Acta Psychologica 150 (2014) 129–135

Contents lists available at ScienceDirect

Acta Psychologica journal homepage: www.elsevier.com/ locate/actpsy

Balancing cognitive control: How observed movements influence motor performance in a task with balance constraints Julius Verrel ⁎, Nina Lisofsky, Simone Kühn Max Planck Institute for Human Development, Center for Lifespan Psychology, Lentzeallee 94, Berlin, Germany

a r t i c l e

i n f o

Article history: Received 19 December 2013 Received in revised form 12 April 2014 Accepted 12 May 2014 Available online 29 May 2014 PsycINFO classification: 2300 2330 2340 Keywords: Action control Perception Common coding Imitation Movement Balance

a b s t r a c t We investigated the influence of observed movements on executed movements in a task requiring lifting one foot from the floor while maintaining whole-body balance. Sixteen young participants (20–30 years) performed foot lift movements, which were either cued symbolically by a letter (L/R, indicating to lift the left/right foot) or by a short movie showing a foot lift movement. In the symbol cue condition, stimuli from the movie cue condition were used as distractors, and vice versa. Anticipatory postural adjustments (APAs) and actual foot lifts were recorded using force plates and optical motion capture. Foot lift responses were generally faster in response to the movie compared to the symbol cue condition. Moreover, incongruent movement distractors interfered with performance in the symbol cue condition, as shown by longer response times and increased number of APAs. Latencies of the first (potentially wrong) APA in a trial were shorter for movie compared to symbol cues but were not affected by cue-distractor congruency. Amplitude of the first APA was smaller when it was followed by additional APAs compared to trials with a single APA. Our results show that automatic imitation tendencies are integrated with postural control in a task with balance constraints. Analysis of the number, timing and amplitude of APAs indicates that conflicts between intended and observed movements are not resolved at a purely cognitive level but directly influence overt motor performance, emphasizing the intimate link between perception, cognition and action. © 2014 Published by Elsevier B.V.

1. Introduction Humans have a remarkable ability and tendency to imitate each other's actions, a capacity which is likely to be crucial for social interaction, skill and language acquisition and cultural evolution (Meltzoff & Moore, 1974; Miller & Dollard, 1941; Tomasello, Kruger, & Ratner, 1993). Imitation poses a complex computational problem, as it requires a transformation between two different domains, from a visual to a motor representation (Brass & Heyes, 2005). Yet, experimental evidence suggests that imitation may at least partly constitute an automatic process, as seeing another person's movement facilitates executing the same movement and interferes with performance of different movements, even if participants are instructed to ignore the other's action (Heyes, 2011). The present study extends previous research on automatic imitation tendencies, which mostly concerned relatively isolated movements of body parts (Brass, Bekkering, & Prinz, 2001; Gillmeister, Catmur, Liepelt, Brass, & Heyes, 2008; Leighton & Heyes, 2010; Stürmer, Aschersleben, & Prinz, 2000) to movements with whole-body balance constraints. Besides addressing the general ⁎ Corresponding author at: MPI for Human Development, Lentzeallee 94, 14195 Berlin, Germany. Tel.: +49 30 82406410. E-mail addresses: [email protected] (J. Verrel), [email protected] (N. Lisofsky), [email protected] (S. Kühn).

http://dx.doi.org/10.1016/j.actpsy.2014.05.010 0001-6918/© 2014 Published by Elsevier B.V.

question how automatic imitation tendencies interact with postural control, anticipatory postural adjustments (APAs), required for maintaining balance, can be measured in this task and allow a more finegrained analysis of the underlying sensorimotor–cognitive interactions. Close links between action and perception have been postulated since the early days of psychology (James, 1890; Lotze, 1852). This proposition is supported by behavioral and neuroimaging studies on action-effect binding, showing bi-directional associations between movements and their sensory consequences (Greenwald, 1970; Hommel, Musseler, Aschersleben, & Prinz, 2001; Kühn, Keizer, Rombouts, & Hommel, 2011; Kühn, Seurinck, Fias, & Waszak, 2010). As a consequence, imitation may be subserved by general mechanisms by which observing an action automatically activates neural circuits involved in performing the action oneself. Such an automatic activation could reflect actions being encoded in terms of their perceptual consequences, as proposed by ideomotor or common coding theories (Prinz, 1990), or, more generally, the result of associative learning (Brass & Heyes, 2005; Elsner, 2007; Heyes, 2001). Behavioral evidence for this view on imitation comes from studies assessing compatibility effects between to-be-performed movements (and their sensory consequences) on the one hand and visual stimuli on the other hand. In one such study, participants had to lift either the index or the middle finger of their dominant right hand in response to a visual cue while ignoring potential visual distractors (Brass et al.,

