Intended emotions influence intentional binding with emotional faces: Larger binding for intended negative emotions

https://doi.org/10.1016/j.concog.2021.103136Get rights and content

Highlights

  • The effect of emotional content on intentional binding (IB) is equivocal.

  • We explicitly use intentions to study the effect of emotional content on IB.

  • Negative intended emotions show a larger binding than positive intended emotions.

  • This emotion effect is present only with intentions and not with instructions.

  • Prospective mechanisms potentially underlie the intended emotion effects on IB.

Abstract

The effect of emotions on Intentional binding (IB) is equivocal. In addition, most studies on IB have not manipulated emotional content of intentions. This study investigates the effect of intended and outcome emotions using emotional faces (happy or disgust face in experiment 1 and a happy or angry face in experiment 3). To see whether the effects are due to priming, we used instructions with a happy-disgust pair in experiment 2 and happy-angry pair in experiment 4. Outcome emotional faces were not predictable. Results showed that intending a negative emotional face resulted in shorter action-outcome interval judgments compared to a happy face irrespective of the emotional content of the outcome face. This effect was absent in experiments 2 and 4 with instructed emotions. In addition to showing the importance of having explicit intentions, the results show that emotional content of our intentions does influence IB possibly due to prospective mechanisms.

Introduction

Sense of agency is a feeling that we are agents who voluntarily perform actions (Moore, 2016). Sense of agency is described as taking control of one’s action, or rather it is the pre-reflective sense that I am in control or the author of my thoughts and actions (Gallagher, 2000). The sense of agency has not only been studied with explicit measures but also with implicit measures like intentional binding (IB: Haggard, Clark, & Kalogeras, 2002). IB is defined as the reduction in estimated time between a voluntary action and its outcome. IB is influenced by many factors including temporal contiguity and action-outcome contingency (Cravo et al., 2011, Haggard, Clark, & Kalogeras, 2002, Moore et al., 2009a) and has been explained using predictive or postdictive mechanisms (Synofzik, Vosgerau, and Voss, 2013).

We perform actions based on consequences we intend to achieve. Intentions are mental representations of the goals, desires or consequences we want to achieve through voluntary movements (Antusch, Aarts, & Custers, 2019). Previous studies have shown that our perception of the outcome depends on whether the outcome was intended or not (Fereday et al., 2019, Makwana & Srinivasan, 2017) pointing to the important role of intentions prior to actions on subsequent perception.

An important outcome of our actions in a social context is emotion elicited in others, given that humans communicate significantly through emotional expressions (Adolphs, 1999). We speak or perform actions with the expectation that specific emotions will be elicited in individuals with whom we interact. These emotional expressions elicited in others enable us to understand the mental state of others, and whether our actions produced to elicit specific emotions have succeeded or not. The effectiveness of our actions in eliciting intended or unintended emotional outcomes in others can be expected to influence our sense of agency and hence IB for emotions and emotional expressions.

The effect of emotions on IB has been studied using auditory emotional stimuli (Christensen et al., 2016, Moreton et al., 2017, Yoshie and Haggard, 2013, Yoshie and Haggard, 2017) and visual emotional stimuli (Moreton et al., 2017). Some studies have investigated the effect of mood on intentional binding (Christensen, Di Costa, Beck, and Haggard, 2019).

In one of the first studies on the effect of emotions on IB (Yoshie & Haggard, 2013), participants viewed a continuously rotating clock and pressed a key to produce a sound after a delay of 250 ms. Participants then judged where the clock hand was at the onset of their key press, and in a separate block where it was at the onset of the sound. Judgement error was calculated for the actual and judged time of the event. Positive judgement error indicated for delayed judgement and negative judgement error indicated for anticipatory judgement. Each key press was followed by a positive (2 achievement and 2 amusement sounds) or a negative sound stimulus (2 fear and 2 disgust sounds). IB was more for positive emotional outcomes compared to negative emotional outcomes.

Further studies have looked into whether the effects of emotional valence on IB are due to predictive or retrospective mechanisms using auditory emotional stimuli (Christensen et al., 2016, Yoshie and Haggard, 2017). Yoshie and Haggard (2017) manipulated the predictability of the emotional outcome. IB was larger for predictable positive emotional vocalizations compared to predictable negative emotional vocalizations. There was no difference in IB for positive and negative emotional vocalizations, when they were unpredictable. The results were taken as evidence that the effect of emotional content of the outcome on IB is based on predictive mechanisms. In another study (Christensen et al., 2016), predictability and emotional valence was manipulated to measure retrospective and prospective components of sense of agency. They found larger retrospective IB for positive emotional stimuli when the emotional outcomes were unpredictable. When they became predictable, both positive and negative emotional stimuli showed lesser binding than neutral stimuli. When a neutral stimulus was expected, the prospective component of binding was lesser for both positive and negative outcome stimuli compared to the neutral stimulus.

