Embodied mood regulation: the impact of body posture on mood recovery, negative thoughts, and mood-congruent recall

ABSTRACT

Previous work has shown that a stooped posture may activate negative mood. Extending this work, the present experiments examine how stooped body posture influences recovery from pre-existing negative mood. In Experiment 1 (n = 229), participants were randomly assigned to receive either a negative or neutral mood induction, after which participants were instructed to take either a stooped, straight, or control posture while writing down their thoughts. Stooped posture (compared to straight or control postures) led to less mood recovery in the negative mood condition, and more negative mood in the neutral mood condition. Furthermore, stooped posture led to more negative thoughts overall compared to straight or control postures. In Experiment 2 (n = 122), all participants underwent a negative mood induction, after which half received cognitive reappraisal instructions and half received no instructions. Mood-congruent cognitions were assessed through autobiographical memory recall. Again, stooped (compared to straight) position led to less mood recovery. Notably, this was independent of regulation instruction. These findings demonstrate for the first time that posture plays an important role in recovering from negative mood.

KEYWORDS:

• Embodiment
• emotions
• mood regulation
• body posture
• reappraisal
• emotion regulation

Emotions are manifested not just in what people are saying, but also, and perhaps even more significantly, in what people are doing. For instance, when people are brooding over a negative event, chances are that their back will become more arched, their head bowed down, and their eyes looking to the ground. This is because negative emotional states are associated with a stooped body posture (e.g. Michalak et al., 2009; Oosterwijk, Rotteveel, Fischer, & Hess, 2009; Riskind & Gotay, 1982). When people are metaphorically feeling down, their body posture is prone to be literally oriented downwards.

Why does feeling down often go together with leaning down? Perhaps the most obvious explanation is that bodily postures like stooping are expressions of people’s emotional states (e.g. Michalak et al., 2009; Oosterwijk et al., 2009). In the present research, however, we address how bodily postures may play a more active role in people’s emotional lives, by shaping emotional processing. In what follows, we start by considering how bodily postures may influence emotion generation, by resulting people to have more or less negative feelings. Building on recent theories of embodied emotion and situated cognition, we then suggest the new possibility that bodily posture may also influence emotion regulation, by changing how people are dealing with negative feelings that have already been activated. Finally, we present two experiments that empirically tested the effects of posture on regulating negative mood.

Posture and emotion generation

Keen observers of human nature have long noted that people’s emotions are linked to specific behavioural expressions. For instance, Charles Darwin described how

After the mind has suffered from an acute paroxysm or grief, [..], we fall into a state of low spirits; or we may be utterly cast down and dejected; [..] the muscles flaccid; the eyelids droop; the head hangs on the contracted chest; the lips, checks, and lower jaw all sink downwards from their own weight. (1872/1965, p. 105)

In a related vein, William James (1890), and generations of emotion theorists after him, have suggested that bodily states not only accompany certain feelings, but bodily states may also have a direct effect on emotion processing (Adelmann & Zajonc, 1989; Damasio, 1994; Niedenthal, 2007; Panksepp, 2005; Strack, Martin, & Stepper, 1988). Thus, emotional states such as feeling down can be a consequence, and not only the cause, of bodily states such as sitting in a stooped position.

Experimental research has confirmed that bodily postures can indeed have a direct effect on emotional processing. For example, when people are led to adopt a straight posture and smile, they are faster at retrieving positive (compared to negative) autobiographical memories (Riskind, 1984). Moreover, assuming a stooped rather than a straight seating posture may lead people to display less pride (Stepper & Strack, 1993), and develop an increased level of helplessness (Riskind & Gotay, 1982). Also, sitting in a stooped rather than upright posture while imagining a positive or negative event, leads people to recall more negative words related to the event afterwards (Michalak, Mischnat, & Teismann, 2014). Not just sitting stooped can cause negative emotional processing, also walking with a more slumped, and forward leaning posture causes people to recall relatively more negative memories than walking with an upright posture (Michalak, Rohde, & Troje, 2015). It thus appears that a stooped posture may play a significant role in evoking experiences and cognitions related to negative emotion.

Classic emotion theories accounted for the pervasive emotional impact of bodily states by assuming that bodily states influence emotion processing through peripheral (i.e. psychophysiological) processes (James, 1890; Lange, 1885; see also Adelmann & Zajonc, 1989; Zajonc, 1998). More recent work, however, suggests that bodily states may also influence more central forms of emotional processing. In particular, theories of embodiment have proposed that human cognition and emotion rely on embodied simulations (e.g. Barsalou, 1999, 2003; Niedenthal, 2007; Semin & Cacioppo, 2008; Smith & Semin, 2007). Just thinking about an emotion, or emotion-eliciting situation, may lead to changes in people their feelings, facial expression, body posture, and behaviour, as it would have been if they actually had encountered the emotion-eliciting situation. For instance, musing about a funny situation can partially activate the facial muscles involved in smiling. This presumed sequence also works the other way around; lower level bodily processes may influence higher level processes, such as thoughts or cognitions. For example, activating facial muscles related to smiling can activate positive cognitions (Strack et al., 1988).

If embodied theories are valid, then the impact of bodily states extends to high-level forms of emotion processing. One form of high-level emotion processing that is of particular interest is emotion regulation. During emotion regulation, people use superordinate control processes to change the spontaneous course of their emotions (Koole, 2009; Oschner & Gross, 2005; Webb, Miles, & Sheeran, 2012). Emotion regulation is a vital process for maintaining physical and psychological health (Aldao & Nolen-Hoeksema, 2012; Koole, 2009; Webb et al., 2012). It is therefore important to establish whether and how bodily states, like posture, may influence emotion regulation.

Posture and emotion regulation

To date, emotion regulation research has mostly focused on cognitive regulation strategies, such as reappraisal and cognitive distraction (Aldao & Nolen-Hoeksema, 2012; Gross, 2015; Webb et al., 2012). Nevertheless, preliminary work suggests that bodily processes may also be helpful in emotion regulation. For example, controlled breathing exercises can be effective in inducing emotional states (Philippot, Chapelle, & Blairy, 2002), and muscle relaxation exercises can reduce anxiety (see Manzoni, Pagnini, Castelnuovo, & Molinari, 2008 for a meta-analysis). Together, these results suggest that changing bodily processes can influence emotion regulation.

