The Power of Suggestion: Posthypnotically Induced Changes in the Temporal Binding of Intentional Action Outcomes

Abstract

The sense of agency is the experience of initiating and controlling one’s voluntary actions and their outcomes. Intentional binding (i.e., when voluntary actions and their outcomes are perceived to occur closer together in time than involuntary actions and their outcomes) is increased in intentional action but requires no explicit reflection on agency. The reported experience of involuntariness is central to hypnotic responding, during which strategic action is experienced as involuntary. We report reduced intentional binding in a hypnotically induced experience of involuntariness, providing an objective correlate of reports of involuntariness. We argue that this reduced binding results from the diminished influence of motor intentions in the generation of the sense of agency when beliefs about whether an action is intended are altered. Thus, intentional binding depends on awareness of intentions. This finding shows that changes in metacognition of intentions affect perception.

Keywords

sense of agency intentional binding hypnosis metacognition open data

The sense of agency is the experience of initiating and controlling voluntary actions and their outcomes (see Haggard & Eitam, 2015). Intentional binding refers to the subjective compression of time between an intentional action and its outcome: People perceive voluntary actions to occur later than they actually do (action binding) and their outcomes to occur earlier than they actually do (outcome binding; Haggard, Clark, & Kalogeras, 2002). The effect is sensitive to intentional action but requires no explicit reflection on agency and may reflect the additional contribution of intentions to causal binding (Buehner, 2012, 2015). Intentional binding has been shown to be affected in a number of disorders of agency—for example, schizophrenia (e.g., Voss et al., 2010) and alien-limb syndrome (Wolpe et al., 2014)—and to be reduced in coerced action (Caspar, Christensen, Cleeremans, & Haggard, 2016).

The “classical suggestion effect” of hypnosis is the experience of involuntariness of an action (Weitzenhoffer, 1974, p. 258), and changes in the sense of agency are central to hypnotic responding (Polito, Barnier, & Woody, 2013). The sense of agency may arise from the integration of internal and external predictive and retrospective cues (Moore & Fletcher, 2012; Synofzik, Vosgerau, & Voss, 2013) and from general beliefs about agency. Indeed, retrospectively manipulating beliefs about agency can alter attributions of agency. Hypnotic involuntariness may therefore reflect a relatively strong weighting of beliefs about hypnosis and a relatively weak weighting of the internal signals provided by motor intentions.

However, participants with high hypnotizability might merely report that a hypnotically suggested movement feels involuntary—even though they may find the experience of the action to be similar to that of any other voluntary action. If so, phenomena that are sensitive to conscious intentions, such as intentional binding, should be normal following hypnosis. Alternatively, if movement under hypnosis represents a shift from relying on internal action signals to relying on experimenter-delivered beliefs about action, then implicit measures that are sensitive to the experience of intentional action might be altered during response to posthypnotic suggestion among highly hypnotizable participants. Beliefs about whether or not one is the cause of an outcome have been shown to influence intentional binding (Desantis, Roussel, & Waszak, 2011). In the current study, we addressed for the first time whether binding is influenced by beliefs about whether an action was intended.

Method

Two hundred sixty-six people were screened for susceptibility to hypnosis using the Sussex Waterloo Susceptibility to Hypnosis Scale (SWASH), which is based on the Waterloo-Stanford Group Scale of Hypnotic Susceptibility, Form C (WSGC; Bowers, 1993). The SWASH is a modified, 10-item version of the WSGC from which age regression and dream suggestions have been removed to avoid participants’ becoming absorbed in negative experiences (Cardeña & Terhune, 2009). The SWASH not only elicits the objective ratings of the WSGC, but also includes a subjective-experience rating for each suggestion. For example, the following is the subjective-rating text for Item 2, “Moving hands together”:

You were next told to hold your hands out in front of you about a foot apart and then told to imagine a force pulling your hands together. On a scale from 0 to 5, how strongly did you feel a force between your hands, where 0 means you felt no force at all and 5 means you felt a force so strong it was as if your hands were real magnets?

Participants

Eighteen participants (4 men, 14 women; mean age = 20.2 years, SD = 2.35) were selected on the basis of their high scores on the SWASH. As requested by the reviewers, a second group of 14 participants was later selected on the basis of their medium scores on the SWASH (4 men, 10 women; mean age = 23.4 years, SD = 5.2). Participants were selected on the basis of their combined hypnotizability score (the simple mean of the objective and subjective scores, each scored on a scale of 0–10), with a minimum cutoff of 5 (which constituted the top 11% of the 266 people screened) for the high-hypnotizability group and a score between 2 and 4.99 for the medium-hypnotizability group (15% of SWASH scores were below 2).