130

J. Verrel et al. / Acta Psychologica 150 (2014) 129–135

2001). Movements were either cued by a short movie of one finger being lifted or by a symbol (“1” or “2”). Compatibility effects were demonstrated in two ways in this study. First, responses were faster for movie cues compared to symbol cues. Second, if participants are symbolically cued to lift one finger while observing a lifting movement of a different finger, response times are longer than when symbolic cue and the movement distractor indicate the same response. Thus, even when participants were explicitly instructed to ignore it, the movie cue influenced the symbolically cued movement. The converse did not hold, that is, movements indicated by a movie cue were immune to interference by the symbolic distractor. Similar compatibility effects have also been demonstrated for movements with other body parts, such as hands, feet, or the mouth (Gillmeister et al., 2008; Heyes, Bird, Johnson, & Haggard, 2005; Leighton & Heyes, 2010; Stürmer et al., 2000), including symbolic gestures (Belot, Crawford, & Heyes, 2013; Cook, Bird, Lünser, Huck, & Heyes, 2012). A number of studies on arm and hand movements employed continuous measures to quantify automatic imitation tendencies in more detail (Bouquet, Gaurier, Shipley, Toussaint, & Blandin, 2007; Kilner, Paulignan, & Blakemore, 2003; Saby, Marshall, Smythe, Bouquet, & Comalli, 2011), demonstrating that (in)compatibility between observed and to-be-performed movements can have subtle effects on movement trajectories. It has been proposed that bidirectional associations between actions and perception (and hence compatibility effects as discussed above) are acquired through sensorimotor experience with self-performed movements and their sensory consequences (Brass, Bekkering, Wohlschläger, & Prinz, 2000; Brass & Heyes, 2005; Catmur et al., 2008; Hommel et al., 2001). This hypothesis was directly addressed in training studies in which participants had to respond with the hands to observed movements of the feet, and vice versa. This training led to a significant reduction in the effector-specificity of compatibility effects (Gillmeister et al., 2008) and to a reversal in neural responses during observation of hand versus foot movements (Catmur et al., 2008). Presumably, the training induced a generalization of visual action effect binding across effectors, emphasizing the role of sensorimotor experience as proposed by associative learning accounts of imitation (Brass & Heyes, 2005; Elsner, 2007; Heyes, 2001). Summing up, there is substantial behavioral evidence that observation of task-irrelevant movements influences motor performance, and neuroimaging studies suggest that this is related to the automatic preactivation of motor programs for the corresponding actions, which may be compatible or incompatible with the to-be performed movement. However, it is currently not known whether automatic imitation tendencies and resulting compatibility effects are confined to relatively isolated movements (e.g., moving a finger or arm, or opening/closing a hand) not requiring preparatory adjustments, or whether they also occur for more complex movements, in particular movements with whole-body balance constraints. The goal of the present study therefore is to investigate compatibility effects between observed and to-be-performed movements in a complex whole-body task: lifting one foot from the floor while maintaining balance. In order to lift one foot from the floor, it is not sufficient to activate the muscles that induce hip and knee flexion, but the whole body needs to be adjusted, shifting the weight to the opposite (standing) leg prior to the focal movement in order to maintain balance. In fact, the APA involves a preparatory movement (pushing to-be lifted foot into to the floor) which to some extent is opposite to the focal movement (lifting the foot). Thus, if observed movements influence motor performance by automatically activating muscles required for the focal movement (lifting the foot) without taking into account balance constraints, observing a foot lift action might actually hamper performance of the preparatory weight shift for lifting the foot on the same side. In contrast, if automatic imitation tendencies are integrated with postural control, observing a foot lift movement should facilitate lifting the ipsilateral foot and hamper lifting the contralateral foot, similar to a previous study on finger lift movements (Brass et al., 2000).

