Metacontrol and joint action: how shared goals transfer from one task to another?

We often work together with other human beings to reach shared goals trying to improve the team performance. But frequently, we also compete against other people either alone (e.g., in a single tennis match) or together with other people (e.g., in tennis doubles) trying to achieve group goals at the expense of others’ goals. When working in independent groups as in bowling or golf, task characteristics of the group performance can be seen as additive (Steiner, 1972). However, in co-acting (e.g., in paddling) or interacting groups (e.g., in basketball), task characteristics of the group performance are typically seen as conjunctive (Steiner, 1972). The aim of this study is to test if competitive/cooperative processing modes (i.e., metacontrol states) between partners change the degree of bodily self-other integration in co-acting dyadic groups. Sebanz, Bekkering, and Knoblich (2006) defined joint action as any form of social interaction, where two or more people coordinate their behavior in space and time to bring about a change in the environment. To coordinate their behavior, two persons need to form a representation of the other person’s action at some level (Liepelt & Prinz, 2011; Vesper, Butterfill, Knoblich, & Sebanz, 2010). Cognitive science investigated if and to what extent people mentally represent their own and other person’s actions (Dittrich, Dolk, Rothe-Wulf, Klauer, & Prinz, 2013; Dittrich, Rothe, & Klauer, 2012; Dolk, Hommel, Prinz, & Liepelt, 2013; Kiernan, Ray, & Welsh, 2012; Liepelt, Wenke, Fischer, & Prinz, 2011; Sebanz, Knoblich, & Prinz, 2003; Wenke, Atmaca, Holländer, Liepelt, Baess, & Prinz 2011) and other person’s tasks (Klempova & Liepelt, 2016; Sebanz, Knoblich, & Prinz, 2005; Yamaguchi, Wall, & Hommel, 2018, 2019), and how this impacts both persons’ behavior. A prominent paradigm of this line of research is the social or joint Simon paradigm (Sebanz, et al., 2003), in which two people share the standard version of the Simon task (Simon, 1969; Simon, Hinrichs, & Craft, 1970). The classic version of the Simon task is a two-choice task, in which one participant has to decide between two different items of a specific category (e.g., color or form) and to respond by pushing one of two laterally located response keys. This stimulus randomly appears laterally on the left or right side of a computer screen, whereby the stimulus location is task irrelevant. Even though stimulus location is task irrelevant, participants show a stimulus–response (S–R) compatibility effect. This S–R effect typically shows faster reaction times when stimulus and response location correspond (compatible trial), compared to when they do not correspond (incompatible trial). This effect is known as the Simon effect (Simon, et al., 1970; Simon & Small, 1969; Simon & Wolf, 1963). Sebanz et al. (2003) transformed the Simon task into an individual go/no go task, so that the individual has to respond to only one of two possible stimuli, while having to withhold the response to the other stimulus. In their study, Sebanz et al. showed that the Simon effect breaks down when performing this go/no go task alone. In addition, a third condition was added, in which the same Simon task was distributed across two persons, each person doing only half of the task, so that both persons now share the two-choice Simon task. As in the go/no go task condition, each person has to respond to only one of the two stimuli and to withhold the response to the other stimulus, whereby now the co-actor takes over the response to the other stimulus. In this joint-task setting, the Simon effect reemerged, but now across both persons sharing the task. This effect was therefore called the social Simon effect (Sebanz et al., 2003) or joint Simon effect—short the JSE.