To assess whether participants were able to maintain an experience of involuntariness for the duration of the task, we recorded verbal ratings of involuntariness on a scale from 0 (completely voluntary) to 5 (completely involuntary) after each block of trials. We excluded 7 participants from the high-hypnotizability group who reported full voluntariness (i.e., an involuntariness score of 0) after any block in the posthypnotic-involuntariness condition (see Procedure section for full information about conditions). Two of these participants did not complete all conditions and therefore provided insufficient data for comparisons. Because the aim was to determine an objective correlate (e.g., intentional binding) of reported feelings of involuntariness, only cases in which there were feelings of involuntariness were relevant for the high-hypnotizability group. Analyses of the results for all participants in the high-hypnotizability group (whether or not they were able to sustain the experience of involuntariness) are shown in Table S4 in the Supplemental Results in the Supplemental Material available online.

The combined hypnotizability scores of participants unable to sustain the suggestion was lower (M = 5.98, SD = 1.11) than those of participants who maintained involuntariness (M = 7.48, SD = 1.24), t(16) = 2.61, p = .019, BF_H[0,3.74] = 3.16.¹ The participants in the medium-hypnotizability group had a mean combined hypnotizability score of 3.19 (SD = 0.88), and none were able to sustain an experience of involuntariness throughout the experiment. One participant from the high-hypnotizability group was excluded because the standard deviation of his baseline-action judgments was more than 3 times the group interquartile range in the passive condition (614.9 ms) and posthypnotic-involuntariness condition (470.2 ms). Therefore, data from 10 participants with high hypnotizability (1 man, 9 women; mean age = 20 years, SD = 1.9) are reported.

Participants with high hypnotizability were recruited during two academic terms, until the participant pool was exhausted or there were no further responses to recruitment e-mails. Participants with medium hypnotizability were recruited during the summer break, until there were no more responses. We used Bayesian analyses to assess the sensitivity of the evidence for Hypothesis 1 (H1) versus the null hypothesis (H0); the measure of evidence is valid regardless of the stopping rule (Rouder, 2014; Schoenbrodt, Wagenmakers, Zehetleitner, & Perugini, 2015).

No power analysis was conducted. We included Bayes factors so that there would be an assessment of the sensitivity of the data to distinguish H0 and H1. Once the data are in, power has no relevance to how sensitive the data are, because power is a property of decision rule in the long run; conversely, Bayes factors indicate the sensitivity of the very data collected to distinguish H1 and H0.

Ethical approval was received from the University of Sussex ethics committee, and we obtained informed consent from the participants. Participants received a cash payment of £18 or course credits.

Procedure

Each session began with a hypnotic induction adapted from the WSGC (included in the Supplemental Method in the Supplemental Material). After the hypnotic induction, participants were given the suggestion that, for blocks of trials that followed a handclap, their fingers would move involuntarily onto the key. Participants were then “counted out” of hypnosis (see Supplemental Method) before performing the experimental task. There were three counterbalanced conditions. In the voluntary condition, participants pressed the key when they wished. In the passive condition, the experimenter pulled participants’ index fingers onto the key using a fabric loop, but the experimenter’s action was out of the participant’s view. In the posthypnotic-involuntariness condition, a single handclap was made approximately 20 s before the start of each 35-trial block (except in baseline-tone trials). After each block in each condition and after 3 trials of the first block of the posthypnotic-involuntariness condition, participants were asked to rate the involuntariness of the action on a scale from 0 to 5. No handclaps were delivered in the voluntary or passive conditions. Participants were informed during the hypnotic induction that the posthypnotic suggestion would be removed when they left the room at the end of the session.

Visual stimuli were displayed at 100 Hz on a 21-in. CRT monitor, and auditory stimuli were presented via headphones (Sennheiser, Wedemark, Germany). Participants were seated at a viewing distance of approximately 60 cm. On each trial, a clock face with the standard hourly markings was presented, subtending a visual angle of 5°; 250 ms later, a dot subtending 0.2° appeared at a pseudorandomized position and began revolving around the clock. Each revolution took 2,560 ms. A computer keyboard was used to record actions (button presses).

There were four trial types, presented in separate blocks. In contingent-tone and contingent-action trials, pressing a key triggered a 1000-Hz, 100-ms tone after a 250-ms delay. In baseline-tone trials, no key was pressed; the tone was triggered pseudorandomly between 2.5 s and 7 s after one revolution of the clock. In baseline-action trials, there was a key press, but no tone.

Participants were asked to look at a fixation cross in the center of the clock and to wait for at least one revolution of the dot before pressing the button at a time of their choosing. The trial was restarted if the button was pressed before one full revolution or after six revolutions. Participants were asked not to plan ahead, not to aim for a particular point on the clock, and to report either the action or the tone. After the tone (or, on baseline-action trials, after the action), the dot continued moving for a pseudorandomized period of time between 1,200 ms and 2,370 ms. The clock was then removed from the screen for a pseudorandomized time interval (500–1,280 ms). When the clock reappeared, participants were able to control the position of the dot using a mouse and were asked to move the dot to match its location at the time of the action or tone and to press the mouse button to record their judgment.