Moreover, balance constraints do not only make the motor task (and the correspondence mapping) more complex, but measuring APAs may also allow a more fine-grained temporal analysis of motor aspects of interference effects (e.g., Cohen, Nutt, & Horak, 2011). If the conflict between to-be-performed and observed movement is resolved at a cognitive level, both foot-lift responses and APAs should be delayed in incongruent conditions compared to congruent conditions. In contrast, if movement observation directly influences the motor system, as suggested by automatic imitation accounts, incongruent movement distractors should lead to wrong initial APAs (corresponding to the observed foot lift movement), which subsequently need to be corrected to lift the correct foot. 2. Material and methods 2.1. Participants Sixteen right-handed young adults aged between 20 and 30 years (mean age: 25.4 years, SD: 3.2 years) took part after providing written informed consent and received a compensation of 10 Euro per hour. The study was approved by the Ethics Committee of the Max Planck Institute for Human Development. 2.2. Setup and data acquisition During the foot lift task (see below), participants stood with their feet on two force plates (Kistler 9286AA, Kistler Instruments, Wintertur, Switzerland) with horizontal dimensions of 60 cm by 40 cm, in order to measure ground reaction force (GRF) separately from the two feet. Foot positions were marked by two pieces of carpet of dimensions 30 cm by 12 cm each, placed at a lateral distance of 20 cm (centers of the back edge) and an angle of 10°. Visual stimuli were back-projected to a screen placed at a distance of 150 cm in front of the force plates (projection design, F20 SX+). The size of the visual stimuli on the screen was 72 cm by 54 cm (size of the symbol: 7 × 8 cm), presented at a height of 40 cm above the floor. Three-dimensional kinematic data were recorded using an 8-camera optical motion capture system (Vicon Motion Systems, Oxford, UK). Reflective markers were attached to relevant landmarks on the participant's body. Only data from the markers on the toes, sacrum and seventh cervical vertebrae (C7) are used in the present analysis. Kinematic and force plate data, were recorded continuously during each block of trials (see below) at sampling rates of 100 Hz and 1000 Hz, respectively. For a second experiment (finger lift task, see below), participants were comfortably seated at a table, with a computer screen at a distance of about 50 cm from their eyes, the dominant right hand resting on a custom-built response, which recorded finger lift movements using infra-red light sensors. Visual stimuli for this task were shown on the screen on a size of about 12 × 9 cm (symbol: 1 × 1 cm). 2.3. Task and procedure Experimental programming was done in Matlab (Matlab R2011b, MathWorks, Natick, MA, USA) using the Psychophysics Toolbox (Brainard, 1997; Kleiner et al., 2007). The foot lift experiment consisted of a symbol cue and a movie cue condition, presented in separate blocks. In both conditions, two lower legs and feet were continuously displayed on the projection screen (Fig. 1, top panel). In the symbol cue condition, the letter L or R was presented between the feet, the task being to lift the corresponding (left or right) foot from the force plate. The symbolic cue was either shown without concurrent foot movement (baseline; Fig. 1, bottom center), with a foot lift movement on the same side (congruent; Fig. 1, bottom left), or with a foot lift movement on the opposite side (incongruent; Fig. 1, bottom right). In the movie cue condition, an animated sequence showing a foot lift was presented, and the task was to

J. Verrel et al. / Acta Psychologica 150 (2014) 129–135

131

accurately as possible. Responses were recorded with a custom-built sensor box, which used light sensors to detect finger lifting movements. The timing of visual stimuli was the same as for the foot lift experiment above, and trials started after the index and middle finger were back in place in the resting position. The finger lift experiment consisted of two blocks of 60 trials (20 baseline/congruent/incongruent trials). 2.4. Data analysis

Fig. 1. Illustration of the symbol cue condition of the foot lift experiment. At the beginning of the trial, the lower part of two legs was displayed. After a variable interval, a symbolic cue (L or R) was shown between the two ankles. This could be accompanied by a congruent or incongruent foot lift movement (lower left and right panel), or by no movement (baseline, lower center).