To understand just how bodily states may influence emotion regulation, we have recently developed a situated cognition perspective on emotion regulation (Koole & Veenstra, 2015; see also Feldman Barrett, Wilson-Mendenhall, & Barsalou, 2014, for a situated conceptualisation of emotions and emotion regulation). According to a situated cognition perspective, people do not rely so much on abstract mental models in regulating their emotions. Rather, people regulate their emotions on the fly, in the spur of the moment, using anything that they can use along the way. A key notion here is that of affordances. An affordance refers to all the possible actions a person can undertake given the restraints of the situation (Gibson, 1979). An affordance is not a mental construction that exists solely in a person’s mind, but a physical reality that exists in the person’s relationship with the environment. Affordances in general refer to the fit between people’s goals and their current situation. Applied to emotion regulation, affordances refer to the fit between people’s emotion-regulatory goals and the possibilities that the current situation provides for meeting these goals.

From a situated cognition perspective, the body represents one of the key affordances for emotion regulation (Koole & Veenstra, 2015). Because different bodily states are mentally associated with specific emotional processes, different bodily states offer different opportunities for regulating one’s emotions. For instance, due to the existing association between stooped posture and negative mood and cognitions (e.g. Riskind & Gotay, 1982), a stooped body posture may prolong people their negative mood, when people are experiencing negative feelings. Conversely, an upright body posture may make it easier for people to recover from negative feelings, because straight posture is related to feelings of power (Carney, Cuddy, & Yap, 2010) and pride (Oosterwijk et al., 2009). In this manner, people’s bodily states may offer different affordances for emotion regulation. From this perspective, emotion regulation is not just an activity of the mind, but rather an activity that relies on a continual interplay between mind and body.

The situated cognition perspective has broad implications for the role of all kinds of bodily processes in emotion regulation. In the present article, however, we focus on a specific bodily process, namely, body posture. For several reasons, body posture is highly suited for testing the role of bodily processes in emotion regulation. First, people’s body always assumes some sort of posture, so that body posture exerts a ubiquitous influence on people’s responses. Second, body posture can be experimentally manipulated in an efficient and unobtrusive manner (e.g. Strack et al., 1988). Third, there already exists a literature on the effects of body posture on emotion generation (e.g. Michalak et al., 2014; Riskind & Gotay, 1982; Stepper & Strack, 1993). This allowed us to build on this literature and to examine the effects of body posture on emotion regulation.

Present research and hypotheses

In two experiments, we examined how body posture influences the regulation of negative mood. Our general prediction across both experiments was that recovery from negative mood would be impaired when participants assumed a stooped rather than a straight seating posture. In Experiment 1, we asked participants to imagine a negative or a neutral situation, and subsequently manipulated their body posture in a stooped, straight, or self-chosen (control) body posture, and asked them to list their thoughts to foster spontaneous mood regulation. We predicted that stooped posture – relative to control and straight postures – would prevent effective mood recovery in the negative imagination condition, and would increase negative affect in the neutral imagination condition. In Experiment 2, we asked all participants to imagine a negative situation, and experimentally manipulated whether participants were requested to engage in cognitive reappraisal, a well-established cognitive emotion regulation strategy (e.g. Sheppes & Meiran, 2007), or to engage in no emotion regulation strategy. The design of Experiment 2 thus allowed us to explore whether and how the effects of posture interact with cognitive emotion regulation.

Experiment 1

In Experiment 1 we examined how stooped posture influences spontaneous emotion regulation, by contrasting stooped posture with straight or control postures, both after an induction of negative mood and without a mood induction. When no negative mood was induced, we expected that participants in a stooped posture would experience more negative mood compared to sitting straight (or control). However, when a negative mood was induced, we expected that posture would make mood recovery more or less effective depending on the specific body posture. More specifically, we expected that participants with a stooped posture would show less mood recovery than participants with a straight (or control) posture, signifying that posture is relevant to mood regulation. We also tested whether posture influenced the valence of participants’ thoughts during the recovery period. Based on the notion that stooped posture interferes with down-regulation of negative mood, we expected that sitting stooped (compared to straight and control) would cause a larger proportion of negative thoughts within the negative mood condition as well as the neutral mood condition.

Method

Participants and design

Two hundred and twenty-nine undergraduate students (163 women; M_age= 21.05, SD_age = 2.66, range = 18–37) at the VU University Amsterdam participated for course credits or money. Participants were randomly assigned to a stooped (n = 76), straight (n = 76), or control (n = 77) posture condition. In addition, half of the participants were randomly assigned to a neutral mood condition (n = 113), and half to a negative mood condition (n = 116). Self-reported negative mood was the main dependent variable.

We determined the required sample size using G*Power (Faul, Erdfelder, Lang, & Buchner, 2007). Due to a lack of previous research on the effects of posture on mood regulation, we assumed a small effect size (Cohen’s f = 0.12) of the present 3 (mood condition) × 2 (posture condition) between-subjects × 3 (time of mood measurement) within-subjects interaction, which resulted in a required sample size of N_required = 210 to detect an interaction with power of 1 – β = 0.85.

Procedure and materials

Mood measurement

We told participants that the study included different parts with interests associated with cognition, behaviour, and personality, and to get a sense of their lab experience, participants would have to rate their mood multiple times. To generate a baseline of mood, at the start of the experiment, participants rated their mood with the Dutch abbreviated version of the Profile of Mood States (POMS; McNair, Lorr, & Droppleman, 1971/1981/1992; Wald & Mellenbergh, 1990) by indicating how they currently felt for each of the 32 items referring to the five subscales of the POMS (e.g. “helpless”, or “sad”) on a computerised 40-point scale with labels on both endpoints (1 = not at all, 40 = very much). The POMS is generally used as a single scale to measure negative affect (McNair et al., 1971/1981/1992; Wald & Mellenbergh, 1990). Thus, to calculate negative mood, we recoded the reversed scored items and averaged all 32 items into one single mood score, for which higher scores indicate a higher level of negative mood (all Cronbach α’s time point 1–3 = .96).

Mood manipulation

To induce negative mood, participants completed a mental imagery task, based on validated mood induction procedures (e.g. Larsen & Ketelaar, 1991). Specifically, in the negative mood induction condition, we asked participants to imagine that their best friend had been deceased from an incurable disease, and instructed them to,

Consider what thoughts you would have in this situation. Experience the feelings you would feel in this situation. Make the imagination of the situation as vividly as you can. Close your eyes if this helps. See detailed images of the situation. Let yourself react as if you were actually there. See the people. Hear the sounds.

