The phenomenology of controlling a moving object with another person

Highlights

• We studied the sense of control during a cooperative joint action.

• The impact of the co-actor’s contribution was task dependent.

• Individual control mattered most when agents performed similar actions.

• Control at the group level mattered most when agents performed complementary actions.

• The sense of control is evaluated at the group level when both agents are pivotal.

Abstract

The phenomenology of controlling what one perceives is influenced by a combination of sensory predictions and inferential processes. While it is known that external perturbations can reduce the sense of control over action effects, there have been few studies investigating the impact of intentional co-actors on the sense of control. In three experiments, we investigated how individuals’ judgments of control (JoC) over a moving object were influenced by sharing control with a second person. Participants used joysticks to keep a cursor centered on a moving target either alone or with a co-actor. When both participants’ actions had similar perceptual consequences, JoC ratings were highest when self-generated movements were the only influence on the cursor, while the appearance of sharing control with a second person decreased JoC ratings. By contrast, when participants performed complementary actions with perceptually distinctive consequences, JoC ratings were highest when both participants were able to influence the cursor. The phenomenology of control during joint action is influenced by low-level visuomotor correlations, the presence of competing causal influences, and group-level performance.

Introduction

To control something is to act in order to bring it to a pre-specified condition, possibly in the face of external forces or changes in the environment that tend to alter it (Powers, 1978). This broad definition encompasses most purposeful human behavior, as voluntary actions are usually performed with the intent of producing a particular change in the environment that can be perceived as a sensory outcome of performance (henceforth “action effects”).

The question of what processes contribute to the phenomenology of controlling what one perceives has motivated much research. The emerging consensus is that the sense of control is not a unitary phenomenon, but rather depends on a combination of efferent motor signals, sensory predictions and higher level cognitive processes (Haggard and Tsakiris, 2009, Pacherie, 2008, Synofzik et al., 2008). The sense of control over body movements is thought to depend on a system of sensorimotor comparators which detects discrepancies between sensory predictions triggered by efferent motor signals, and actually executed movements (Blakemore et al., 1998, Frith, 2012, Tsakiris et al., 2005). Sensory predictions also influence the sense of control over distal events outside the body. For example, auditory stimuli triggered by keystrokes are more likely to be attributed to external sources when the timing or frequency is different from what was expected (Knoblich and Repp, 2009, Sato and Yasuda, 2005).

The sense of control can also be influenced by inferential processes. For example, priming unintended action effects has been shown to increase feelings of authorship, which suggests one’s own causal role may be inferred post hoc from the match between a prior mental state and a subsequent action effect (Aarts et al., 2005, Moore et al., 2009, Sato, 2009, Wegner and Wheatley, 1999). The sense of control may be further modulated by the fluency of action selection (Haggard & Chambon, 2012), and by the magnitude and valence of action effects (Aarts et al., 2006, Kawabe, 2013).

Although there has been progress in understanding the mechanisms which contribute to a sense of control, the types of task environments that have been studied are limited in scope. Many investigators have focused on the phenomenology of causal initiation, i.e. the sense of agency. In these experiments, participants are typically asked to rate their agreement that a brief event such as a tone (e.g. Engbert et al., 2008, Sato and Yasuda, 2005), the sudden appearance of a visual stimulus (e.g. Linser and Goschke, 2007, Sato, 2009), or the sudden stopping of a previously moving stimulus (e.g. Aarts et al., 2005, Jones et al., 2007, Wegner and Wheatley, 1999) was caused by their own prior action. Although interesting in its own right, causal initiation does not guarantee that an entire action will be experienced as controlled (Pacherie, 2007). One may initiate an event, but lose control as it unfolds over time, as for example when one loses control of an automobile while driving. Yet there have been relatively few studies investigating the sense of control for events lasting longer than a few milliseconds (but see Dewey et al., 2010, Metcalfe and Greene, 2007).