mirror the observed movement, that is, to lift the foot on the same side as the model on the screen. The movie cues were presented either without distractors, or with a letter L or R presented between the start positions of the feet, as in the symbol cue condition. Participants were instructed to respond as quickly and as accurately as possible. During the experiment, participants had their arms in a comfortable position along the sides of their body. The animated movie sequence consisted of four images, timed in terms of the projector's refresh cycle (60 Hz): the first image (Fig. 1, top) was shown for a variable, pseudorandomly determined duration of 30–54 cycles (500–900 ms) at the beginning of each trial; the second and the third image were shown for two refresh cycles (33.3 ms) each; the fourth picture was shown for 30 cycles (500 ms). The symbol (cue or distractor) was presented simultaneously with the beginning of the visual movement, that is, the second image of the sequence. At all other times (i.e., at the end of and between trials), the first image was displayed. The movie sequences were generated from mirrorsymmetric versions of a single foot which either remained on the ground or were lifted. It was therefore impossible to detect any postural adjustments from the visual stimuli. The beginning of each trial was initiated automatically when the difference between the GRF of the left and right foot remained below 20% of the participants' body weight (BW) for at least 300 ms, in order to ensure a symmetric body posture at the beginning of each trial. The cue stimulus (second movie frame or symbol) was displayed after a variable delay following trial onset (30–54 refresh cycles, 500–900 ms). Symbol and movie cue conditions were blocked and presented in an ABBA sequence, counter-balanced across participants. Each experimental block consisted of 60 trials (20 baseline, 20 congruent, 20 incongruent). Twelve practice trials were provided before the first and second block (first occurrence of each Cue Type condition). Participants were instructed to lift the cued foot as quickly as possible to a height about mid-way between ankle and knee of the stance leg. Prior to the foot lift experiment, participants performed a finger lift experiment, which was essentially the same as the one used in earlier studies on automatic imitation tendencies (Brass et al., 2001). As earlier experiments showed no distractor effects in the movie cue condition, only the symbol cue condition was used in this task: the digit “1” or “2” was shown between the index and the middle finger of a hand displayed on a computer screen. At the same time, a short picture sequence was presented, showing the lifting of one of the two fingers (or only the start image for a baseline condition). Participants were instructed to respond to the symbolic cue by briefly lifting either the index (1) or the middle (2) finger of their right hand, as quickly and

Motion capture data from the foot lift experiment were preprocessed and manually corrected in Vicon Nexus 1.7.1 (Vicon Motion Systems, Oxford, UK). Custom-written Matlab (Matlab R2011b, MathWorks, Natick, MA, USA) routines were used to analyze force plate and kinematic data, as described below. Trials were excluded if the force asymmetry at cue onset exceeded 20% of the body weight and if the foot lift occurred earlier than 200 ms or later than 2000 ms after cue onset, or if the wrong foot was lifted. Based on these criteria, 97.7% of all trials were analyzed. Kinematic and force plate data for each trial were lowpass-filtered using bidirectional Butterworth filter (cut-off frequency 20 Hz, order 5). The analysis of force plate data is illustrated in Fig. 2. The foot lift time was determined as the first time point at which the vertical force of one of the force plates dropped below 10 N. The onset of the first anticipatory postural adjustment (APA) was defined as the time after cue presentation, when the absolute difference between the vertical forces of the two feet increased by more than 5% of the participants' body weight, relative to the force difference at cue onset. Subsequent APA onsets were scored at zero-crossings of the force difference which were followed by an absolute excursion of at least 5% of the body weight. Based on this, the number of APAs was determined for each trial. Moreover, the amplitude of an APA was defined as the absolute maximal force difference during the APA. The movement of the body in space was assessed by the range of the vertical motion of the toe marker and the range of the roll angle of the trunk (determined from the markers placed on C7 and sacrum). 2.5. Statistical analysis Statistical analyses were performed in R (R Development Core Team, 2008). Dependent variables for the foot lift experiment were averaged across trials of the cells defined by the factors Cue Type (symbol, movie) and Congruency (congruent, baseline, incongruent), and submitted to a 2 × 3 repeated measures ANOVA with within-subject factors Cue Type and Congruency. Response times from the finger lift experiment were analyzed with a repeated-measures ANOVA with within-subject factor condition (congruent, baseline, incongruent). Interference in the foot and finger lift experiment was quantified, separately for each subject, by computing unpaired t-scores contrasting performance in the congruent and incongruent condition. The resulting individual interference scores for different performance measures were correlated between subjects to test for potential links between the underlying cognitive processes. The level for statistical significance was 0.05. Significant effects were followed up by paired t-tests, corrected for multiple comparisons (Holm, 1979). Effect sizes are reported as generalized eta-squared score η2G (Bakeman, 2005). 3. Results 3.1. Foot lift experiment Sample data from the foot lift experiment are shown in Fig. 2. As can be seen in the trial with a single APA (left panel), the force difference first decreases, corresponding to an increasing ground reaction force of the to-be-lifted foot, that is, a pushing force accelerating the body towards the standing foot. Subsequently, the force difference increases