In the neutral mood condition, we asked participants to recall what they did from the moment that they came out of bed this morning until they left the house. Participants had four minutes to read the instructions, close their eyes, and imagine being in the situation. After the mental imagery task, participants answered four questions about their imagination ability, to bolster the cover story for the procedure, and rated their mood for a second time.

Posture manipulation

To reduce expectation effects of posture on emotional experience, we told participants that the next part of the study concerned the effects of body posture on performing a task. To manipulate body posture, we instructed one-third of the participants to adopt a stooped body posture, one-third to adopt a straight body posture, and the final third to adopt a comfortable body posture (control). Participants in the stooped body posture condition were instructed to take the clipboard and pen lying on the desk, move their chair back and put the clipboard on their lap. They were asked to (a) move their upper body forward so that their back was curved, (b) bend their head downward so their gaze was directed to their lap, and (c) stay in this position until the instructions on the computer stated otherwise.

Participants in the straight posture condition were instructed to put the clipboard and pen on the desk in front of them and move their chair close to the table. Furthermore, participants were asked to sit with their back against the backrest of the chair, put their feet on the ground and let their hands rest comfortably on the desk. They were instructed to keep their head straight up, as if there was a book on top of it, and stay in this position until the instructions on the computer told otherwise. A line figure illustrated the instructed stooped and straight body postures (see Figure 1).

Figure 1. Line figures illustrating a stooped and straight body posture.

Participants in the control condition were told that they were assigned to the control body posture condition, and asked to adopt a comfortable body position and keep it until the instructions told otherwise. After the posture instructions, we asked participants to call the experimenter before they continued, so she could make a note of their body posture. If necessary, the experimenter repeated some of the instructions similar to the instructions on the screen, until the participant adopted the desired seating position.

Thought-listing task

While being seated in the instructed body posture, all participants received a thought-listing task to induce spontaneous mood regulation. The instructions were printed on paper, which was attached to a clipboard, such that all posture conditions could keep the adopted body posture while reading the instructions. We instructed participants to write down all thoughts that came to their mind during the next four minutes. Participants were told to use a new line for each thought and instructed to retain the instructed seating position throughout the task. Finally, participants were instructed to adopt their natural posture while filling out the third and final mood measurement. After this mood rating, participants evaluated the valence of the thoughts they had listed. For each thought, participants indicated whether it was positive, negative, or neutral, by placing a plus sign (+), a minus sign (−), or a star (*) in front of each thought. This allowed us to test whether posture influenced the valence of participants’ thoughts.

Manipulation checks

As a manipulation check of participants’ body posture, the experimenter rated the participants’ body posture on a 7-point scale (1 = strongly stooped, 7 = strongly slouched) (see Figure 2). At the end of the experiment, participants themselves also indicated the body posture they had adopted during the task on a similar scale. The reported posture effects did not change when we used either the experimenter’s or participants’ ratings.

Figure 2. Line figures presented in a scale to indicate which body position participants adopted.

In addition, we asked participants how much physical tension they experienced when they were writing in the instructed body posture, and how comfortable they found the body posture on a visual analogue slider scale (1 = not at all, 100 = very much) to rule it out as alternative explanation of the body posture effects. Participants perceived the stooped body posture as more physically challenging than the other posture conditions, and rated both body posture instructions (straight and stooped) as less comfortable than the control condition (see supplemental material for the statistical results). However, statistically controlling for these differences did not change our results, nor did these differences interact with the manipulation of posture.

Finally, participants were asked for their demographic information, were thanked, debriefed, and received course credits or money for participation. Altogether, the experiment lasted approximately 40 minutes.¹

Results

Mood

Our general prediction was that mood recovery would be impaired when participants assumed a stooped rather than a straight (or control) seating posture. In addition, in accordance with previous work, we hypothesised that sitting stooped causes relatively more negative mood compared to sitting straight (or control) in the absence of a negative mood induction. Thus, we conducted a mixed-model ANOVA with body posture (stooped vs. straight vs. control) and mood condition (neutral vs. negative) as between-subjects factors, and time of mood measurement (before mood induction-t1 vs. after mood induction-t2 vs. after recovery period-t3) as a within-subjects factor. Means of all groups are provided in Table 1. To further test our hypotheses, we conducted planned contrast analyses.

Table 1. Means and standard deviations (M (SD)) of negative mood on three time points (t1: before mood induction, t2: after mood induction, t3: after thought listing) for the two different mood conditions (neutral vs. negative), and three body posture conditions (stooped vs. straight vs. control) (Experiment 1).

	Neutral				Negative				Total
Posture	t1	t2	t3	N	t1	t2	t3	N	t1	t2	t3	N
Stooped	21.68 (15.64)	21.86 (15.80)	25.60 (16.70)	38	24.27 (16.80)	31.46 (18.22)	29.69 (17.76)	39	22.99 (16.18)	26.72 (17.63)	27.67 (17.26)	77
Straight	22.48 (12,78)	23.07 (13.03)	22.77 (13.22)	37	22.67 (14,13)	30.34 (16.15)	24.78 (15.21)	39	22.57 (13.40)	26.80 (15.07	23.81 (14.22)	76
Control	21.73 (13.14)	21.80 (15.67)	21.30 (14.44)	38	24.78 (11.96)	34.83 (15.48)	26.84 (14.81)	38	23.26 (12.57)	28.31 (16.80)	24.07 (14.80)	76
Total	21.96 (13.80)	22.23 (14.78)	23.23 (14.85)	113	23.90 (14.35)	32.21 (16.63)	27.11 (15.98)	116	22.94 (14.09)	27.28 (16.48)	25.19 (15.52)	229

Manipulation check

First, we checked whether our mood manipulation was successful. As expected, the overall interaction between mood condition and time was significant, F(2, 446) = 21.76, p < .0001, η² = .089. The increase in negative mood during the mood induction (t1–t2) was significantly larger after the negative mood induction, MD_t2–t1= 8.31, Cohen’s d_z = 0.68, 95%CI d_z[0.48, 0.88], than after the neutral mood induction, MD_t2−t1 = 0.27, Cohen’s d_z = 0.05, 95%CI d_z[−0.14, 0.23], F(1, 223) = 39.53, p < .0001, η² = .151, for the contrast of the interaction. To check whether the predicted posture effects were not influenced by a priori mood differences, we conducted contrast analyses between t1 and t2 for the overall significant two-way interaction between body posture and time, F(4, 446) = 3.42, p = .009, η² = .030. Planned contrasts revealed no significant differences between the body posture conditions during the mood induction period (t1−t2), F(1, 223) = 0.40, p = .672, η² = .004, indicating that there were no a priori mood differences in the posture conditions. Thus, our mood manipulation was successful.