The standard Simon effect is typically explained by the assumption that the spatial response dimension (left/right key press) overlaps with the spatial stimulus dimension (left/right). Hence, task irrelevant spatial stimulus features prime corresponding response features decreasing reaction times when the task irrelevant stimulus dimension is compatible with the required response. Due to a conflict between the instructed and the primed response on S–R incompatible trials reaction times are increased, as conflict resolution takes time. The individual go/no-go Simon effect is abolished because there is only a single response and hence, no left/right response dimension is present anymore. Along these lines the Simon effect is re-established in the joint go/no-go Simon task condition (JSE), because a salient spatial response dimension is restored by the other individual controlling the other response key (Liepelt, et al., 2011). Therefore, Sebanz et al. (2003) concluded that other’s action is regarded as similar to ones own action and other’s task is regarded as own task, making the JSE a measure of the co-representation of other person’s action (Sebanz, et al., 2005). Based on studies showing that JSE-like compatibility effects are also established when one of the two human co-actors is replaced by an event-producing object (Dolk, et al., 2013; Puffe, Dittrich, & Klauer, 2017; Stenzel & Liepelt, 2016), for instance a puppet (Müller, et al., 2011) or a humanoid robot (Stenzel et al., 2012; Stenzel, Chinellato, del Pobil, Lappe, & Liepelt, 2013), a sole mechanism of action co-representation accounting for the JSE and JSE-like effects has been questioned (Dolk, et al., 2011, 2014; Klempova & Liepelt, 2016). Based on the theory of event coding (Hommel, Müsseler, Aschersleben, & Prinz, 2001), the referential coding account has been proposed for joint action (Dolk, et al., 2014, 2013). The basic assumption is that late stages of perception and early stages of action planning and control are cognitively represented by the same kinds of perceptual codes (Prinz, 1997, 2015). Given a similarity between internally used and externally activated events (e.g., events produced by another responding human co-actor) an action discrimination conflict arises during the shared go/no-go decision (Liepelt, et al., 2011). According to the theory of event coding (Hommel, et al., 2001), action selection refers to the selection of a given event that is associated with the to be performed action from a pool of all activated event representations (Dolk, et al., 2013). When sharing the joint Simon task, perceived (or imagined) events produced by the co-actor possessing a high similarity to those events used to select one’s own action produce an action discrimination conflict. According to the referential coding account (Dolk, et al., 2013), this conflict can be resolved by changing the selection criteria through increasing the task relevance of discriminating task features. In the case of the joint Simon task, one dominant discriminating feature is location information (left–right feature codes). For example, strengthening the right response coding (of the right co-actor) strengthens the spatial response dimension, and hence increase the dimensional overlap with the task-irrelevant spatial stimulus dimension (Kornblum, Hasbroucq, & Osman, 1990), producing the JSE. Higher perceptual or conceptual similarity between both the co-actors or their given responses may strengthen the discrimination problem and hence the JSE (Dolk & Prinz, 2016).

Even though different accounts for the JSE were proposed, almost all available accounts do agree that the JSE can be seen as a measure of bodily self-other integration (Iani, Anelli, Nicoletti, Arcuri, & Rubinchi, 2011; Tomasello, Carpenter, Call, Behne, & Moll, 2005). The degree of self-other integration seems to depend on the cognitive states of the individuals sharing the task (Colzato, van den Wildenberg, & Hommel, 2013; Colzato, de Bruijn, & Hommel, 2012; Colzato, Zech, Hommel, Verdonschot, van den Wildenberg, & Hsieh, 2012).

Hommel and Wiers (2017) proposed that cognitive states related to persistence or flexibility directly affect action control (for an earlier account of cognitive control state theory see Goschke, 2003). According to Hommel (2015), differences in cognitive states can be characterized by more (or less) top-down influence of the current action goal leading to less (or more) self-other integration, and strong (or weak) mutual competition between alternative action representations (Ma & Hommel, 2018). Having a task goal related to persistence should therefore lead to less self-other integration, while shared task goals related to flexibility should lead to more self-other integration. The present study tested if shared goals (competition vs. cooperation) two persons have while sharing one task are transferred to a new joint Simon task. Further, and more exploratory, we tested whether shared group experiences shape a potential goal transfer.

