Introduction
In our everyday life, we regularly coordinate our actions with others, e.g., to divide labor and save energy, or to make use of each other’s expertise and accomplish more than we could on our own (Beckes & Coan,
2011; Coan, Schaefer, & Davidson,
2006; Schnall, Harber, Stefanucci, & Proffitt,
2008; Sebanz, Knoblich, & Prinz,
2003). However, the joint coordination of actions also introduces ambiguity in action control. That is, merely observing other people’s actions activates the same brain areas as when we perform the action ourselves (Carr, Iacoboni, Dubeau, Mazziotta, & Lenzi,
2003; Cochin, Barthelemy, Roux, & Martineau,
1999; Hari et al.,
1998; Mukamel, Ekstrom, Kaplan, Iacoboni, & Fried,
2010). Hence, during action coordination, it may be hard to distinguish our own actions and their consequences from those of others (Brass, Ruby, & Spengler,
2009; Liepelt et al.,
2016). If people integrate (and potentially confuse) their own and other people’s actions during joint action control, how then, do they experience control over jointly coordinated actions?
Over the past decade, research has revealed much on how people experience control over their own behavior in social isolation (Aarts, Custers, & Wegner,
2005; Frith, Blakemore, & Wolpert,
2000; Moore, Lagnado, Deal, & Haggard, 2009; van der Weiden, Aarts, & Ruys,
2013; Wegner & Wheatley,
1999). This research has demonstrated that experienced control crucially depends on how people represent their behavior (Aarts et al.,
2012; Aarts, Custers, & Marien,
2009; David et al.,
2006; Pacherie,
2008; Vallacher & Wegner,
1989; van der Weiden, Aarts, & Ruys,
2010; van der Weiden, Ruys, & Aarts,
2013). That is, people typically experience control over an action (e.g., playing piano) when its triggers (e.g., notes on the bass clef), performance (e.g., a left-hand movement), and consequences (e.g., low tones) match the prior internal representation of the action (e.g., when the action was represented in terms of notes on the bass clef, left-hand movements, and low tones; Moore et al.,
2009; Morsella et al.,
2009; Sidarus & Haggard,
2016; van der Weiden et al.,
2010; van der Weiden, Aarts, & Ruys,
2011; Wenke, Fleming, & Haggard,
2010).
As actions are typically represented in terms of desired outcomes (i.e., in terms of overarching goals; Vallacher & Wegner,
1987; van der Weiden et al.,
2010), people usually experience control when they attain their (conscious or unconscious) goals, even in the absence of actual control (Aarts et al.,
2009,
2005; Preston & Newport,
2010; J. L. Preston, Ritter, & Wegner,
2011; van der Weiden, Ruys, et al.,
2013). But also earlier action selection processes that take place before goal attainment play an important role, such that experienced control is strongest when a matching representation of the required action is pre-activated (i.e., in the absence of response conflict; Morsella et al.,
2009; Sidarus & Haggard,
2016; Wenke et al.,
2010). In one study that tested the role of action selection processes in experienced control, participants performed a Stroop task (Stroop,
1935) in which they had to name the font color of stimulus words (e.g., blue), while the stimulus words sometimes triggered a different naming response (e.g., when the word was “red”), inducing response conflict. After each trial, participants indicated to what extent they experienced control over their response. Results showed that participants experienced more control over their actions when a representation of the required response (e.g., blue) was triggered by a matching stimulus word (e.g., “blue” vs. “red”).
Such response selection processes become more complicated when coordinating multiple (e.g., left and right hand) actions simultaneously. That is, each action requires different planning and execution and may be associated with different triggers and with different expectations regarding its consequences (e.g., when playing piano, notes on the bass clef trigger left-hand actions that produce low tones, while notes on the treble clef trigger right-hand actions that produce high tones). In order to successfully distinguish and control multiple actions, people tend to represent (
or code) their actions in reference to each other (e.g., as “bass,” “left,” “low” vs. “treble,” “right,” “high”; Craft & Simon,
1970; Elsner & Hommel,
2001; Hommel,
1993,
1996b; Simon,
1990), also referred to as
referential coding. As such, people can represent (and experience control over) multiple actions at the same time. With regard to experienced control, one would expect the strongest experiences of control when one’s actions are compatible with these referential representations (e.g., when left-hand movements produce low rather than high tones).
