experiments could be thus explained by a difference in perceived change of the game-related subjects' status quo. At the beginning of each DG trial, indeed, our experimental subjects had no token, so that the unfair choice, in which the DC did not share his sum with the RS, represented a maintenance of his initial status quo. Fair choices, on the other hand, represented an event which changed RS' starting condition: there was indeed a movement of half of the DC's sum to the RS's total amount of tokens. The opposite was true for the TG, where fair choices preserved the RS' initial condition (RS had 50 tokens and the TF didn't steal them) while unfair ones represented a modification of the RS' starting status (the TF took RS' sum). Interestingly, this status quo change is mainly salient in blocks where its maintenance is more frequent, namely the Loss block in the DG and the Gain block in the TG. MF modulation, indeed, increased in these two blocks and in the trials where the status quo changed (Cox et al., 2013; Ritov and Baron, 1995; Samuelson and Zeckhauser, 1988). Previous research highlighted what is commonly defined as the “status quo bias”, i.e. the tendency of having a stronger preference in decision making for the option which does not change the initial condition (Connolly and Zeelenberg, 2002; Kahneman and Miller, 1986), which in turn has been linked to the “omission bias”, a judgment discrepancy between unfair omission and commission (Baron and Ritov, 1994; Ritov and Baron, 1992, 1995). It has been shown, indeed, both in experimental and ecological settings, that regardless of the outcome, choices that

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

imply a maintenance of the status quo are selected more frequently (Samuelson and Zeckhauser, 1988), and when an unfair choice, such as not helping someone, is the product of an omission (thus a choice that does not change the status quo), it is considered less negative than when the same outcome is achieved through an active action (Baron and Ritov, 1994; Ritov and Baron, 1992, 1995). These biases apply also in economic games as shown by a study by Cox et al. (2013). The authors investigated, in a dictator game setting, whether both fair and unfair active choices of the dictator corresponded to stronger fair or unfair responses of the recipient, as compared to fair or unfair omissions. In particular they showed that in a scenario where the RS's gain condition corresponded to the DC's active fair action (as in our Experiment 1), the RS was more prone to actively reward the DC as compared to a scenario in which the DC's fair choice was the product of an omission (as in our Experiment 2). The same was true for unfair choices: when these were the outcome of an active choice, DCs were punished more frequently, as compared to when they followed an omission. Similarly in our experiments we found that the status quo plays a relevant role. In particular, we found a greater cortical excitability enhancement when fair behavior was the product of an active choice, a change of the status quo, a condition exaggerated by the low ratio with which it happened (Loss block of Experiment 1), and in the mirrored condition, i.e. when an unfair behavior resulted from an active choice that disrupted the initial and most frequent condition (Gain block of Experiment 2). The cortico-spinal excitability modulation found in the present study may also depend on the emotional effect that the proposer's choices elicit in the RS. A recent TMS study (Borgomaneri et al., 2013) investigated how motor excitability, assessed by MEPs recording, was influenced during observation and categorization of positive, neutral and negative pictures. They found an increase of MEP amplitude for negative and positive pictures