Mood recovery

Our general prediction was that mood recovery would be impaired when participants assumed a stooped rather than a straight (or control) seating posture. Importantly, the contrast of the interaction between time (t2–t3) and posture condition showed that changes in negative mood differed between posture conditions, F(1, 223) = 5.98, p = .003, η² = .051 (see Figure 3). Further contrasts revealed that the decrease of negative mood was smaller in the stooped posture condition, MD_t3−t2= 0.95, Cohen’s d_z = 0.10, 95%CI d_z [−0.13, 0.32], than in the control posture condition, MD_t3–t2= −4.24, Cohen’s d_z = 0.42, 95%CI d_z [0.19, 0.65], F(1, 223) = 11.04, p = .001, η² = .047, and in the straight posture condition, MD_t3–t2= −4.24, Cohen’s d_z = 0.28, 95%CI d_z [0.05, 0.51], F(1, 223) = 6.17, p = .014, η² = .014. Mood decreases were similar in the straight and control conditions, F(1, 223) = 0.70, p = .405, η² = .003. Together, these results indicate that adopting a stooped posture resulted in less mood recovery than adopting a straight or control posture.

Figure 3. Posture effect on negative mood. Negative mood on three time points (t1: before mood induction, t2: after mood induction, t3: after thought listing) for three body posture conditions (stooped versus straight versus control) with standard error bars (1-SE) adjusted for within variability (Experiment 1).

Mood generation

In accordance with previous work (e.g. Michalak et al., 2014; Riskind & Gotay, 1982; Stepper & Strack, 1993), we also hypothesised that sitting stooped causes relatively more negative mood compared to sitting straight (or control) in the absence of a negative mood induction. We examined whether stooped (compared to straight or control) posture caused negative mood in the neutral condition. This was only the case when we compared the mood difference of stooped posture with the average of mood difference of straight and control posture. More specifically, stooped posture increased negative mood, MD_t3−t2 = 3.74, Cohen’s d_z = 0.49, 95%CI d_z [0.15, 0.83], compared to the average of straight and control posture, MD_t3–t2 = −0.40, Cohen’s d_z = 0.05, 95%CI d_z [−0.27, 0.37], F(1, 223) = 4.57, p = .034, η² = .020. Together, these results indicate that stooped posture increased negative mood in the neutral mood condition.²

Thought ratings

Finally, we examined whether posture influenced the valence of participants’ thoughts. We predicted a larger proportion of negative thoughts for participants in a stooped (compared to straight and control) seating posture, regardless of mood condition. To test this prediction, we conducted a factorial ANOVA with body posture (straight vs. control vs. stooped) and mood condition (neutral vs. negative) as between-subjects factors, and proportion of negative thoughts as dependent variable. Means are provided in Table 2.

Table 2. M (SD) of the proportions of negative thoughts for the different mood conditions (neutral vs. negative), and body postures (straight vs. stooped vs. control) (Experiment 1).

Posture	Neutral	N	Negative	N	Total	N
Stooped	.42 (0.23)	37	.46 (0.26)	37	.44 (.25)	74
Straight	.38 (0.21)	35	.30 (0.19)	38	.34 (.20)	73
Control	.34 (0.24)	38	.32 (0.17)	38	.33 (.21)	76
Total	.38 (0.23)	110	.36 (0.22)	113	.37 (.22)	223

As predicted, posture had a significant effect on the proportion of negative thoughts, F(2, 217) = 5.46, p = .005, η² = .048 (see Figure 4). More specifically, contrast results showed that participants with a stooped posture reported a significantly larger proportion of negative thoughts than participants with a straight posture, MD = 0.10, Cohen’s d = 0.43, 95%CI d [0.10, 0.75], F(1, 217) = 7.11, p = .008, η² = .032, or participants with a control posture, MD = 0.09, Cohen’s d = 0.47, 95%CI d [0.15, 0.79], F(1, 217) = 9.16, p = .003, η ²= .040. Participants in the control and straight body posture conditions reported the same proportion of negative thoughts, MD = 0.01, Cohen’s d = 0.05, 95%CI d [−0.27, 0.37], F(1, 217) = 0.11, p = .741, η² = .001. There was no main effect of mood condition, F(1, 217) = 0.55, p = .458, η² = .003, nor an interaction effect between body posture and mood condition, F(2, 217) = 1.36, p = .258, η² = .012.

Figure 4. Posture effect on valence of thoughts. Proportion of negative thoughts for the different body posture conditions (straight versus control versus stooped), and mood induction conditions (negative versus neutral) with standard error bars (1-SE) (Experiment 1).

Entering proportion of negative thoughts as covariate did not diminish the effects of body posture or mood induction condition on negative mood, nor did proportion of negative thoughts interact with the observed effects for mood.³ Together, these results show that stooped posture (compared to straight or control postures) led to an increase in negative thoughts in both mood conditions, but this increase was unrelated to the effects of stooped posture on mood.

Discussion

The results of Experiment 1 showed that a stooped compared to straight or control posture influenced both mood regulation and mood generation. In the absence of a mood induction, people in a stooped posture reported larger increases in negative mood than those sitting upright. The latter effect is consistent with prior research showing that posture influences mood generation (e.g. Michalak et al., 2014; Riskind & Gotay, 1982; Stepper & Strack, 1993). Going beyond prior research, Experiment 1 showed further that posture also influences mood recovery. Specifically, after a negative mood induction, people in a stooped posture recovered less from their negative mood than those sitting upright. Notably, participants in a stooped posture also had more negative thoughts than participants in a straight or control body posture. However, the latter effects on thought valence were unrelated to the effects of posture on mood.

Experiment 2

In Experiment 1, we investigated spontaneous mood regulation; we did not instruct people to deliberately regulate their emotions. However, when people feel negative they can use all kind of cognitive strategies to get rid of these feelings, for example, distract themselves, reinterpret the situation, or suppress the expression of their feelings (e.g. Gross, 1998). Reinterpreting the emotion-eliciting situation, also known as reappraisal, has often been found to be an effective regulation strategy (Webb et al., 2012). The question thus arises whether, and if so, how posture still influences emotional recovery when people use such a deliberate emotion regulation strategy. We sought to answer this question in Experiment 2.