A replication attempt with sufficient power failed to replicate the effect of emotional outcomes on IB (Moreton et al., 2017). The study involved different kinds of emotional stimuli including emotional auditory vocalizations and faces. The second experiment of this study involved two separate experimental blocks of each emotions i.e. one being positive and the other negative, so overall the action-effects were predictable in nature. They used three different A-O intervals, that is, A-O interval between a key press and faces appearing on the screen (100 ms, 400 ms, and 700 ms). They used 36 disgust faces and 4 fearful faces. They did not find any effect of emotional outcomes on IB. Experiment four of the same study tried to directly replicate the Yoshie and Haggard (2013) to investigate the replicabililty of the original finding. Similarly, the temporal binding was not significantly modulated by positive and negative sound vocalization as action outcomes. Overall, this study indicated that emotional content of the outcome may not influence IB.

A recent study investigated action-outcome binding as a function of freedom of choice, predictability and emotional valence of the outcome (Tanaka & Kawabata, 2021) using the Libet’s clock task. The study manipulated choice by asking participants to press one of eight available keys (free choice) or press the key as instructed (no-choice). In experiment 1, when the outcome was unpredictable, they found larger action shift for negative compared to positive outcome irrespective of nature of choice. However, for outcome shift, the shift was larger for negative outcome only when the choice was free. In experiment 2, larger binding was seen for negative outcomes when they were unexpected.

To summarize, the results of IB studies manipulating emotional outcomes, predictability and choice have been equivocal. The results range from a lack of effect of valence (Moreton et al., 2017), larger binding for positive outcomes under certain conditions (Christensen et al., 2016, Yoshie and Haggard, 2017) and larger binding for negative outcomes under certain conditions (Tanaka & Kawabata, 2021). Given these conflicting results, there is a further need to investigate IB with emotional stimuli. One significant issue with studies on IB especially those investigating the effect of emotions on IB is the lack of “intentions” (Makwana & Srinivasan, 2017). The emotional expression in a given trial is not intended per se (Christensen et al., 2016, Moreton et al., 2017, Yoshie and Haggard, 2013, Yoshie and Haggard, 2017). Agency depends both on the process of action choice as well as actual cause of action outcome (Frith, 2014). In our study, participants were asked to intend (experiments 1 and 3) and perform an appropriate action to obtain the intended emotional outcome. Second issue is the lack of control with respect to the positive and negative emotional stimuli. It is not clear whether the emotional stimuli in some of the earlier studies (Yoshie and Haggard, 2013, Gentsch et al., 2015; Christensen et al., 2016; Moreton et al., 2017) were matched for perceived emotional intensity or arousal, and even with a recent study doing a pre and post rating of stimuli for each subject (Tanaka & Kawabata, 2021). Third, the negative emotional stimuli were mixed, consisting of different negative emotions (Moreton et al., 2017, Tanaka & Kawabata, 2021, Yoshie and Haggard, 2013, Yoshie and Haggard, 2017). For example, in experiment 2 of an earlier study (Moreton et al., 2017), negative stimuli consisted of disgust and fearful faces.

Given that most of the earlier studies have used auditory emotional stimuli, we decided to focus on visual stimuli: emotional faces. We paired happy faces with either disgust faces (experiments 1 and 2) or angry faces (experiments 3 and 4). Most studies on emotions and IB do not ask participants to explicitly form intentions regarding the emotional content of the outcome (Moreton et al., 2017, Tanaka & Kawabata, 2021, Yoshie and Haggard, 2013, Yoshie and Haggard, 2017). While the Tanaka and Kawabata (2021) study had freedom of choice, their choice did not correspond to the emotional content of the outcome. To fully understand whether intentions influence IB in the context of emotions, we explicitly told participants to intend (Makwana & Srinivasan, 2017) an emotional expression and press an appropriate key to indicate the intended expression in experiments 1 and 3. To ensure that any effects on IB due to emotional expressions are a consequence of forming intentions (experiments 1 and 3), in experiments 2 and 4, participants were instructed to press the key appropriate for the emotional expression shown on the screen by the experimenter. In all the experiments, A-O interval was varied and participants verbally reported the A-O interval between their action and the outcome emotional face.