at different timings, providing thus a direct support for the notion that emotion perception is linked to action system. Emotional arousal has been found also to be strictly linked with the status quo bias itself. Nicolle et al. (2011), in a study investigating the psychological and neural bases of this bias, highlighted a stronger activation of areas usually linked with emotion processing, such as left and right insula and medial prefrontal cortex, for errors after changes against the status quo, as well as a greater subjective regret rate, as compared to errors related to choices which maintained the status quo. This enhanced negative emotional feeling could be the drive for subjects' subsequent avoidance of this type of choices, thus accounting for the bias occurrence. Moreover, status quo modification has been strictly linked with basal ganglia activity as well. Specifically, Fleming et al. (2010) highlighted increased right inferior frontal gyrus and bilateral sub-thalamic nucleus activity after rejection of the status quo conditions for difficult decision contexts, confirming the role of these structures in decision making and action selection. Our data are congruent with these results, since we found that cortical excitability is linked to emotions elicited by social interaction, and especially by those changing the baseline relational balance, and with learning components, which arise when interactions are embedded in a frame that involves multiple exchanges between confederates. It could be speculated that this increased excitability after status quo modification might help in signaling a change in the normal balance between partners that may be potentially harmful or fruitful for the subject and thus prompting potential actions to respond accordingly. Further researches are needed to better investigate these implications. In line with previous studies (Guo et al., 2013; Kahneman and Tversky, 1979), the interaction between game framing and action fairness also depends on gain context, thus supporting the role of reward contingencies and social context in economic choices and interactions. As a last warning footnote, we have to point out that no data on participants' belief of being involved in an economic interaction with a real person was recorded. It has to be noted, however, that since interaction with a computer usually led to similar but attenuated responses both at

157

the behavioral and neurophysiological levels as compared to human interactions (Rilling et al., 2004), if our participants realized that they were playing with a computer and not with a real person our results would be weakened rather than strengthened. In conclusion, in this study we assessed whether action meaning, defined as economically fair or unfair, and context, in terms of different interaction styles, may modulate MF. Economic exchange, indeed, especially when imposed by only one part (proposer), can elicit positive or negative emotions according to the context in which it occurs, implying different moral evaluation and reactions according to whether an action is linked with an active choice breaking the status quo vs a passive omission. Our results demonstrate that fair or unfair choices differently modulate MEP amplitude in relation to different social contexts (Loss vs Gain) and action meaning (fair vs unfair), and that this modulation is strictly linked to the reference game frame. Since MF is considered to be a phenomenon mediated by the HMNs, our results suggest that this system may be influenced by action meaning, unveiling a potential neurophysiological marker of the well documented status quo bias. Future research should further investigate whether other context modulations (participants' personality features, direct vs indirect involvement