In Experiment 2, we investigated the joint influence of body posture and deliberate mood regulation on mood recovery. We again used a mental imagery task to induce negative mood. However, we dropped the neutral mood condition from our design. Thus, all participants imagined a negative situation, after which we manipulated whether participants were requested to engage in cognitive reappraisal (e.g. Sheppes & Meiran, 2007), or to engage in no emotion regulation strategy. Next, we asked participants to adopt either a stooped or straight body posture. We further added an autobiographical recall task to the self-reported mood measurement. Mood-congruent recall has been often used to gauge the passive priming of negative thought contents (e.g. Joormann & Siemer, 2004). Thus, the autobiographical recall task assessed the activation of mood-congruent cognitions in a more implicit manner.

As in Experiment 1, we predicted that participants with a stooped body posture would display less recovery from their negative mood than participants with a straight body posture. Based on previous work on the effectiveness of emotion regulation strategies (e.g. Gross, 1998; Webb et al., 2012), we further predicted that participants who actively regulated their emotions by using reappraisal would recover better from their negative mood than participants who regulated their emotions spontaneously.

To our knowledge, no previous research has examined the joint effects of posture and deliberate emotion regulation. Nevertheless, on theoretical grounds, we can discern three possibilities. First, posture and deliberate emotion regulation might be separate processes. In this case, posture should have an effect on emotional recovery, independent of whether people deliberately try to regulate their emotions. Second, effects of posture might be amplified by deliberate regulation, such that stooped posture interferes with, and/or straight posture facilitates the effects of cognitive reappraisal on negative affect (see Balcetis & Cole, 2009; Riskind, 1984). This would lead to an interaction between posture and regulation strategy, such that body posture has a greater effect on mood recovery when people deliberately reappraise their present feelings compared to spontaneous regulation. Third and last, the effects of body posture, which are presumably based on more bottom-up automatic processes, might be weakened by the top-down, deliberate down-regulation of negative affect. The latter would also lead to an interaction between posture and regulation strategy, but one in which body posture has a greater effect when people spontaneously regulate their feelings than when they deliberately regulate their feelings.

Method

Participants and design

One hundred and twenty-two undergraduate students (75 women; M_age= 20.45, SD_age = 2.85, range = 17–33) from the VU University Amsterdam participated for course credits or monetary rewards. Due to limited available lab space, it was not clear whether we would be able to run the whole design before we started the study. Consequently, we first ran the reappraisal regulation condition (n = 61), and a few weeks later, ran the spontaneous regulation condition (n = 61). We always assigned participants randomly to one of two body posture conditions (stooped, n = 61 vs. straight, n = 61), with the restriction that each of the experimental groups would have an equal number of participants. Self-reported negative mood, and the valence of recalled autobiographical memories were the main dependent variables. Using G*Power (Faul et al., 2007) with an assumed small-medium effect size (estimation based on the effect size of Experiment 1; Cohen’s f = 0.16) of the present 3 (within-subjects) × 2 × 2 (between-subjects) interaction, we determined a required sample size of N_required = 100 to detect an interaction with power of 1 – β = 0.85.

Procedure and materials

The equipment and setting were similar to those of Experiment 1, except that we omitted the neutral condition of the mood manipulation, and the self-chosen posture condition of the body posture manipulation. The differences in measurements and materials are explained below.

Emotion regulation manipulation

In the reappraisal condition, we asked participants to reappraise the previously imagined sad situation, by asking them:

In an earlier part of this study, you have imagined how it would be to lose a dear friend. This would obviously be a very negative event. However, it is possible to think about this event in a way that it is perhaps less negative and sad. Try to think about the event in such a way that your feelings and thoughts would be less negative and sad.

In the spontaneous regulation condition, the instruction was similar to the instructions of Experiment 1, that is, participants were asked to list all thoughts that came to mind.

Mood measurement

Four items about physical experiences were added to the POMS to make the cover story about participants’ general lab experience more plausible (e.g. “head ache”, “cold”). To reduce the length of the experimental session and prevent fatigue among participants, one of the five subscales (fatigue) of the POMS was excluded. To calculate negative mood, we recoded the reversed scored items and averaged all 26 items into one single mood score, for which higher scores indicate a higher level of negative mood (all Cronbach α’s at time points 1–3 = .96–.97).

Mood-congruent recall

Besides the self-report measures of mood, participants completed a mood-congruent recall task (e.g. Joormann & Siemer, 2004; Rusting & DeHart, 2000), in which they were instructed to recall five different memories from their high school period. On the next screen, they evaluated each of the recalled memories on a computerised slider scale with an affective label on both ends (1 = negative, 100 = positive). The five ratings were averaged into one evaluation score.

Results

Manipulation check for regulation instruction

To investigate whether participants in the reappraisal condition actually prompted to reappraise, each thought of each participant was rated by two independent blind coders as positive, negative, or neutral. The mean inter-rater reliability of the two coders was .84 (SD = 0.16). The average number of thoughts participants had written down was 10.00 (SD = 5.71), and the average proportion of positive thoughts was .47 (SD = 0.41). An independent t-test showed that the proportion of positive thoughts was larger in the reappraisal condition (M = .84, SD = 0.23), than in the spontaneous regulation condition (M = .10, SD = 0.12), t(120) = 22.86, p < .0001, Cohen’s d = 4.07, 95%CI d [3.50, 4.77]. We concluded that our instruction led participants to reappraise the negative imagination, confirming that our manipulation of regulation strategy was successful.

Mood

To test our hypotheses, we conducted a mixed-model ANOVA with body posture (straight vs. stooped) and regulation strategy (reappraisal vs. spontaneous) as between-subjects factors, and time of mood measurement (before mood induction-t1 vs. after mood induction-t2 vs. after recovery period-t3) as within-subjects factor. Relevant means are shown in Table 3.

Table 3. M (SD) of negative mood on three different time points (t1: before mood induction, t2: after mood induction, t3: after writing task) for the two different body posture conditions (straight vs. stooped), and two regulation conditions (spontaneous vs. reappraisal) (Experiment 2).