Although experiences of control have been associated with social behavior and moral responsibility (Damen, van Baaren, Brass, Aarts, & Dijksterhuis,
2015; Frith,
2013; Moretto, Walsh, & Haggard,
2011; Ruys & Aarts,
2012), relatively little is known about how such processes affect people’s experiences of control when interacting with others. First evidence on experiences of control in joint action suggests that similar to individual action, people experience more control over jointly coordinated actions that lead to goal attainment (Dewey & Carr,
2013; Dewey, Pacherie, & Knoblich,
2014; van der Wel, Sebanz, & Knoblich,
2012). Yet, the question remains whether experiences of control are affected by the way people represent their own actions and those of their interaction partner.
In essence, the coordination of actions with others is not so different from the coordination of one’s own actions. Indeed, people also referentially code their own actions in reference to their interaction partner, which has repeatedly been demonstrated in a go/no-go task that is commonly used to measure referential coding (i.e., the Simon task). In a standard version of this task participants respond with different actions to different colored stimuli that pointed to the left, middle, or right of the computer screen. For example, participants are required to respond as fast as possible to green stimuli with a left button press and to red stimuli with a right button press. This typically results in a Simon effect (Craft & Simon,
1970; Simon,
1990), i.e., slowed reaction times to stimuli that are incompatible with the spatial referential coding of actions (e.g., responding with a
right button press to a stimulus pointing to the
left). Intriguingly, while there is typically no Simon effect when people have to respond to only one stimulus in an individual go/no-go version of the task (e.g., only responding to green stimuli with left button presses; Hommel,
1996a; Sebanz et al.,
2003), the Simon effect is reinstated when another person or object is associated with the other stimulus (e.g., when one’s interaction partner responds to the red stimuli with right button presses; Sebanz et al.,
2003). This reinstatement of the Simon effect suggests that just like representing their own (e.g., left and right hand) actions in reference to each other, people tend to represent their own actions in reference to other events in their direct environment, such as their interaction partner’s actions when jointly coordinating actions (Dolk et al.,
2014; Dolk, Hommel, Prinz, & Liepelt,
2013; Sebanz et al.,
2003; Tsai, Kuo, Hung, & Tzeng,
2008). To the extent that people represent their own as well as their interaction partner’s actions during action coordination, people may experience control over their own as well as others’ actions that are compatible with these representations.
Specifically, we expect people to experience more control over their own as well as their interaction partner’s actions when these actions (e.g., left key presses) are triggered by a compatible (e.g., left-presented) rather than an incompatible (e.g., right-presented) stimulus. However, as referential coding (Dittrich, Rothe, & Klauer,
2012; Liepelt, Wenke, Fischer, & Prinz,
2011; Vlainic, Liepelt, Colzato, Prinz, & Hommel,
2010) and neural activation (Kilner, Friston, & Frith,
2007; Mukamel et al.,
2010; Veluw & Chance,
2014) are generally stronger for own compared to others’ action performance, we do expect people to experience more control over their own actions compared with their interaction partner’s actions. To test these hypotheses, we employed a joint go/no-go task in which we assessed experiences of control over (1) self-performed actions on go-trials, (2) other-performed actions on no-go trials, and 3) action inhibition on no-go trials.
Conclusion and discussion
In the present study, we investigated whether the internal representation of an interaction partner’s actions would lead to vicarious experiences of control. Although participants did represent their actions in reference to their interaction partner (as demonstrated by a strong main effect of spatial compatibility on reaction times), this joint action representation did not predict experiences of control over their own or their interaction partner’s actions. In fact, Bayesian analyses indicated that our data provided evidence for a null-effect of compatibility on experienced control. This null-finding appears to be inconsistent with research suggesting that experienced control over self-produced actions is strongest when action selection is smooth (i.e., on compatible compared with incompatible trials; Morsella et al.,
2009; Sidarus & Haggard,
2016; Wenke et al.,
2010).
