External commands influence the sense of agency
The present research explored whether the commands by another agent influence perceptions of one‘s own sense of agency. Using different measures, we found that obedience or acting compatible to a verbal command decreased the sense of agency. To some degree, these effects were to be expected as earlier research has shown that individuals experience less responsibility after forced compliance (Festinger & Carlsmith, 1959), and personal low power will lead to less agentic global attributions (Keltner, et al., 2008). However, the present research extends these findings by showing that even our immediate action experiences are influenced by acting upon commands (Study 2.1); that commands influence action related time perception (Studies 2.2 & 2.4); are even able to do so when individuals have already
started their action (Study 2.4); and that such effects are likely to occur not because of a limitation in choice opportunity, but are likely due to the contextual presence of an external command (Study 2.3).
Study 2.3 suggests that the reduction due to obedience was mostly due to the presence of the command itself, and is not the consequence of a removal in choice opportunity. As a recent study has shown that a reduction in the number of action alternatives is related to a decrease in the sense of agency (Barlas & Obhi, 2013), a reduction in the number of action alternatives to one possible action – essentially what a command does – could have been the process underlying the reduction in agency due to obedience. However, the findings from Study 2.3 suggest this is not the case: commands that did not alter choice opportunity led to lower agency levels, just as commands that did limit the number of action opportunities. This suggests this process is not due to the presence or absence of action alternatives, but that it is the presence of the command itself that influences the sense of agency.
The findings from Study 2.4 further support this explanation. Here, commands did not relate to choice opportunity as the commands were given after participants‘ actions had already been initiated. The results showed that these commands were again able to influence the sense of agency, depending on their compatibility with the ongoing action: verbal commands that were compatible with the action reduced agency compared to unrelated or opposite commands. While Moore et al. (2009) showed that supraliminal effect primes (a preview of a tone), increased agency when the effects were compatible rather than incompatible (for similar findings see Aarts et al., 2005), the present research shows that the presence of action related commands decreases the sense of agency when those commands are compatible. The findings from Study 2.4 also expand on a recent study by Damen and colleagues (2014), who showed that compatible supraliminal action primes given prior to action performance lead to lower agency ratings compared to incompatible primes. The present findings show that this process can also occur after action initiation. The conscious awareness that we may have been influenced or are actually doing something in line with the suggestions or commands of another will reduce our sense of agency (Damen et al., 2014). Imagine driving and taking a left turn, and someone in the passenger seat tells you to keep going left. You are then likely to experience decreased agency for taking the left turn compared to when the passenger was actually telling you to take a right turn instead.
Since previous literature has argued that explicit judgments and implicit feelings of agency reflect different levels of agency (Synofzik et al., 2008), both concepts of agency were investigated by using both explicit and implicit measures. In general, our results show that commands influence agency at both levels; receiving a command affects both our immediate feelings of agency, as well as our reflective judgments of being a likely agent.
However, the question remains why external commands would actually influence time perception, an implicit measure that has been considered to reflect a more primary or lower level of agency assumed to be unaffected by contextual cues (unrelated to motor prediction). Given the present findings, the most likely explanation is that processes of motor prediction can in fact be influenced by contextual cues: If you suddenly receive a voice command during your motor preparation, as in the present studies, it is not hard to imagine that you will be distracted and subsequently your sense of agency becomes reduced. Such an explanation would be in line with a number of recent findings that indeed challenge the view that motor prediction processes are inoculate to inferential cognitions and contextual cues, and suggest that factors such as belief states and contextual cues may have a much more fundamental effect than previously thought. For example, in a study by Rigoni and colleagues (2012), individuals‘ belief in free will and personal causality was shown to influence the degree of pre-motor activation in the brain; individuals who disbelieved in free will showed a reduction in readiness potential amplitudes. In similar vein, the readiness potentials can perhaps also be influenced by the presence of a verbal command; an interesting avenue to investigate in future research.
A second - and intriguing - explanation is that inferences and beliefs, perhaps also influenced by contextual cues, can influence feelings of agency and intentional binding more directly then previously considered. Desantis and colleagues (2011) showed that inducing participants with high beliefs about personal agency led to increased intentional binding compared to individuals who were led to believe their actions were caused by another individual. However, such an effect may still be caused by a difference in motor prediction. As mentioned above, Rigoni and colleagues (2012) have shown that beliefs about free will did also influence pre-motor activity. However, the results from Study 2.4 open the possibility of another explanation: that of a direct effect of contextual inference on time perception.
Previous studies in the domains of agency and binding have regularly required their participants to perform a single short action (e.g., a button-press) and to observe a single event (e.g., a tone). Study 2.4 is, to our knowledge, the first study to show that time estimation can be influenced on a task involving an ongoing action. The context of an ongoing action and over several seconds, makes it difficult to explain the present findings as an effect related to motor prediction. While one may reasonably argue that in Studies 2.1-2.3 ones motor preparation is disrupted due the context of a verbal command, and therefore the amount of sensory information the participants are receiving during the motor preparation period is different for the separate conditions, such an explanation is less likely to have occurred in Study 2.4. The results from Study 2.4 therefore suggest an effect that is inferential rather than related to motor prediction.