	Spontaneous				Reappraisal				Total
Posture	t1	t2	t3	N	t1	t2	t3	N	t1	t2	t3	N
Straight	10.04 (5.40)	13.78 (6.65)	9.96 (5.71)	30	9.70 (5.49)	12.53 (6.79)	10.88 (5.96)	31	9.87 (5.40)	13.14 (6.69)	10.43 (5.81)	61
Stooped	10.82 (4.70)	13.66 (6.60)	12.01 (6.14)	31	10.17 (5.82)	11.77 (6.70)	11.64 (6.20)	30	10.50 (5.24)	12.73 (6.66)	11.83 (6.12)	61
Total	10.44 (5.03)	13.71 (6.57)	11.00 (5.97)	61	9.93 (12.16)	12.16 (6.70)	11.25 (6.04)	61	10.18 (5.31)	12.94 (6.65)	11.13 (5.98)	122

Manipulation check mood induction

We checked whether our negative mood induction was successful, by looking at the t1–t2 (before mood induction vs. after mood induction) mood effects in the mixed-model ANOVA described above. There was an overall significant effect of time, F(2, 236) = 24.50, p < .0001, η² = .172.⁴ And indeed, planned contrasts showed that participants’ average negative mood increased significantly from before (t1) to after the imagination task (t2), MD = 2.76, Cohen’s d_z = 0.69, 95%CI d_z [0.48, 0.88], F(1, 118) = 57.52, p < .0001, η² = .328. We also checked whether the mood induction was the same for all posture and regulation conditions. First, there was an overall marginal interaction between body posture and time, F(2, 236) = 2.68, p = .071, η² = .022. However, planned contrasts between time 1 and time 2, showed no a priori differences between the body posture conditions during the mood induction period (t1–t2), F(1, 118) = 2.16, p = .145, η² = .018. In addition, there was an overall marginal interaction between regulation condition and time, F(2, 236) = 2.67, p = .071, η² = .022. However, again there were no a priori mood differences between the regulation conditions during the mood induction (t1–t2), F(1, 118) = 2.20, p = .141, η² = .018. Finally, there was no overall three-way interaction between body posture, regulation strategy and mood recovery, F(2, 236) = 0.19, p = .824, η² = .002, neither was there a significant three-way interaction for the planned contrasts between time 1 and time 2.

Mood recovery

To test our hypotheses, we further analysed the mixed-model ANOVA with body posture (straight vs. stooped) and regulation strategy (reappraisal vs. spontaneous) as between-subjects factors, and time of mood measurement (before mood induction-t1 vs. after mood induction-t2 vs. after recovery period-t3) as within-subjects factor. As in Experiment 1, we predicted that participants who adopted a stooped posture would recover less from their negative mood than participants who adopted a straight posture. Indeed, following up on the overall interaction between body posture and time, posture affected mood recovery significantly from t2 to t3, F(1, 118) = 5.13, p = .026, η² = .042 (see Figure 5). More specifically, the decrease in negative mood was larger for participants sitting in a straight position, MD_t3–t2 = −2.71, Cohen’s d_z = 0.79, 95%CI d_z [0.50, 1.07], than for participants sitting in a stooped position, MD_t3–t2 = 0.93, Cohen’s d_z = 0.26, 95%CI d_z [−0.09, 0.42]. Thus, overall, participants with a stooped body posture recovered less from their negative mood than participants with a straight body posture, showing that stooped posture interfered with mood recovery, replicating the findings of Experiment 1.

Figure 5. Posture effect on mood. Negative mood on three time points (t1: before mood induction, t2: after mood induction, t3: after thought listing) for body posture conditions (stooped versus straight) with standard error bars (1-SE) adjusted for within variability (Experiment 2).

Furthermore, based on previous research (Webb et al., 2012), we expected that reappraisal would lead to more mood recovery than spontaneous regulation. Unpacking the two-way interaction between regulation condition and time, there was a significant interaction between regulation instruction and time during the recovery period (t2–t3), F(1, 118) = 5.13, p = .025, η² = .042 (see Figure 6). However, this effect was in the opposite direction of the predicted effect: Participants who used reappraisal showed negligible recovery from their negative mood, MD_t3−t2 = 0.91, Cohen’s d_z = 0.19, 95%CI d_z [−0.06, 0.45], whereas participants who used spontaneous mood regulation showed mood recovery, MD_t3−t2 = −2.71, Cohen’s d_z = 0.61, 95%CI d_z [0.33, 0.88]. Thus, unexpectedly, cognitive reappraisal led to less mood recovery than spontaneous regulation.⁵

Figure 6. Regulation effect on mood. Negative mood on three time points (t1: before mood induction, t2: after mood induction, t3: after thought listing) for two regulation conditions (reappraisal versus spontaneous) with standard error bars (1-SE) adjusted for within variability (Experiment 2).

Finally, we explored the combined effect of cognitive reappraisal and body posture on mood recovery. There was no three-way interaction between body posture, regulation strategy and mood recovery, neither was there a significant three-way interaction for the planned contrast between time 2 and time 3. This indicates that regulation type and posture had distinct effects on self-reported mood recovery, and that cognitive regulation did not amplify or weakened the effects of posture on mood recovery.

Mood-congruent recall

To test the effects of posture and mood regulation strategy on the implicit mood measure, we conducted a 2 (regulation strategy: reappraisal vs. spontaneous) × 2 (posture: stooped vs. straight) factorial ANOVA with mean valence of the recalled memories as dependent variable. Relevant means are shown in Table 4.

Table 4. M (SD) of valence of the recalled memories for the different posture (straight vs. stooped), and regulation (spontaneous vs. reappraisal) conditions (Experiment 2).

Posture	Spontaneous	N	Reappraisal	N	Total	N
Straight	71.85 (14.09)	30	68.95 (15.62)	31	70.38 (14.84)	61
Stooped	64.72 (15.29)	31	73.65 (16.00)	30	69.11 (16.16)	61
Total	68.22 (15.03)	61	71.27 (15.86)	61	69.74 (15.46)	122

First, we predicted that stooped posture would cause recall of less positive memories than straight posture. However, the main effect of posture on mood-congruent recall was not significant, F(1, 118) = 0.19, p = .661, η² = .002. In addition, we expected that cognitive reappraisal would lead to greater recall of positive memories than spontaneous emotion regulation. However, there was no main effect of regulation condition, F(1, 118) = 1.19, p = .277, η² = .010, indicating that regulating negative affect through reappraisal was no more effective than spontaneous emotion regulation.