Possibly, we found no effects of compatibility on experienced control because the difference between spatially compatible and incompatible trials was too subtle. That is, in contrast to a recent study by Sidarus (
2016), the source of the response conflict in our study is implicit. Furthermore, although we found a strong compatibility effect in our study, the difference in reaction times for compatible and incompatible trials was only 16.65 ms, or even 12.92 ms if you only consider the second trials over which experienced control was assessed. Yet, reaction times in the studies by Morsella (
2009; non-verbal condition) and Wenke (
2010) differed by 133.4 and 49.9 ms between compatible and incompatible trials, respectively. Hence, even though participants may have been unaware of the source of the response conflict (e.g., incompatible action primes), they may have noticed the resulting differences in their own reaction times and used this information to base their experience of control on. Reaction time differences may thus affect experienced control only if they can be consciously detected by the actor. Conscious detection may, however, be especially relevant when explicitly assessing experienced control. A remaining question for future research is whether people do experience more control on an implicit level as a result of representing one’s own actions in reference to their interaction partner, and whether such an implicit sense of control affects behavior regulation (Steinhauser & Kiesel,
2011), (observational) learning, or feelings of responsibility (Frith,
2013; Haggard & Tsakiris,
2009; Moretto et al.,
2011) for jointly produced actions and outcomes.
Further exploratory analyses did not reveal the typical sequential modulation of the joint Simon effect (i.e., the compatibility effect being larger after compatible versus incompatible trials; Liepelt et al.,
2011) in our adapted version of the joint Simon task, which was regularly interrupted by a question about experienced control following each second trial. It could be that the facilitating effects of stimulus–response repetition are attenuated by the relatively large performance improvements on each second trial in a sequence. Such performance improvements are usually less pronounced as reaction times and sequential effects are typically measured over longer trial sequences, sometimes even excluding performance on the first trial from analyses (e.g., Liepelt et al.,
2013,
2011; Notebaert et al.,
2006; Verbruggen et al.,
2006). More importantly, the presence of a joint Simon effect in the absence of the typical sequential modulation supports the assumption of the feature-integration account that the (social) Simon effect is independent from the integration processes that produce the sequential effects (Hommel et al.,
2004).
It is worth noting that we were able to reliably detect the joint Simon effect even when interrupting the task every other trial with introspective questions. Although reaction times were slower on first trials compared to second trials, the spatial compatibility effect was present on first as well as second trials within a sequence. This opens new possibilities for assessing the consequences of joint action representations within the context of the (joint) Simon task. For example, it might be possible to add implicit measures of experienced control (e.g., such as temporal binding or sensory attenuation; see Hughes, Desantis, & Waszak,
2013) that may be more susceptible to subtle response selection processes. Also, it might be interesting to gain more insight in when and how people learn from the observation of other people’s actions, e.g., by measuring memory for self-performed versus observed reactions to compatible versus incompatible stimuli on a trial level.
In conclusion, in line with the notion that referential coding (Dittrich et al.,
2012; Liepelt et al.,
2011; Vlainic et al.,
2010) and neural activation (Kilner et al.,
2007; Mukamel et al.,
2010; Veluw & Chance,
2014) are generally stronger for own compared to others’ action performance, the present findings indicate that experienced control is higher for self-produced versus other-produced actions. In that sense, experienced control is a matter of you versus me. However, the present study further suggests that the ability to discriminate one’s own and others’ actions by means of spatial coding does not affect explicit experiences of control. Furthermore, we showed that the joint Simon effect is independent from the integration processes that produce typical sequential effects, and that the joint Simon effect remains intact despite frequent interruptions by additional measurements (i.e., introspective questions on experienced control). As such, the present research opens new avenues for future research on the downstream consequences of joint action representation.