in the game, third party games) might affect the observed results. Acknowledgments This study was supported by the Center for Interdisciplinary Studies in Economics, Psychology & Social Sciences (CISEPS). We also acknowledge the University of Verona (2011 Joint Projects on “Punishment and Decision-making: Neuroeconomic Foundations, Behavioural Experiments and Implications for Law and Economics”) for financial support. We thank the two anonymous Reviewers for insightful suggestions and comments. Appendix A. Supplementary material Supplementary materials related to this article can be found online at http://dx.doi.org/10.1016/j.neuroimage.2014.06.056. References Aziz-Zadeh, L., Maeda, F., Zaidel, E., Mazziotta, J., Iacoboni, M., 2002. Lateralization in motor facilitation during action observation: a TMS study. Exp. Brain Res. 144 (1), 127–131. Baron, J., Ritov, I., 1994. Reference points and omission bias. Organ. Behav. Hum. Decis. Process. 59, 475–498. Becchio, C., Cavallo, A., Begliomini, C., Sartori, L., Feltrin, G., Castiello, U., 2012. Social grasping: from mirroring to mentalizing. NeuroImage 61, 240–248. Blakemore, S.J., Decety, J., 2001. From the perception of action to the understanding of intention. Nat. Rev. Neurosci. 2, 561–567. Borgomaneri, S., Gazzola, V., Avenanti, A., 2013. Temporal dynamics of motor cortex excitability during perception of natural emotional scenes. Soc. Cogn. Affect. Neurosci. 12 (Sep). Brasil-Neto, J.P., Cohen, L.G., Panizza, M., Nilsson, J., Roth, B.J., Hallett, M., 1992. Optimal focal transcranial magnetic activation of the human motor cortex: effects of coil orientation, shape of the induced current pulse, and stimulus intensity. J. Clin. Neurophysiol. 9, 132–136. Buccino, G., Binkofski, F., Fink, G.R., Fadiga, L., Fogassi, L., Gallese, V., Seitz, R.J., Zilles, K., Rizzolatti, G., Freund, H.J., 2001. Action observation activates premotor and parietal areas in a somatotopic manner: an fMRI study. Eur. J. Neurosci. 13, 400–404. Camerer, C.F., 2003. Behavioral Game Theory: Experiments in Strategic Interaction. Princeton, Princeton University Press. Caspers, S., Zilles, K., Laird, A.R., Eickhoff, S.B., 2010. ALE meta-analysis of action observation and imitation in the human brain. NeuroImage 15, 1148–1167. Clark, S., Tremblay, F., Ste-Marie, D., 2003. Differential modulation of corticospinal excitability during observation, mental imagery and imitation of hand actions. Neuropsychologia 42, 105–112. Connolly, T., Zeelenberg, M., 2002. Regret in decision making. Curr. Dir. Psychol. Sci. 11, 212–216. Courville, A.C., Daw, N.D., Touretzky, D.S., 2006. Bayesian theories of conditioning in a changing world. Trends Cogn. Sci. 10, 294–300. Cox, J.C., Servatka, M., Vadovic, R., 2013. Status quo effects in fairness games: reciprocal responses to acts of commission vs. Acts of omission. Experimental Economics Center Working Paper Series 2012-03. Experimental Economics Center, Andrew Young School of Policy Studies, Georgia State University.

158

A. Pisoni et al. / NeuroImage 101 (2014) 150–158

Cushman, F., Murray, D., Gordon-McKeon, S., Wharton, S., Greene, J.D., 2012. Judgment before principle: engagement of the frontoparietal control network in condemning harms of omission. Soc. Cogn. Affect. 7, 888–895. De Gennaro, L., Marzano, C., Veniero, D., Moroni, F., Fratello, F., Curcio, G., Ferrara, M., Ferlazzo, F., Novelli, L., Concetta Pellicciari, M., Bertini, M., Rossini, P.M., 2007. Neurophysiological correlates of sleepiness: a combined TMS and EEG study. Neuroimage 13, 1277–1287. Emery, N.J., Clayton, N.S., 2004. The mentality of crows: convergent evolution of intelligence in corvids and apes. Science 10, 1903–1907. Fadiga, L., Fogassi, L., Pavesi, G., Rizzolatti, G., 1995. Motor facilitation during action observation: a magnetic stimulation study. J. Neurophysiol. 