132

J. Verrel et al. / Acta Psychologica 150 (2014) 129–135

Fig. 2. Sample force plate data and illustration of APA analysis. The force difference (stance foot minus lift foot), referenced to the beginning of cue onset, is plotted from cue onset (t = 0) till foot lift (lift foot force dropping below 10 N). The APA onset is scored when the force difference increases by 5% of the body weight, relative to cue onset. The amplitude of an APA is defined as the peak deviation from the force difference at cue onset. A. Trial with a single APA. B. Trial with two APAs, indicating an initial weight shift in the wrong direction.

to 100%, when the foot that is being lifted leaves the ground. In the trial with two APAs (right panel), the force difference initially increases, reflecting a pushing force with the standing foot, which results in acceleration of the body towards the to-be-lifted foot, that is, in the wrong direction. This is followed by a corrective pushing with the tobe-lifted foot (decrease of the force differences) shifting the weight to the standing leg, and finally lifting of the correct leg (force difference reaching 100%).

3.1.1. Foot lift times and number of APAs Summary statistics for the foot lift time and the number of APAs are shown in Fig. 3. For the foot lift time, the ANOVA yielded significant main effects of Congruency, F(2,30) = 29.0, p b 0.001, η2G = 0.04, and Cue Type, F(1,15) = 155.5, p b 0.001, η2G = 0.22, and a significant twoway interaction, F(2,30) = 35.8, p b 0.001, η2G = 0.04. Pairwise comparisons showed that, across Congruency conditions, foot lift times were shorter for movie cues compared to symbol cues, t(15) N 5, p b 0.001 in all cases. For movie cues, no differences were found between Congruency conditions, t(15) b 1, p N 0.5. In contrast, for symbol cues, foot lift times differed significantly between each two Congruency conditions (incongruent N baseline N congruent), t(15) N 4, p b 0.005. For the number of APAs, a similar pattern was found. The ANOVA showed significant main effects of Congruency, F(2,30) = 52.1, p b 0.001, η2G = 0.41, and Cue Type, F(1,15) = 175.2, p b 0.001, η2G = 0.57, and a significant two-way interaction, F(2,30) = 26.6, p b 0.001, η2G = 0.32. The number of APAs was smaller for movie cues compared to symbol cues, across Congruency conditions t(15) N 3, p b 0.02. Congruency did not influence the number of APAs in the movie cue condition t(15) b 2, p N 0.1, but did so in the symbol cue condition, with significant differences between each pair of Congruency levels (incongruent N baseline N congruent), t(15) N 3, p b 0.01.

Of all valid trials of all participants, 78.1% had one APA, 20.5% had two APAs, and 1.2% had more than two APAs. As trials with more than two APAs might indicate a more complex process than basic errorcorrection, we repeated the statistical analyses for the foot lift time and number of APAs, including only trials with one or two APAs. This did not change the pattern of statistically significant effects.

3.1.2. Timing and amplitudes of APAs Fig. 4 shows onset times and amplitudes of the first and last APAs in a trial. For the onset of the first APA, only a main effect of Cue Type was found, F(1,15) = 69.0, p b 0.001, η2G = 0.26, with earlier APA onset in the movie cue compared to the symbol cue condition. In contrast, the onset of the last APA showed a pattern similar to the foot lift time and the number of APAs, with significant main effects of Congruency, F(2,30) = 39.2, p b 0.001, η 2G = 0.27, and Cue Type, F(1,15) = 191.0, p b 0.001, η 2G = 0.66, and a significant two-way interaction, F(2,30) = 37.2, p b 0.001, η 2G = 0.26. Onset of the last APA occurred earlier in the movie compared to the symbol cue condition, t(15) N 6, p b 0.001 and differed between each pair of levels of the factor Congruency in the symbol cue condition (incongruent N baseline N congruent), t(15) N 5, p b 0.001 in all cases. For the amplitudes of the first APA, we found main effects of Congruency, F(2,30) = 30.1, p b 0.001, η2G = 0.08, and Cue Type, F(1,15) = 86.7, p b 0.001, η2G = 0.26, and a significant two-way interaction, F(2,30) = 13.72, p b 0.001, η2G = 0.05. Amplitudes were smaller for the symbol compared to the movie cue in all three conditions, t(15) N 3.9, p b 0.002. For symbol cues, pairwise differences between Congruency conditions were significant (incongruent b baseline b congruent), t(15) N 3, p b 0.01 in all cases. For the amplitudes of the last APA, we found a main effect of Congruency, F(2,30) = 8.90, p b 0.001, η 2G = 0.01, and a two-way interaction,

Fig. 3. Main outcome measures for foot lift experiment: A. time of foot lift, B. number of APAs. Error bars indicate between-subject standard error.