Recall that body posture and cognitive mood regulation could either be separate processes (no interaction), or could amplify, or weaken one another (interaction). Addressing this, the interaction effect between regulation strategy and body posture, F(1, 118) = 4.57, p = .035, η² = .037, was significant, such that in the stooped posture condition, reappraisal resulted in more positive memories than spontaneous regulation, MD = 8.83, Cohen’s d = 0.58, 95%CI d [0.06, 1.08], F(1, 118) = 5.22, p = .024, η² = .042. By contrast, in the straight posture condition, the valence of memories were similar in the reappraisal and spontaneous regulation conditions, MD = 2.90, Cohen’s d = 0.20, 95%CI d [−0.31, 0.70], F(1, 118) = 0.55, p = .461, η² = .005. Thus, in contrast to our expectations – that cognitive regulation could amplify or weaken the posture effects – the posture manipulation seemed to moderate the regulation effect. More specifically, the results indicate that cognitive reappraisal led to more positive autobiographical recall than spontaneous regulation, but only when participants assumed a stooped posture. However, regulation did not affect implicit mood when participants assumed a straight posture. For a visual display of the means, see Figure 7.

Figure 7. Posture and regulation effect on valence of memory. Mean valence of the recalled memories for the different body posture conditions (straight vs. stooped), and regulation conditions (reappraisal versus spontaneous) with standard error bars (1-SE) (Experiment 2).⁶

Discussion

Replicating the main findings of Experiment 1, participants who assumed a stooped posture recovered less from their negative mood than participants assuming a straight posture. In addition, Experiment 2 yielded the novel finding that posture had this effect regardless of whether participants engaged in cognitive reappraisal. It thus appears that the effects of posture on mood operated through a different mechanism than cognitive reappraisal. Posture might thus be regarded as an alternative channel through which people can regulate their emotional states, a channel that operates in parallel to cognitive forms of emotion regulation.

Unexpectedly, participants who engaged in cognitive reappraisal recovered less from their negative mood than participants who received no specific regulation instructions. The ineffectiveness of reappraisal in our study might be explained by prior findings in the reappraisal literature showing that the effects of reappraisal are context-sensitive. First, reappraisal works better when participants are instructed before than after confrontation with emotional stimuli or an emotion induction (Sheppes & Meiran, 2007). Second, research has found that reappraisal is more effective when emotional intensity is low rather than high (Sheppes, Scheibe, Suri, & Gross, 2011). Both of the latter limiting conditions were present in our experiment; (1) we instructed participants to reappraise after the emotion induction instead of before the emotion induction; (2) our mood induction involved imagining the death of a close friend, which is an intensely negative event for anyone. A third constraint on the effectiveness of cognitive reappraisal is formed by people’s inventiveness and flexibility in generating new ways of looking at the emotion-eliciting situation (Weber, Loureiro de Assunção, Martin, Westmeyer, & Geisler, 2014). It is conceivable that our emotion induction, which involved imagining the death of a close friend, afforded relatively few possibilities for generating alternative appraisals.

Although there was no effect of regulation strategy on participants’ self-reported mood, regulation strategy did affect autobiographical recall, depending on which posture participants assumed. Specifically, reappraisal resulted in more positive recall than spontaneous regulation when participants adopted a stooped, but not when they adopted a straight posture. Although we can only speculate as to why this might be the case, one explanation may be that sitting in a stooped position, in combination with the difficulty of thinking of positive thoughts, might have upset participants during the recovery period. Consequently, this might have motivated participants in the reappraisal condition to recall more positive memories in the subsequent task. Of course, the latter explanation is tentative and requires further empirical testing.

General discussion

In the present research, we examined how body posture influences the generation and regulation of negative mood. Experiment 1 found that a stooped seating posture – compared to a straight or control posture – increased negative mood, consistent with earlier findings that body posture contributes to emotion generation (e.g. Michalak et al., 2014; Riskind & Gotay, 1982; Stepper & Strack, 1993). More importantly, Experiment 1 also yielded the new finding that stooped seating posture – compared to a straight or control posture – interferes with recovery from a negative mood induction. Thus, we established for the first time that body posture contributes to emotion regulation. Across mood induction conditions, stooped seating posture also caused more negative thoughts, but this effect was unrelated to mood. Experiment 2 replicated the effect of posture on mood recovery, and showed that posture affected mood recovery independent of whether people deliberately regulated their negative feelings through reappraisal or not. Thus, the present findings provided consistent evidence that posture can influence mood regulation.

We observed effects of body posture consistently in two experiments and across three different dependent measures, namely, mood, thought listing, and autobiographical recall. As such, the effects of body posture seem to be robust. Indeed, body posture influenced mood in Experiment 2, even when cognitive reappraisal led to relatively less mood recovery. Though cognitive reappraisal generally assists in emotion regulation (Webb et al., 2012), it is less effective in regulating more intense emotional states (Sheppes & Meiran, 2007), and requires some level of inventiveness (Weber et al., 2014). By comparison, changing one’s body posture is likely to be cognitively less demanding, and may be effective even for changing mood states that resist attempts at cognitive change. Consequently, our findings suggest that it may often be worthwhile for people to use their posture as a relatively efficient and robust way of regulating their moods.

An alternative explanation to our findings might be that the physical discomfort from a stooped posture induced negative mood and as such, interfering with subsequent mood recovery, was the driving force behind our findings. However, the subjective discomfort people reported as a result of their stooped body posture could not explain the posture effects on mood. Alternatively, posture might have primed negative mood, due to an existing association between stooped posture and negative mood (e.g. Riskind & Gotay, 1982). If so, then straight posture would lead to positive priming effects as well, because straight posture is related to feelings of power (Carney et al., 2010) and pride (Oosterwijk et al., 2009). If this were true, we would expect that sitting straight would lead to more effective mood recovery than a control posture. However, we did not find a facilitating effect of a straight posture compared to a control posture, which might be explained by the less profound difference between the control and straight body posture.