73, 2608–2611. Fadiga, L., Craighero, L., Olivier, E., 2005. Human motor cortex excitability during the perception of others' action. Curr. Opin. Neurobiol. 15, 213–218. Fecteau, S., Tormos, J.M., Gangitano, M., Theoret, H., Pascual-Leone, A., 2010. Modulation of cortical motor outputs by the symbolic meaning of visual stimuli. Eur. J. Neurosci. 32, 172–177. Ferrari, P.F., Gallese, V., Rizzolatti, G., Fogassi, L., 2003. Mirror neurons responding to the observation of ingestive and communicative mouth actions in the monkey ventral premotor cortex. Eur. J. Neurosci. 17, 1703–1714. Fleming, S.M., Thomas, C.L., Dolan, R.J., 2010. Overcoming status quo bias in the human brain. Proc. Natl. Acad. Sci. U. S. A. 30, 6005–6009. Gallese, V., Fadiga, L., Fogassi, L., Rizzolatti, G., 1996. Action recognition in the premotor cortex. Brain 119, 593–609. Gangitano, M., Mottaghy, F.M., Pascual-Leone, A., 2001. Phase-specific modulation of cortical motor output during movement observation. Neuroreport 12, 1489–1492. Gangitano, M., Mottaghy, F.M., Pascual-Leone, A., 2004. Modulation of premotor mirror neuron activity during observation of unpredictable grasping movements. Eur. J. Neurosci. 20, 2193–2202. Grafton, S.T., 2009. Embodied cognition and the simulation of action to understand others. Ann. N. Y. Acad. Sci. 1156, 97–117. Grèzes, J., Decety, J., 2001. Functional anatomy of execution, mental simulation, observation, and verb generation of actions: a meta-analysis. Hum. Brain Mapp. 12, 1–19. Hayden, B.Y., Heilbronner, S.R., Pearson, J.M., Platt, M.L., 2011. Surprise signals in anterior cingulate cortex: neuronal encoding of unsigned reward prediction errors driving adjustment in behavior. J. Neurosci. 16, 4178–4187. Hertwig, R., Ortmann, A., 2001. Experimental practices in economics: a methodological challenge for psychologists? Behav. Brain Sci. 24, 383–451. Hesse, M.D., Sparing, R., Fink, G.R., 2008. End or means—the “what” and “how” of observed intentional actions. J. Cogn. Neurosci. 21, 776–790. Hogeveen, J., Obhi, S.S., 2012. Social interaction enhances motor resonance for observed human actions. J. Neurosci. 25, 5984–5989. Iacoboni, M., Molnar-Szakacs, I., Gallese, V., Buccino, G., Mazziotta, J.C., Rizzolatti, G., 2005. Grasping the intentions of others with one's own mirror neuron system. PLoS Biol. 3, 529–535. Kahneman, D., Miller, D.T., 1986. Norm theory: comparing reality to its alternatives. Psychol. Rev. 93, 136–153. Kahneman, D., Tversky, A., 1979. Prospect theory: an analysis of decision under risk. Econometrica 47, 263–291. Kahneman, D., Knetsch, J., Thaler, R.H., 1986. Fairness and the assumptions of economics. J. Bus. 59, 285–300. Kaplan, J.T., Iacoboni, M., 2006. Getting a grip on other minds: mirror neurons, intention understanding, and cognitive empathy. Soc. Neurosci. 1, 175–183. Kaye, H., Pearce, J.M., 1984. The strength of the orienting response during Pavlovian conditioning. J. Exp. Psychol. Anim. Behav. Process. 10, 90–109. Kourtis, D., Sebanz, N., Knoblich, G., 2010. Favouritism in the motor system: social interaction modulates action simulation. Biol. Lett. 23, 758–761. Loporto, M., Holmes, P.S., Wright, D.J., McAllister, C.J., 2013. Reflecting on mirror mechanisms: motor resonance effects during action observation only present with lowintensity transcranial magnetic stimulation. PLoS One 28, e64911. Maeda, F., Chang, V.Y., Mazziotta, J., Iacoboni, M., 2001. Experience dependent modulation of motor corticospinal excitability during action observation. Exp. Brain Res. 140, 24–44. Maeda, F., Kleiner-Fisman, G., Pascual-Leone, A., 2002. Motor facilitation while observing hand actions: specificity of the effect and role of observer's orientation. J. Neurophysiol. 87, 1329–1335.