Each block consisted of 35 repetitions of one trial type, except for baseline-tone trials; there was a block of 13 baseline-tone trials in each condition (i.e., voluntary, passive, and posthypnotic involuntariness), and they were subsequently combined into a single block of 39 trials. The baseline-tone trials were spread across the conditions in this way to minimize the experiments’ duration and to reduce the possibility of participants becoming fatigued. Because the baseline-tone trials required no action to take place, the different experimental conditions should not influence these timing judgments. Blocks were separated by 30-s rest periods and presented in counterbalanced order. Before the session began, all participants were trained with 4 practice trials in the baseline-tone blocks and 4 practice trials in the baseline-action blocks. To reduce the effects of fatigue, the experimental task was split across two experimental sessions, with two conditions performed in the first session and one in the second. Participants were led through the hypnotic induction and count-out procedure at the start of each session. Sessions took place on separate days or after a gap of at least 2 hr. In total, the sessions took approximately 2 hr 30 min, including training and debriefing. All stimuli were generated with MATLAB (Version R2014a; The MathWorks, Natick, MA) running Psychophysics Toolbox (Version 3; Kleiner, Brainard, & Pelli, 2007).

Analysis

Mean judgment errors (measured in milliseconds) were calculated for each group on each trial type by subtracting the time of the action or tone event from the judged time of the event. Individual judgments more than 3.5 SD away from the mean for each participant on each judgment type were then excluded before mean judgment errors were calculated for each participant, as specified in Lush, Parkinson, and Dienes (2016). Twenty judgments were excluded across all participants and trials (0.52% of judgments). Participants’ judgments in the baseline-action and baseline-tone blocks in each condition were subtracted from their judgments in the contingent-action and contingent-tone blocks, respectively, in each condition to calculate action and outcome binding. Outcome binding was subtracted from action binding to produce a total-binding measure.

Repeated measures analyses of variance were performed for the action, outcome, and total-binding measures. Mean baseline-action judgments and within-participant standard deviations of baseline-action judgments were also compared. Where there was evidence for violation of sphericity, Greenhouse-Geisser corrected degrees of freedom were used. Significant F values were followed up with Fisher’s least-significant-difference post hoc comparisons.

Bayes factors (BFs) were used to assess strength of evidence for effects with 1 degree of freedom. A BF above 3 indicates substantial evidence for the alternative hypothesis, a BF below 1/3 indicates substantial evidence for the null, and a BF between 1/3 and 3 indicates data insensitivity (see Dienes, 2014; cf. Jeffreys, 1939). Thus, all Bayes factors reported here represent the evidence for H1 relative to H0; to find the evidence for H0 relative to H1, take 1/BF. BF_H(0,x) refers to a Bayes factor in which the predictions of H1 were modeled as a half-normal distribution with a standard deviation of x (see Dienes, 2014); the half-normal can be used when a theory makes a directional prediction in which x scales the size of effect that could be expected (so x can be chosen from, e.g., relevant past studies). BF_N(0,x) indicates that H1 was specified as a normal distribution with mean of 0 and a standard deviation of x (for nondirectional predictions).

Since the 19th century, researchers have proposed that a shared mechanism underlies functional motor disorders (motor disorders with no known neurological cause) and hypnotic involuntariness (for a recent review, see Bell, Oakley, Halligan, & Deeley, 2011). Kranick et al. (2013) provided an estimated effect size for the difference in intentional binding between functional motor-disorder patients and healthy volunteers; the difference between groups in outcome binding was approximately half the effect found in control participants. Bayes factors for differences in each measure were therefore calculated using a half-normal distribution with standard deviation based on half the mean in the voluntary condition. BF_U[0,max] refers to a Bayes factor in which the predictions of H1 were modeled as a uniform distribution from 0 to max. We used this model for the rating of involuntariness, which is on a scale from 0 to 5; thus the maximum possible population mean difference between conditions was 5. A Bayes factor for the regression of the difference in outcome binding between the voluntary and posthypnotic-involuntariness conditions on reported involuntariness in the posthypnotic-involuntariness condition was calculated using a half-normal distribution with a standard deviation that was based on the quotient of the mean outcome binding in the medium group (as an independent estimate of the rough amount of binding that could exist in highs) and the range of the involuntariness rating scale (i.e., 120/6). Bayes factors for simple interactions between two conditions and group were calculated by modeling H1 using half the mean binding in both groups for the relevant binding component.