Another limitation of research in this area has been the focus on individuals performing tasks in isolation. In everyday life people often act in a social context, performing joint actions with others. Joint action can be defined as a social interaction where individuals coordinate their actions to bring about a change in the environment (Sebanz, Bekkering, & Knoblich, 2006). There are some studies which have investigated the sense of agency and related processes in social contexts (e.g. Desantis et al., 2012, Dewey and Carr, 2013, Obhi and Hall, 2011a, Obhi and Hall, 2011b, Wegner and Wheatley, 1999). Typically, participants perform a task either with or without a partner, but only one agent actually controls the stimulus at any given time. By contrast, we were interested in situations in which two actors share control. For example, consider white water rafting with a group. In this scenario, the motion of the raft is jointly determined by several people working together with a more or less common purpose, plus some unpredictable perturbations caused by the water currents. In a situation like this, what are the implications for the individual’s sense of control? Can people distinguish their own contributions from the contributions of their co-actors? Are the contributions of co-actors perceived as perturbations that reduce the individual’s sense of control? Or do the contributions of co-actors increase the individual’s sense of control by facilitating attainment of shared goals? To begin addressing these questions, in the present study we investigated the phenomenology of control during a cooperative joint action lasting several seconds.

A fundamental question is whether the sense of control during cooperative joint actions engages the same processes which shape the sense of control during individual action, or if it is in some sense a special case. One possibility is that the sense of control is essentially egocentric, depending on the perception of a causal relationship between one’s motor inputs and the perceived action effect. In this case the contributions of a co-actor might be experienced as external perturbations if both agents tried to manipulate an object at the same time. On the other hand, if each agent’s contribution was perceptually distinctive (for example, the two agents take turns manipulating an object), the egocentric hypothesis predicts that the co-actor’s actions would have little impact on the sense of control.

An alternative to the egocentric hypothesis is that the contributions of a cooperative co-actor might increase the sense of control due to the agents’ shared intentions. There is evidence that when individuals feel themselves to be part of a group, this can influence action-perception links, including response times (Tsai, Sebanz, & Knoblich, 2011), the perceived timing of actions and their effects (Obhi & Hall, 2011a), and the sensory attenuation of effects generated by another person (Weiss, Herwig, & Schütz-Bosbach, 2011). A catch-all term for these effects of shared intentionality is the “we-mode” (Gallotti & Frith, 2013). Cognition in the we-mode might lead individuals to evaluate control at a group level, e.g. based on the success of the joint action. We will refer to this as the joint control hypothesis. With this background, we considered three non-exclusive ways in which actions performed by agent B might influence agent A’s sense of controlling the action effect during a cooperative joint action.

Action effect predictability. When individuals perform tasks alone, the sense of control is influenced by congruence between intended, predicted, and actually perceived action effects. Thus, one way agent B could influence agent A’s sense of control is by altering the objective correlation between agent A’s motor inputs and the action effect, whether positively or negatively (Fig. 1a). This could be characterized as an impact at the level of egocentric sensory predictions. There could also be an impact of action effect predictability at a perceptual level that does not depend on motor signals. In that case, predictable contributions from agent B might increase agent A’s sense of control even if they did not correlate with agent A’s motor inputs. The latter possibility would be consistent with the joint control hypothesis.

Performance cues. A second possibility is that agent B’s contribution to a joint action could modulate agent A’s sense of control by causing the joint action to be more or less successful (Fig. 1a). Positive outcomes can lead to illusions of control, particularly when people are led to believe the outcome is skill dependent (Langer, 1975). For example, acquisition of a goal can influence judgments of control over moving objects, even leading individuals to overlook minor discrepancies between predicted and observed action effects (Dewey et al., 2010, Metcalfe and Greene, 2007). Performance cues can also have a significant impact on the sense of control during joint action tasks (Van der Wel, Sebanz, & Knoblich, 2012). A caveat to this point is that participants in these studies were simply asked to rate their feelings of control (or agency) without specific instructions as to what that meant. So it unclear whether participants perceived that they contributed more on the more successful trials, or simply felt that having success is part of what it means to be in control. With respect to the question of whether the sense of control is egocentric, an important question is whether individuals’ sense of control is influenced when a joint action either succeeds or fails due to the co-actor’s contribution.