The bodily self-other integration process (Colzato, et al., 2013; Liepelt et al., 2012) has been shown to depend on the relationship between both co-actors sharing the task (Hommel, Colzato, & van den Wildenberg, 2009; Iani, et al., 2011; Mendl, Fröber, & Dolk, 2018; Ruissen & de Bruijn, 2016; Ruys & Aarts, 2010). When co-actors were friendly, inviting for cooperation, their actions were more strongly integrated than actions of intimidating and unfriendly co-actors (Hommel, et al., 2009) indicating that self-other integration is sensitive to the social relationship between co-actors. Ruys and Aarts (2010) aimed to investigate the independence between actors sharing the joint Simon task. When only the best subject won a reward, the JSE was decreased as compared to a condition in which the team won a reward (cooperation) and when a random selection of team winners earned a reward (competition). This study suggests that interdependency leads to more self-other integration. Iani et al. (2011) argued that cooperation is directly implied by the positive interdependence as two persons sharing a task need to work together for attainment of a common goal. On the other hand, competition is implied by negative interdependence, as both co-actors work against each other for the attainment of a personal goal (see Iani, et al., 2011, p. 442). Further, they argued that the random winner selection used in the competition condition in the study of Ruys and Aarts (2010) was not satisfying, as competition is typically defined as a condition in which one individual attempts to outperform another individual in a zero-sum situation (Iani, et al., 2011). Therefore, Iani et al. (2011) more directly manipulated cooperation vs. competition by instructing participants that either the best couple would receive an extra reward (to induce cooperation) or that the best participant of a couple would receive an extra reward (to induce competition). When performing the joint Simon task together with a person with whom the reward was shared (cooperation), the JSE was increased as compared to a situation, in which the task was played against another person (Iani, et al., 2011) and only the best person won the reward (competition). Combining this logic with an analysis of the sequential modulation of the JSE (Liepelt, Wenke, & Fischer, 2013; Liepelt, et al., 2011; Yamaguchi, Wall & Hommel, 2017), Mendl et al. (2018) found the difference between cooperation and competition conditions only for actor repetition trials, not for trials in which the actor switched. This suggests that processing adjustments primarily affected the processing of one’s own actions. When competing with another person, both actors seem to focus more on their own action leading to less self-other integration. Ruissen and de Bruijn (2016) decoupled the cooperation-competition manipulation from the joint Simon task by letting participants play Tetris before running the actual Simon task. Couples who cooperatively played Tetris with each other showed an increased JSE as compared to couples who competed in the Tetris game. This finding is in line with studies showing that self-construal priming drawing attention to independence reduced the JSE relative to a task drawing attention to interdependence (Colzato, de Bruijn, et al., 2012; Colzato, Zech, et al., 2012). Evidence for a carry-over effect of competition was shown by Iani, Anelli, Nicoletti and Rubichi (2014). In their study, pairs of participants performed a joint Simon task before and after a joint Flanker task. In their first experiment, the joint Simon task was performed under neutral task instructions, while participants performed the joint Flanker task either under cooperation or competition. In the second experiment, participants were required to compete in the joint Flanker task and to cooperate during the subsequent joint Simon task. While the competition effect transferred from the joint Flanker task to the joint Simon task indicated by an absence of a joint Simon task in their first experiment, prior competition did no longer affect joint Simon task performance when a new cooperation goal was introduced in the joint Simon task (Iani, et al., 2014). Cooperation goals and competition goals seem to affect the performance in a joint Simon task and the findings of Iani et al. (2014) indicate that competition seem to extend beyond a specific setting affecting the following interaction. These manipulations, playing Tetris, self-construal priming, or a competition induction, changed the task goal of both persons who are sharing the task, which may explain observed changes in the degree of bodily self-other integration (Colzato, Zech, et al., 2012).

To induce a cooperative goal, we adopted the manipulation of Iani et al. (2011), instructing pairs of participants that the best couple will win an extra reward. During the instruction, we did not specify what participants would receive as the “extra reward”. The winner could choose from a random selection of sweets. This condition was contrasted to a condition in which the best individual of each pair will win, earning an extra reward (same type of reward), thereby setting participants in a competitive task mode. However, in contrast to most previous studies testing the impact of cooperation and competition directly in the joint Simon task (Iani, et al., 2011; Ruys & Aarts, 2010) or between two different compatibility tasks (Iani, et al., 2014) showing evidence for a carry-over effect of competition, we induced competitive or cooperative task goals in a separate dyadic game (Colzato, et al., 2013; Ruissen & de Bruijn, 2016) called “Mindflex”. This was done to set participants either in an abstract higher-level competitive or cooperative cognitive state and test the impact of the respective state on bodily self-other integration in a separate task (i.e. abstract goal transfer). The effect of goal transfer was tested with the joint Simon task. Further, and more exploratory, we manipulated whether goal induction was performed alone or together with the partner (i.e. jointly) testing if a possible goal transfer depends on a shared group experience.

If shared competitive task goals can be transferred from one task to another, we predict less self-other integration when both participants are previously set to a competition goal compared to a cooperation goal (JSE_{_competition goal} < JSE_{_cooperation goal}). More exploratory, we tested if this effect depends on shared group experience, which may predict competition goal transfer to be stronger when participants experience the induction together compared to being alone.