Mills, K.R., Boniface, S.J., Schubert, M., 1992. Magnetic brain stimulation with a double coil: the importance of coil orientation. Electroencephalogr. Clin. Neurophysiol. 85, 17–21. Newman-Norlunda, R., van Schieb, H.T., van Hoekc, M.E.C., Cuijpersc, R.H., Bekkeringd, H., 2010. The role of inferior frontal and parietal areas in differentiating meaningful and meaningless object-directed actions. Brain Res. 22, 63–74. Nicolle, A., Fleming, S.M., Bach, D.R., Driver, J., Dolan, R.J., 2011. A regret-induced status quo bias. J. Neurosci. 2, 3320–3327. Oberman, L.M., Pineda, J.A., Ramachandran, V.S., 2007. The human mirror neuron system: a link between action observation and social skills. Soc. Cogn. Affect. Neurosci. 2, 62–66. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97–113. Pack, A.A., Herman, L.M., 2007. The dolphin's (Tursiops truncatus) understanding of human gazing and pointing: knowing what and where. J. Comp. Psychol. 121, 34–45. Rilling, J.K., Sanfey, A.G., Aronson, J.A., Nystrom, L.E., Cohen, J.D., 2004. The neural correlates of theory of mind within interpersonal interactions. Neuroimage 22, 1694–1703. Rinaldi, L., Gallo, P., Calabrese, M., Ranzato, F., Luise, D., Colavito, D., Motta, M., Guglielmo, A., Del Giudice, E., Romualdi, C., Ragazzi, E., D'Arrigo, A., Dalle Carbonare, M., Leontino, B., Leon, A., 2006. Longitudinal analysis of immune cell phenotypes in early stage multiple sclerosis: distinctive patterns characterize MRI-active patients. Brain 129, 1993–2007. Ritov, I., Baron, J., 1992. Status-quo and omission bias. J. Risk Uncertain. 5, 49–61. Ritov, I., Baron, J., 1995. Outcome knowledge, regret, and omission bias. Organ. Behav. Hum. Decis. Process. 64, 119–127. Rizzolatti, G., Craighero, L., 2004. The mirror neuron system. Annu. Rev. Neurosci. 27, 169–192. Rizzolatti, G., Fadiga, L., Gallese, V., Fogassi, L., 1996. Premotor cortex and the recognition of motor actions. Brain Res. Cogn. Brain Res. 3, 131–141. Rossini, P.M., Barker, A.T., Berardelli, A., Caramia, M.D., Caruso, G., Cracco, R.Q., Dimitrijević, M.R., Hallett, M., Katayama, Y., Lücking, C.H., et al., 1994. Non-invasive electrical and magnetic stimulation of the brain, spinal cord and roots: basic principles and procedures for routine clinical application. Report of an IFCN committee. Electroencephalogr. Clin. Neurophysiol. 91, 79–92. Sale, M.V., Ridding, M.C., Nordstrom, M.A., 2007. Factors influencing the magnitude and reproducibility of corticomotor excitability changes induced by paired associative stimulation. Exp. Brain Res. 18, 615–626. Samuelson, W., Zeckhauser, R., 1988. Status quo bias in decision making. J. Risk Uncertain. 1, 7–59. Sanfey, A.G., 2007. Social decision-making: insights from game theory and neuroscience. Science 26, 598–602. Sartori, L., Cavallo, A., Bucchioni, G., Castiello, U., 2012. From simulation to reciprocity: the case of complementary actions. Soc. Neurosci. 7, 146–158. Sartori, L., Betti, S., Castiello, U., 2013. Corticospinal excitability modulation during action observation. J. Vis. Exp. 31, 51001. Strafella, A.P., Paus, T., 2000. Modulation of cortical excitability during action observation: a transcranial magnetic stimulation study. Neuroreport 11, 2289–2292. Swan, J.A., Pearce, J.M., 1988. The orienting response as an index of stimulus associability in rats. J. Exp. Psychol. Anim. Behav. Process. 14, 292–301. Tversky, A., Kahneman, D., 1981. The framing of decisions and the psychology of choice. Science 30, 453–458. Vercoulen, J.H., Swanink, C., Fennis, J.F., Galama, J., van der Meer, J.W., Bleijenberg, G., 1994. Dimensional assessment of chronic fatigue syndrome. J. Psychosom. Res. 38, 383–392. Vogt, S., Buccino, G., Wohlschlager, A.M., Canessa, N., Shah, N.J., Zilles, K., Fink, G.R., 2007. Prefrontal involvement in imitation learning of hand actions: effects of practice and expertise. NeuroImage 37, 1371–1383. Wassermann, E.M., 1998. Risk and safety of repetitive transcranial magnetic stimulation: report and suggested guidelines from the International Workshop on the Safety of Repetitive Transcranial Magnetic Stimulation (June 5–7, 1996. 108, 1–16. Figure captions).