The priming explanation of the observed posture effects is in accordance with theories of embodiment, and a situated conceptualisation of emotion regulation (Feldman Barrett et al., 2014; Koole & Veenstra, 2015). More specifically, because of the bidirectional relationship between bodily processes and emotional representations (e.g. Neumann & Strack, 2000; Niedenthal, Winkielman, Mondillon, & Vermeulen, 2009; Riskind, 1984; Strack et al., 1988), the body can work as an affordance of an emotional experience. Recall, that affordances refer to the fit between people’s emotion-regulatory goals and the possibilities that the current situation provides for meeting these goals. The present results suggest that posture can indeed work as an affordance of the emotion regulation goal, that is, the automatic tendency to assume a stooped posture when feeling negative, potentially maintains or even strengthens negative mood, working as a circular feedback loop between the body and the experienced affective state. Consequently, stooped posture might inhibit effective mood recovery or regulation, because a stooped posture is incongruent with the desired neutral or positive state.

The present findings are the first to explore the impact of body posture and mood regulation. As such, this work opens up many questions for future inquiry. One question pertains to the effect of straight body posture. If a stooped posture inhibits mood regulation because it is incongruent with the desired state, one might expect that a straight posture facilitates emotion regulation because it is congruent with the desired state. However, as mentioned earlier, we did not find this facilitating effect of straight posture. One reason for this might be that the control condition (self-chosen) in our study was very similar to the explicit straight posture condition, suggesting that sitting straight might be a default. Consequently, sitting stooped provides a clear change in bodily signals, increasing its influence on mood recovery. However, because the straight posture condition was close to the default, it may have produced a far weaker bodily signal, reducing its influence on mood recovery. Future work might find conditions under which straight posture does facilitate mood recovery.

Another unanswered question is whether and how body posture interacts with other deliberate forms of emotion regulation, such as distraction (Van Dillen & Koole, 2007), or expressive suppression (Butler et al., 2003). Future research could look into the contextual conditions in which different emotion regulation strategies and body posture might interact more strongly. For example, for physically oriented strategies such as expressive suppression, body postures might even have more profound effects than during cognitive strategies such as reappraisal. These future directions are particularly relevant with regard to therapeutic purposes. For instance, combining regulation strategies with less stooped body postures might help people with depression to decrease their persistent depressive feelings.

Emotion theorists have long maintained that sensory-motor processes make a vital contribution to people’s emotional lives (e.g. Adelmann & Zajonc, 1989; Damasio, 1994; James, 1890; Niedenthal, 2007). So far, however, the effects of bodily processes on emotion processing have been only investigated in relation to emotion generation, but not in relation to emotion regulation. The present study demonstrates for the first time that negative mood is not just associated with a stooped body posture, stooped body posture also resulted in relatively less effective negative mood regulation than straight or control postures. The effects of body posture were largely independent of the effects of cognitive reappraisal, indicating that bodily processes make an independent contribution to mood regulation. Furthermore, the present results show that sensory-motor processes are not just side effects, but rather are vital in instantiating and regulating a desired emotional state, and thus to the effectiveness and efficiency of emotion regulation. The next time you are feeling down, consider that it may be your stooped posture that keeps your spirits from going up.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCiD

Lotte Veenstra http://orcid.org/0000-0003-2725-0408

Additional information

Funding

This work was supported by a Consolidator Grant from the European Research Council [grant number ERC-2011-StG_20101124], awarded to Sander L. Koole.

Notes

1. The present research also included additional measures for exploratory reasons, which are further elaborated on in the online supplemental materials.

2. Further results of the mixed-model ANOVA showed an overall main effect of time, F(2, 446)  = 22.87, p < .0001, η² = .093. Contrast analyses showed that negative mood significantly increased from before the mood induction (t1) to after the mood induction (t2), F(1, 223) = 45.20, p < .0001, η² = .169, and decreased from before the recovery period (t2) to after the recovery period (t3), F(1, 223) = 10.28, p = .002, η² = .044. In addition, following up on the overall two-way interaction between mood condition and time, contrast analyses showed that during the recovery period (t2–t3) the decrease in negative mood was larger in the negative mood condition, Cohen’s d_z = 0.42, 95%CI d_z [0.23, 0.61], than in the neutral mood condition, Cohen’s d_z = 0.14, 95%CI d_z [−0.04, 0.33], F(1, 223) = 22.44, p < .0001, η² = .091. This indicates that participants that imagined a negative situation recovered from their negative mood during the spontaneous regulation period. Overall, participants who imagined a neutral situation showed no mood changes over time. Finally, the three-way interaction between mood manipulation, body posture, and time was not significant, F(4, 446) = 0.32, p = .867, η² = .003.

3. In addition to the proportion of negative thoughts, we also analysed the proportion of positive and neutral thoughts. There was no body posture effect on the proportion of positive or neutral thoughts. There was only one main effect of mood condition on proportion of neutral thoughts, F(2, 217) = 5.46, p = .005, η² = .05. Participants in the negative mood condition (M = 0.34, SD = 0.22) reported a larger proportion of neutral thoughts than participants in the control condition (M = 0.27, SD = 0.21), t(221) = 2.40, p = .017, 95%CI MD [0.01; 0.13], Cohen’s d = 0.33, 95%CI d [0.05, 0.59].

4. Planned contrasts showed that there was also a main effect of time during the recovery period (t2–t3), F(1, 118) = 19.82, p < .0001, η² = .144, participants’ negative mood decreased during the regulation period.

5. Note that although the mood induction was similar for both regulation conditions, the level of negative mood after the induction was higher for participants in the spontaneous than in the reappraisal condition (see Table 3 and Figure 6), which indicates that the recovery in the reappraisal condition might be lower due to their lower level of negative affect. However, we suggest that other factors are more likely to have caused the ineffectiveness of reappraisal. We discuss this issue further in the Discussion of Experiment 2.

6. When inspecting the pattern of results in Figure 7, it seems as if type of regulation strategy also moderated the effect of posture, such that the effect of posture on mood-congruent recall was inhibited in the cognitive regulation condition, but not in the spontaneous regulation condition. Indeed, in the reappraisal condition, participants with a stooped posture did not report significantly more negative memories, than participants in the straight posture condition, MD = 4.70, Cohen’s d = 0.30, 95%CI d [−0.21, 0.80], F(1, 118) = 1.44, p = .232, η² = .012. However, the effect of posture in the spontaneous regulation condition was also not very strong, participants with a stooped posture reported marginally significant more negative memories, than participants with a straight posture, MD = 7.13, Cohen’s d = 0.46, 95%CI d [−0.03, 0.99], F(1, 118) = 3.32, p = .071, η² = .027. Therefore, we can only carefully conclude that cognitive reappraisal eliminated the effect of posture on autobiographical recall.