Predictions

We tested participants with high hypnotizability and medium hypnotizability on an intentional binding task in one voluntary condition and two involuntary conditions; in the latter conditions, the action was passive or was reported to be experienced as involuntary following a posthypnotic suggestion (in which response occurs following hypnosis; Barnier & McConkey, 1998) of action involuntariness. Because binding is sensitive to agency, binding should be stronger in the voluntary condition and weaker in the passive condition. If the experience of involuntariness reported in hypnotic responding by highly hypnotizable participants reflects real changes in the experience of agency, intentional binding should also be weakened in posthypnotic involuntariness in highly hypnotizable participants. Compared with participants with medium hypnotizability, those with high hypnotizability should have should have a greater difference in binding between the voluntary and posthypnotic-involuntariness conditions and between the passive and posthypnotic-involuntariness conditions. We make no prediction regarding whether participants with high hypnotizability are different from participants with medium hypnotizability in the difference between the voluntary and passive conditions.

Results

Involuntariness ratings

Table 1 shows mean involuntariness ratings for each hypnotizability group in each condition. The effects of hypnotizability on reported involuntariness were analyzed using hypnotizability (high vs. medium) as a between-participants factor and condition (voluntary vs. posthypnotic involuntariness vs. passive) as a within-participants factor. Note that there was a significant interaction between condition and group on reported involuntariness, F(1, 22) = 50.85, p < .001, η_p² = .698. The interaction was decomposed into the simple effect of condition for each hypnotizability group. For the high-hypnotizability group, there was a significant effect of condition on involuntariness, F(2, 18) = 135.2, p < .001, η_p² = .94. Participants reported more involuntariness in the passive condition than in the voluntary condition, p < .001, BF_U[0,5] = 3.84 × 10²⁸, 95% confidence interval (CI) for the difference between conditions = [−4.98, −3.62], d_z = 5.53, and more involuntariness in the passive condition than in the posthypnotic-involuntariness condition, p < .001, BF_U[0,5] = 3.53 × 10¹¹, 95% CI for the difference between conditions = [−3.62, −2.23], d_z = 3.01. However, action in the passive condition was reported to be more involuntary than action in the posthypnotic-involuntariness condition, p < .001, BF_U[0,5] = 2.15 × 10⁸, 95% CI for the difference between conditions = [0.99, 1.76], d_z = 2.53. For the medium-hypnotizability group, there was a significant effect of condition on involuntariness, F(2, 26) = 413.08, p < .001, η_p² = .97. Participants reported less involuntariness in voluntary action than in passive action, p < .001, BF_U[0,5] = 6.95 × 10¹²⁵, 95% CI for the difference between conditions = [−5.00, −4.28], d_z = 7.33. This provides evidence that there was no difference between voluntariness ratings in the voluntary and posthypnotic-involuntariness conditions, p > .250, BF_U[0,5] = 0.10, 95% CI for the difference between conditions = [0.53, 0.24], d_z = .22. Actions in the passive condition were rated as more involuntary than actions in the posthypnotic-involuntariness condition, p < .001, BF_U[0,5] = 2.43 × 10⁸⁴, 95% CI for the difference between conditions = [4.06, 4.94], d_z = 5.92.

Table 1.

Mean Involuntariness Ratings for the High- and Medium-Hypnotizability Groups in the Three Experimental Conditions

		Condition
Group	Voluntary	Posthypnotic involuntariness	Passive
High hypnotizability	0.7 (0.30)	3.3 (0.17)	5 (0.00)
Medium hypnotizability	0.3 (0.16)	0.4 (0.20)	4.9 (0.07)

Note: The rating scale ranged from 0, completely voluntary, to 5, completely involuntary. Values in parentheses are standard errors.

Total binding

Analyses of the total binding measure are reported in the Supplemental Results in the Supplemental Material.

Outcome binding

Table 2 shows the binding measures in each condition for both hypnotizability groups, and Table 3 shows p values, Bayes factors, 95% CIs, and effect sizes of post hoc comparisons for each main effect. The effects of hypnotizability on outcome binding were analyzed using hypnotizability (high vs. medium) as a between-participants factor and condition (voluntary vs. posthypnotic involuntariness vs. passive) as a within-participants factor. There was a significant main effect of condition on outcome binding, corrected F(1.42, 31.30) = 10.30, p = .001, η_p² = .319, but no significant main effect of hypnotizability on this measure, F(1, 22) = .929, p > .250, η_p² = .041. There was a marginally significant interaction between condition and group on outcome binding, corrected F(1.42, 31.30) = 2.81, p = .091, η_p² = .11. The theory that hypnotic response is experienced as passive predicts two key partial interactions. Specifically, as predicted, when only the voluntary and the posthypnotic-involuntariness conditions were compared, there was a sensitive interaction effect of group and condition on outcome binding, F(1, 22) = 9.18, p = .006, BF_H(0,62.5) = 39.01, η_p² = .29. There was no evidence one way or the other for a predicted interaction between the passive and posthypnotic-involuntariness conditions, F(1, 22) = 0.222, p > .250, BF_H(0,62.5) = 0.67, η_p² = .01. Finally, there was no sensitive evidence for an interaction between group and condition on outcome binding when only the voluntary and passive conditions were compared, F(1, 22) = 3.52, p = .074, BF_N(0,62.5) = 1.63, η_p² = .14.