The relationship between intention and outcome was kept unpredictable because earlier studies with auditory stimuli either did not find any effect for emotional valence when stimuli are unpredictable (Yoshie & Haggard, 2017) or showed that the nature of action choice seems to matter (Tanaka & Kawabata, 2021). In addition, we also used verbal estimates instead of the Libet’s clock task unlike most of the earlier studies (Christensen et al., 2016, Moreton et al., 2017, Tanaka & Kawabata, 2021, Yoshie and Haggard, 2013, Yoshie and Haggard, 2017).

We hypothesized having an explicit intention would matter and there would be an effect of intended emotion on IB. However, we were not sure whether having an intention would mean greater or lesser IB for positive or negative emotions. In terms of actual emotional content of the outcome, we did not have any specific hypothesis given the conflicting results with emotional valence in the previous studies (Moreton et al., 2017, Tanaka & Kawabata, 2021, Yoshie and Haggard, 2017). Given that we asked participants to explicitly intend an emotional expression, the design would allow us to see the effects of intended emotion separately from the effects of emotional outcome. If emotional content of the outcome influences IB, then this would constitute evidence for retrospective mechanisms and if only emotional content of the intention influences IB, then this would constitute evidence for prospective mechanisms.

Section snippets

Experiment 1

The first study focused on the effects of intended and actual emotional happy and disgust faces. We chose disgust faces because an earlier failed replication used a large number of disgust faces (Moreton et al., 2017). In addition, disgust sounds have been used in earlier studies (Yoshie & Haggard, 2017).

Experiment 2

Experiment 1 showed an effect of intended emotion. It is important to establish that the effect is due to intentions and not simply due to passive activation of outcome representations. Hence, we used instruction-based actions in this study. The action and outcome phase was the same as in experiment 1. If the effect of intended emotion is due to activation of representations specific to an intention, then there would be no effect of instructed emotion in this experiment.

Experiment 3

Experiment 1 used happy and disgust emotional faces. It is important to replicate the effect with a different emotion given discrepancies in earlier studies and a previous failure to replicate emotional effects on IB (Moreton et al., 2017). In this experiment, we used anger instead of disgust since anger expression has not been used in previous studies on emotions and IB (Moreton et al., 2017).

Experiment 4

We used instruction-based actions as in experiment 2 but with the emotional word “angry” instead of “disgust” to check whether the effect in experiment 3 is due to intentions. Rest of the details were the same as in experiment 2 and same action-outcome combinations as in experiment 3.

General discussion

Given discrepancies in earlier findings on the effect of emotions on IB (Christensen et al., 2016, Moreton et al., 2017, Tanaka & Kawabata, 2021, Yoshie and Haggard, 2013, Yoshie and Haggard, 2017), we conducted the current study to investigate the effect of emotions using emotional faces. The results showed that the binding was more (estimated temporal judgments were shorter) when negative expression was intended compared to positive expressions (experiments 1 and 3). This differential binding

CRediT authorship contribution statement

Dhrubajyoti Sarma: Conceptualization, Software, Formal analysis. Narayanan Srinivasan: Conceptualization, Methodology, Formal analysis.

References (28)

  • C. Stetson et al.

    Motor-sensory recalibration leads to an illusory reversal of action and sensation

    Neuron

    (2006)
  • M. Yoshie et al.

    Negative emotional outcomes attenuate sense of agency over voluntary actions

    Current Biology

    (2013)
  • S. Antusch et al.

    The role of intentional strength in shaping the sense of agency

    Frontiers in Psychology

    (2019)
  • A. Bandura

    Social cognitive theory: An agentic perspective

    Annual Review of Psychology

    (2001)
  • Cited by (3)

    • Social, affective, and non-motoric bodily cues to the Sense of Agency: A systematic review of the experience of control

      2022, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      No significant difference was found even when sounds were replaced by emoticons or human faces portraying positive and negative emotions. Similarly, Sarma and Srinivasan (2021) did not find any modulation of IB by outcome valence, although they did find stronger IB when participants intended to generate negative stimuli with their button presses, irrespective of the actual outcome. Majchrowicz and Wierzchoń (2021) employed positive and negative vocalizations as action outcomes in a study where they aimed to understand the effects of valence on SA and explicit SoA ratings.

    View full text