Causal discounting. Causal discounting is a phenomenon in causal reasoning where the presence of one causal explanation casts doubt on another (Einhorn and Hogarth, 1986, Khemlani and Oppenheimer, 2011). Applying this idea to the phenomenology of voluntary action, Wegner and Wheatley (1999) proposed that perceiving one’s own thoughts or actions as the exclusive or most probable cause of an external event is a necessary precondition for the sense of agency. Thus, a third way agent B could influence agent A’s sense of control is by leading agent A to infer that he or she has less control compared to when agent A performs the task alone (Fig. 1b). If causal discounting had a strong influence on the sense of control during joint action, this would tend to support the hypothesis that the sense of control is egocentric, as it implies that agent B’s control comes at the expense of agent A.

Broadly, the goal of the present study was to characterize how the sense of control during cooperative joint actions is influenced by contributions (real or imagined) from a co-actor. Of particular interest was the question of whether the sense of control is essentially egocentric, depending on cues which indicate a direct a causal link between one’s own motor inputs and the action effect, or alternatively, if the sense of control depends more on performance at the group level.

To disentangle egocentric sensory predictions from group performance, we varied the perceptual distinctiveness of two agents’ contribution to a jointly controlled action effect. We considered two scenarios. In the first, agents A and B performed the same action at the same time, which made their respective contributions to the action effect difficult to distinguish. A real-world example would be two people sitting in a boat and paddling in synchrony. In the second scenario, A and B took turns performing complementary but different actions, which made their unique contributions easier to distinguish. A real-world example would be two people sitting on opposite sides of the boat, one side being responsible for making left turns and the other side for right turns.

The reasoning behind this manipulation was that if the sense of control is egocentric, then agent A’s sense of control should be modulated by agent B’s contributions in the perceptually ambiguous scenario but not in the perceptually distinctive scenario. However, if the sense of control depends on control at the group level, then increasing agent B’s control should increase agent A’s sense of control if it results in a better outcome, even in the perceptually distinctive scenario. This assumes that both agents’ actions are predictable and serve a common goal.

To implement and test the two scenarios just described, we devised a visual tracking task in which participants used joysticks to keep a cursor centered on a moving target. The joysticks could be turned on or off independently, so that neither, one, or both participants had some control over the cursor. To add further variability to performance and the sense of control, random perturbations were added to the cursor during some but not all trials. We hypothesized that individuals’ sense of control would be influenced by a combination of action effect predictability, causal discounting, and group performance. We further hypothesized that the relative impacts of egocentric sensory predictions and group performance would depend on the perceptual distinctiveness of each agent’s contribution.

In Experiment 1, we tested the scenario in which both agents’ contributions to the joint action were perceptually similar. Here, we predicted the agents’ actions would overlap and interfere with one another, so individuals’ sense of control would be highest when their own joystick was the only influence on the cursor. The causal discounting hypothesis further predicted that the mere appearance of sharing control with another would decrease the sense of control, even if the co-actor had no actual influence on the outcome, because the co-actor would be perceived as a competing causal explanation.

In Experiment 2, we tested the scenario in which agents’ contributions were complementary and perceptually distinctive. Here, we predicted that the co-actor’s contribution would no longer be perceived as interference, so individuals’ sense of control would be highest when both agents’ joysticks were able to influence the cursor. In line with the joint control hypothesis, this would demonstrate that the sense of control during joint action can transcend egocentric sensory predictions, in such a way that the sense of control aligns with the performance of the group as a whole. Although it has previously been shown that the sense of control is influenced by performance when control is ambiguous, complementary joint actions represent an interesting case for further study for at least two reasons: (1) the presence of a co-actor may bring causal discounting into play; and (2) it is unclear whether the performance of a co-actor should influence the sense of control when the agents’ respective contributions are perceptually unambiguous.

In Experiment 3, we replicated Experiment 2, and also compared judgments of self control to judgments of the co-actor’s control. The joint control hypothesis predicted symmetry between the two sets of control ratings, i.e. a shared sense that “we” do or do not have control. An alternative hypothesis would be that individuals show a self-serving bias to take more credit for successful outcomes while attributing less control to the co-actor.