Table 2.

Mean Binding for the High- and Medium-Hypnotizability Groups in the Three Experimental Conditions

	Condition
Group and binding measure	Voluntary	Posthypnotic involuntariness	Passive
High hypnotizability
Action binding	28.0 (25.0)	24.6 (56.7)	23.2 (53.7)
Outcome binding	−130.7 (45.4)	−69.4 (56.0)	−50.5 (89.1)
Medium hypnotizability
Action binding	9.9 (28.2)	25.6 (38.4)	3.1 (42.9)
Outcome binding	−120.2 (66.3)	−117.9 (67.9)	−83.5 (79.7)

Note: Values are mean times in milliseconds. Values in parentheses are standard deviations.

Table 3.

Post Hoc Comparisons of the Binding Measures of the High- and Medium-Hypnotizability Groups

Binding measure and comparison	p	BF	95% CI for the difference between conditions	d_z
High-hypnotizability group
Action binding
Voluntary vs. passive	> .250	BF_H(0,14) = 0.91	[–29.3, 38.8]	0.10
Voluntary vs. posthypnotic involuntariness	> .250	BF_H(0,14) = 0.89	[–32.9, 39.7]	0.07
Passive vs. posthypnotic involuntariness	> .250	BF_H(0,14) = 0.93	[–59.4, 56.8]	0.02
Outcome binding
Voluntary vs. passive	.003	BF_H(0,65) > 79.51	[–124.4, –35.9]	1.08
Voluntary vs. posthypnotic involuntariness	.009	BF_H(0,65) > 14.70	[–103.4, –19.2]	0.93
Passive vs. posthypnotic involuntariness	> .250	BF_H(0,65) = 0.76	[–59.0, 96.8]	0.17

Medium-hypnotizability group
Action binding
Voluntary vs. passive	> .250	BF_H(0,5) = 1.08	[–16.8, 30.5]	0.17
Voluntary vs. posthypnotic involuntariness	> .250	BF_H(0,5) = 0.77	[–44.3, 12.8]	0.31
Passive vs. posthypnotic involuntariness	.121	BF_H(0,5) = 1.33	[–52.0, –6.9]	0.43
Outcome binding
Voluntary vs. passive	.019	BF_H(0,60) = 7.07	[–66.4, –7.0]	0.67
Voluntary vs. posthypnotic involuntariness	> .250	BF_H(0,60) = 0.22	[–23.5, 18.8]	0.06
Passive vs. posthypnotic involuntariness	.022	BF_H(0,60) = 6.11	[–62.8, –5.9]	0.65

Note: Bayes factors (BFs) > 3 or < 1/3 are sensitive (indicated in the table by boldface type).

We conducted a planned simple-effects analysis. For the high-hypnotizability group, the simple effect of condition on outcome binding was significant, corrected F(1.15, 10.37) = 5.50, p = .037, η_p² = .38. Outcome binding was lower in the passive and posthypnotic-involuntariness conditions than in the voluntary condition. For the medium-hypnotizability group, the simple effect of condition on outcome binding was significant, F(2, 24) = 5.52, p = .010, η_p² = .30. Outcome binding was higher in the voluntary and posthypnotic-involuntariness conditions than in the passive condition. There was sensitive evidence for no difference in outcome binding between the voluntary and posthypnotic-involuntariness conditions.

Action binding

The effects of hypnotizability on action binding were analyzed using hypnotizability (high vs. medium) as a between-participants factor and condition (voluntary vs. passive vs. posthypnotic involuntariness) as a within-participants factor. There was no significant main effect of condition on action binding, F(2, 44) = 0.579, p > .250, η_p² = .026, nor was there a significant effect of group, F(1, 22) = 1.165, p > .250, η_p² = .050.

The interaction between condition and group on action binding was also not significant, F(2, 44) = 0.579, p > .250, η_p² = .03. The more precise partial interactions were all nonevidential; no conclusions were possible. Specifically, comparison of the voluntary and posthypnotic-involuntariness conditions revealed only insensitive evidence for an interaction effect of group and condition on action binding, F(1, 22) = 0.859, p > .250, BF_H(0,19) = 1.19, η_p² = .038. The same was true for analyses comparing the voluntary and passive conditions, F(1, 22) = 0.013, p > .250, BF_N(0,19) = 0.68, η_p² = .001, and the passive and posthypnotic-involuntariness conditions, F(1, 22) = 0.623, p > .250, BF_H(0,19) = 1.22, η_p² = .03. We conducted another planned simple-effects analysis, and the simple effect of condition was not significant for the high-hypnotizability group, F(1.30, 11.72) = 0.032, p > .250, η_p² = .004.