The joint Simon effect is considered as a marker for bodily self-other integration in joint action (Colzato, Zech, et al., 2012). This study tested two questions with regard to self-other integration. First, does a competitive processing mode (i.e. metacontrol state) between partners change the degree of self-other integration in joint action? Second, and more exploratory, does competition transfer depend on a shared group experience with a joint action partner?

The data of our manipulation check indicates that the induction of the different processing modes (competition and cooperation) effectively worked via shared game experience. Our results showed that a competitive processing mode produces less bodily self-other integration than a cooperative processing mode (Hommel, et al., 2009; Iani, et al., 2011, 2014) as indexed by a transfer effect to a Joint Simon Task that was performed after the state induction. In line with previous studies (Iani, et al., 2014; Ruissen & de Bruijn, 2016) our findings indicate that the specific metacontrol state a dyadic pair was set to (i.e., competition or cooperation) was transferable to a fully new dyadic task. The transferable metacontrol state can therefore be considered as an abstract control policy acting on a lower level process of bodily self-other integration. The finding that less bodily self-other integration is observed under a competitive processing mode than under a cooperative processing mode (Hommel, et al., 2009; Iani, et al., 2011, 2014) in a consecutively performed joint Simon Task is in line with findings from social psychology (Bargh & Chartrand, 2000; Sassenberg, Moskowitz, Jacoby, & Jansen, 2007). These studies show that participant’s seem to behave in line with a primed mindset even in a new and task unrelated context. Our first question can, therefore, be clearly answered with yes. A competitive processing mode (i.e. metacontrol state) between partners reduces the degree of bodily self-other integration as compared to a cooperative processing mode. Even though the joint Simon effect was significantly reduced following a competitive induction, it was nevertheless present after competition with the same co-actor. This is different to findings of Iani et al. (2014) showing a complete absence of the joint Simon effect after having performed a joint Flanker task in competition. One difference between our study and the study of Iani et al. (2014), which could explain this difference, is the similarity between the induction task and the transfer task applied. The joint Simon task is a spatial compatibility task and the joint Flanker task is non-spatial in nature. However, these two tasks have more similarities than the Mindflex task of our study and the joint Simon task. The joint Simon and the joint Flanker task are both S–R compatibility tasks with overt spatially distributed manual responses (the joint Flanker task containing an additional S–S compatibility effect). Opposed to this, our Mindflex task is much more abstract and mental in nature possessing much less overlap with the joint Simon task. Applied task sets that are activated or deactivated by the respective goal will therefore probably lead to a much more complete “competition transfer” in the study of Iani et al. (2014) than in ours. This speculation could explain why the reduced joint Simon effect following a competitive induction was still significant in our study. Furthermore, based on the findings of Iani et al. (2014), the conclusion that the metacontrol transfer, which we observed in our study, might be better explained with a transfer of competition (not a transfer of cooperation) seems justified.

In contrast to the competition transfer that we observed in the joint Simon task, the competitive state vs. cooperative state induction did not affect the subjective conceptual self-other relation measured with the IOS scale (Aron, et al., 1992). In a recent review (Quintard, Jouffre, Hommel, & Bouquet, 2020), the impact of romantic love and corresponding positive affects due to romantic feelings have been proposed to bias metacontrol states towards flexibility vs. integration also affecting overlapping bodily self-other representations. In line with this proposal, we found evidence for such metacontrol biases on bodily self-other representations (i.e. the JSE). However, the short-term competitive state vs. cooperative state induction that we used to implement different metacontrol state biases, did not affect the conceptual self-other relation between the two co-actors in a way that would be predicted with long-term varying degrees of romantic love and passion felt for the partner as shown in couples (Quintard, Jouffre, Croizet, Bouquet, 2020). This seems to suggest that the degree of bodily self-other representations may be more flexible and easier to change online than conceptual self-other relations, as the latter have been shown to be the product of long-term experiences and long-lasting relations (Quintard, Jouffre, Croizet, et al., 2020). The different degrees of bodily self-other representations varied through different types of metacontrol states, as done in our study, seem to be sensitive to the current online interaction situation. On the other hand, changing degrees of conceptual self-other relation varied through different degrees of passion and romantic love, may be more strongly related to the type of the long-term interactive social relation between persons.