Although the action binding shifts in the voluntary condition for participants with high hypnotizability are comparable with other reported results (e.g., 20 ms reported in Haggard et al., 2002), we found no sensitive evidence for a difference in action binding between conditions to parallel the shift in outcome binding. However, as Table 2 shows, neither is there substantial evidence for no difference between any two conditions; the data are simply insensitive and provide support for neither the experimental hypothesis nor the null hypothesis. We can therefore draw no conclusions about action binding from the current results. The insensitivity is not surprising; as we found, outcome binding is typically a larger effect than action binding (e.g., Desantis et al., 2011; Kranick et al., 2013; Lush, Parkinson, & Dienes, 2016). Given that action binding is characterized by a smaller shift than tone binding, a larger sample might be required to reveal differences in this measure.

Figure 1 shows the interval between the action and tone events in each condition, derived from participants’ judgments. Because outcome binding was reduced but not eliminated in passive actions, these results are broadly consistent with evidence that intentional binding is a special case of a general causal binding (Buehner, 2012). Because actions in the passive condition were reported to be more involuntary than actions in the posthypnotic-involuntariness condition for highly hypnotizable participants, we would expect a difference in magnitude of binding between these two conditions. Table 2 shows that the mean values followed this expected pattern. However, because the comparisons between these two conditions were insensitive, we can draw no firm conclusions about this pattern of results (Table 3).

Fig. 1.

Derived time intervals between the action and tone events in the high-hypnotizability group in the voluntary, passive, and posthypnotic-involuntariness conditions. The top line indicates the actual length of time between the action (pressing a button) and the tone. The three lines below that indicate, for each condition, the mean times that participants judged to have passed between the action and the tone.

To investigate the relationship between the experience of involuntariness and binding, we first calculated a difference score for outcome binding between the voluntary and posthypnotic-involuntariness conditions. We then regressed this score on reported involuntariness in the posthypnotic-involuntariness condition alone. All participants with high or medium hypnotizability (including those excluded from other analyses because they were unable to maintain involuntariness) were included in this analysis.

Reported involuntariness predicted the difference in outcome binding between the voluntary and the posthypnotic-involuntariness conditions; the raw slope was 19 ms per rating unit, t(27) = 2.37, p = .025, BF_H(0,20) = 6.48. Therefore, outcome binding was reduced in the posthypnotic-involuntariness condition compared with the voluntary condition as reported involuntariness increased, which supports the hypothesis that binding difference is related to subjective experience.

Discussion

As in previous studies, causal binding was stronger for voluntary action than for passive action (Buehner, 2015; Haggard et al., 2002). Crucially, binding was also reduced in participants with high hypnotizability after a posthypnotic suggestion of involuntariness, providing evidence for hypnotically induced changes in sense of agency.

We found evidence for changes only in outcome binding. The prediction of the sensory outcome of an action may provide cues for sense of agency by comparing a predicted sensory outcome with the actual outcome, and hypnotic suggestion may disrupt this mechanism by preventing motor intentions from activating sensorimotor predictions (Blakemore, Oakley, & Frith, 2003). Therefore, reduced outcome binding may arise from disrupting the formation or use of the sensorimotor prerepresentation of the action outcome.

An alternative account proposes that, by analogy with cross-modal cue combination (Ernst & Banks, 2002; Körding et al., 2007), the timing judgments of intentional actions and their outcomes may be a weighted average of the action and outcome cues (Kawabe, Roseboom, & Nishida, 2013); the weighting would depend on the estimated precision with which each cue is individually timed. The decreased outcome-judgment shift reported here may therefore arise from the increased weighting of the outcome cue over the action cue in estimating the time of the outcome event when motor intention information is discounted and the estimated precision of the action cue consequently decreases. (In the Supplemental Results, we report lower within-participants standard deviations in the voluntary condition than in the posthypnotic-involuntariness condition for participants with high hypnotizability, a finding that is consistent with cue combination. We also report sensitive evidence of no within-participants difference for participants with medium hypnotizability.) This would occur in the passive condition because motor intention information is absent and in the posthypnotic-involuntariness condition because hypnotist-induced beliefs reduce the relative weighting of motor intentions in generating a sense of agency. A cue-combination mechanism is predictive of an increase in action binding when motor-intention information is reduced, because lower precision of action should result in a relatively higher weighting of the outcome cue in outcome-timing judgments and consequently a greater shift of the weighted average of the two events toward the action cue. This might run contrary to our prediction of reduction in overall binding, because the two opposing shifts would, to at least some degree, cancel each other out. However, because we report no sensitive evidence for differences in action binding, the results of the current study do not bear on this prediction either way.

Although the current study is the first to show the relevance of beliefs about intentions for binding, outcome binding is also reduced when participants incorrectly believe that an outcome is triggered by another’s action (Desantis et al., 2011). This may reflect a reduced contribution of motor intentions to outcome timing judgments when, in accordance with participants’ beliefs, such information is not relevant to event timing. Binding has also been shown to be reduced when participants are instructed to press a particular key at a particular time (Caspar et al., 2016). By contrast, in the current study, participants were free to press the button when they wished and were merely instructed that they would not feel that they had intended the action.

It might be argued that hypnotic responding occurs in the absence of intentions (e.g., Woody & Bowers, 1994). However, given that hypnotic actions are performed in appropriate and flexible ways, intentions appear to be undisrupted in hypnotic responding; rather, it is the metacognition of intentions that is disrupted (e.g., see Woody & Sadler, 2008). Thus, the difference between a hypnotic action and a nonhypnotic action may lie in the awareness of intentions (Dienes, 2012; Lush, Naish, & Dienes, 2016). Accordingly, if an intention becomes conscious, that may increase its availability to other processes (Cleeremans & Jiménez, 2002) and thus to the process of timing its associated action. Consistent with this idea, mindfulness meditators, who may have more accurate metacognition of motor intentions (Dreyfus, 2011), show stronger outcome binding (Lush, Parkinson, & Dienes, 2016). It should be noted that highly hypnotizable people are a highly selected group, and these results may not generalize to the general population.

We report that hypnotically suggested actions behave more like genuinely involuntary actions than like voluntary actions in an implicit measure sensitive to agency. This finding provides objective evidence for hypnotically suggested changes in agentic experience and demonstrates that beliefs about whether an action is intended influence binding.

Footnotes

Action Editor

Marc J. Buehner served as action editor for this article.

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Funding

P. Lush was supported by Belgian Federal Science Policy Office (BELSPO) Interuniversity Poles of Attraction Grant P7/33 and The Sackler Centre for Consciousness Science. P. Haggard was supported by European Research Council (ERC) Advanced Grant 323943 (Human Volition, Agency and Responsibility Project). E. A. Caspar was supported by the Belgian National Fund for Scientific Research (FRNS)–Fonds de la Recherche Scientifique. A. Cleeremans is a Research Director with the FRNS. A. Cleeremans was supported by BELSPO Interuniversity Poles of Attraction Grant P7/33 and by ERC Advanced Grant “Radical.”

Supplemental Material

Additional supporting information can be found at

Open Practices

All data have been made publicly available via the Open Science Framework and can be accessed at https://osf.io/abq5f/. The complete Open Practices Disclosure for this article can be found at https://journals-sagepub-com.web.bisu.edu.cn/doi/suppl/10.1177/0956797616687015. This article has received the badge for Open Data. More information about the Open Practices badges can be found at .

Notes

References

Barnier

A. J.

McConkey

K. M.

(1998). Posthypnotic responding away from the hypnotic setting. Psychological Science, 9, 256–262.

Bell

Oakley

D. A.

Halligan

P. W.

Deeley

(2011). Dissociation in hysteria and hypnosis: Evidence from cognitive neuroscience. Journal of Neurology, Neurosurgery & Psychiatry, 82, 332–339.

Blakemore

S. J.

Oakley

D. A.

Frith

C. D.

(2003). Delusions of alien control in the normal brain. Neuropsychologia, 41, 1058–1067.

Bowers

K. S.

(1993). The Waterloo-Stanford Group C (WSGC) scale of hypnotic suggestibility: Normative and comparative data. International Journal of Clinical and Experimental Hypnosis, 41, 35–46.

Buehner

M. J.

(2012). Understanding the past, predicting the future: causation, not intentional action, is the root of temporal binding. Psychological Science, 23, 1490–1497.

Buehner

M. J.

(2015). Awareness of voluntary and involuntary causal actions and their outcomes. Psychology of Consciousness: Theory, Research, and Practice, 2, 237–252.

Cardeña

Terhune

D. B.

(2009). A note of caution on the Waterloo-Stanford Group Scale of hypnotic susceptibility: A brief communication. International Journal of Clinical and Experimental Hypnosis, 57, 222–226.

Caspar

E. A.

Christensen

J. F.

Cleeremans

Haggard

(2016). Coercion changes the sense of agency in the human brain. Current Biology, 26, 585–592.

Cleeremans

Jiménez

(2002). Implicit learning and consciousness: A graded, dynamic perspective. In French

R. M.

Cleeremans

(Eds.), Implicit learning and consciousness: An empirical, computational and philosophical consensus in the making? (pp. 1–40). Hove, England: Psychology Press.

10.

Desantis

Roussel

Waszak

(2011). On the influence of causal beliefs on the feeling of agency. Consciousness and Cognition, 20, 1211–1220.

11.

Dienes

(2012). Is hypnotic responding the strategic relinquishment of metacognition? In Beran

Brandl

Perner

Proust

(Eds.), The foundations of metacognition (pp. 267–277). Oxford, England: Oxford University Press.

12.

Dienes

(2014). Using Bayes to get the most out of non-significant results. Frontiers in Psychology, 5, Article 781. doi:10.3389/fpsyg.2014.00781

13.

Dreyfus

(2011). Is mindfulness present-centred and non-judgmental? A discussion of the cognitive dimensions of mindfulness. Contemporary Buddhism, 12, 41–54.

14.

Ernst

M. O.

Banks

M. S.

(2002). Humans integrate visual and haptic information in a statistically optimal fashion. Nature, 415, 429–433.

15.

Haggard

Clark

Kalogeras

(2002). Voluntary action and conscious awareness. Nature Neuroscience, 5, 382–385.

16.

Haggard

Eitam

(2015). The sense of agency. New York, NY: Oxford University Press.

17.

Jeffreys

(1939). A theory of probability. New York, NY: Oxford University Press.

18.

Kawabe

Roseboom

Nishida

(2013). The sense of agency is action-effect causality perception based on cross-modal grouping. Proceedings of the Royal Society B: Biological Sciences, 280, Article 991. doi:10.1098/rspb.2013.0991

19.

Kleiner

Brainard

Pelli

(2007). What’s new in psychtoolbox-3 ? [Abstract]. Perception, 36(Suppl. 1), 14. doi:10.1177/03010066070360S101

20.

Körding

K. P.

Beierholm

W. J.

Quartz

Tenenbaum

J. B.

Shams

(2007). Causal inference in multisensory perception. PLoS ONE, 2(9), Article e943. doi:10.1371/journal.pone.0000943

21.

Kranick

S. M.

Moore

J. W.

Yusuf

Martinez

V. T.

LaFaver

Edwards

M. J.

. . . Hallett

(2013). Action-effect binding is decreased in motor conversion disorder: Implications for sense of agency. Movement Disorders, 28, 1110–1116.

22.

Lush

Naish

Dienes

(2016). Metacognition of intentions in mindfulness and hypnosis. Neuroscience of Consciousness, 2016(1), Article niw007. doi:10.1093/nc/niw007

23.

Lush

Parkinson

Dienes

(2016). Illusory temporal binding in meditators. Mindfulness, 7, 1416–1422.

24.

Moore

J. W.

Fletcher

P. C.

(2012). Sense of agency in health and disease: A review of cue integration approaches. Consciousness and Cognition, 21, 59–68.

25.

Polito

Barnier

A. J.

Woody

E. Z.

(2013). Developing the Sense of Agency Rating Scale (SOARS): An empirical measure of agency disruption in hypnosis. Consciousness and Cognition, 22, 684–696.

26.

Rouder

J. N.

(2014). Optional stopping: No problem for Bayesians. Psychonomic Bulletin & Review, 21, 301–308.

27.

Schoenbrodt

F. D.

Wagenmakers

E.-J.

Zehetleitner

Perugini

(2015). Sequential hypothesis testing with Bayes factors: Efficiently testing mean differences. Psychological Methods. Advance online publication. doi:10.1037/met0000061

28.

Synofzik

Vosgerau

Voss

(2013). The experience of agency: An interplay between prediction and postdiction. Frontiers in Psychology, 15(4), Article 127. doi:10.3389/fpsyg.2013.00127

29.

Voss

Moore

Hauser

Gallinat

Heinz

Haggard

(2010). Altered awareness of action in schizophrenia: a specific deficit in predicting action consequences. Brain, 133, 3104–3112.

30.

Weitzenhoffer

A. M.

(1974). When is an “instruction” an “instruction”? International Journal of Clinical and Experimental Hypnosis, 22, 258–269.

31.

Wolpe

Moore

J. W.

Rae

C. L.

Rittman

Altena

Haggard

Rowe

J. B.

(2014). The medial frontal-prefrontal network for altered awareness and control of action in corticobasal syndrome. Brain, 137, 208–220.

32.

Woody

E. Z.

Bowers

K. S.

(1994). A frontal assault on dissociated control. In Lynn

S. J.

Rhue

J. W.

(Eds.), Dissociation: Clinical and theoretical perspectives. New York, NY: Guilford Press.

33.

Woody

E. Z.

Sadler

(2008). Dissociation theories of hypnosis. In The Oxford handbook of hypnosis: Theory, research, and practice (pp. 81–110). Oxford, UK: Oxford University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.58 MB

0.09 MB

0.31 MB