J. Verrel et al. / Acta Psychologica 150 (2014) 129–135

133

Fig. 4. Additional APA measures for foot lift experiment: onset and amplitudes of first and last APAs. Error bars indicate between-subject standard error.

F(2,30) = 14.7, p b 0.001, η2G = 0.01. Pairwise differences between congruency conditions were only significant in the symbol cue condition, with a larger amplitude for incongruent compared to congruent and baseline stimuli, t(15) N 4, p b 0.01. 3.1.3. Influence of response conflict on APA timing and amplitudes To further elucidate the mechanisms underlying the increased number of APAs under response conflict, we analyzed APA latencies and amplitudes in incongruent symbol cue trials only, separately for trials with a single versus multiple APAs. Fig. 5 shows onset times and amplitudes for trials with a single APA, and for the first and the last APA for trials with multiple APAs. Paired t-tests showed that the onset of the APA in trials with a single APA occurred later than the onset of the first APA in trials with multiple APAs, t(15) = 5.48, p b 0.001, and earlier than the onset of the last APA (in trials with multiple APAs), t(15) = 18.61, p b 0.001. Moreover,

in trials with multiple APAs, the first APA (trivially) occurred before the last APA t(15) = 21.0, p b 0.001. An additional t-test showed that the latency of first APAs in incongruent symbol-cue trials with multiple APAs did not significantly differ from APA latency in movie cue trials with a single APA, t(15) = 0.99. p = 0.33. APA amplitudes were found to be smaller in first APAs, compared to both single APAs, t(15) = 16.0, p b 0.001, and last APAs, t(15) = 32.9, p b 0.001. Moreover, amplitudes were larger on average in the last (corrective) compared to single (correct) APAs, t(15) = 8.77, p b 0.001. The vertical toe lift range showed a significant interaction of Cue Type and Congruency, F(2,30) = 3.33, p = 0.049, η 2G = 0.00, but follow-up tests did not reveal any significant differences between conditions. The trunk roll angle range showed no significant effects. Likewise, the foot load asymmetry at the time of cue onset did not significantly differ between conditions. Thus, these control measures do not show any systematic differences between experimental

Fig. 5. Onset (A) and amplitude (B) of APAs, plotted separately for trials with a single or multiple (first, last) APAs, in the incongruent symbol cue condition. Data were averaged across conditions within subjects. Boxes and whiskers indicate the between-subject inter-quartile range and range, respectively.

134

J. Verrel et al. / Acta Psychologica 150 (2014) 129–135

conditions that might explain the observed effects in the main dependent variables reported above. 3.2. Finger lift experiment and between-task correlations Correct responses and reaction times from the finger lift experiment replicated the pattern found previously (Brass et al., 2000). For the proportion of correct responses, the ANOVA showed a main effect of Congruency, F(2,30) = 32.78, p b 0.001, η2G = 0.45. Participants had higher accuracy in the congruent (mean ± SD: 92.8 ± 4.1%) and the baseline (89.8 ± 6.9%) compared to the incongruent condition (80.3 ± 6.5%), both t(15) N 5, p b 0.001. The ANOVA for reaction time (only correct trials) showed a main effect of Congruency, F(2,30) = 24.88, p b 0.001, η2G = 0.17. Pairwise differences between conditions were significant, with shorter reaction times in the congruent (416 ± 28 ms) compared to baseline (433 ± 25 ms), and in baseline compared to incongruent condition (447 ± 32 ms), t(15) N 3, p b 0.05. Correlations of interference scores of the finger lift and foot lift task showed the following pattern: the correlation was not significant between interference scores for foot lift time and reaction time (finger task), ρ = 0.13, p N 0.5. Significant or marginally significant correlations were found for foot lift time and accuracy in the finger lift task, ρ = − 0.55, p = 0.027, number of APAs (foot) and reaction time (finger), ρ = 0.43, p = 0.099, and number of APAs (foot) and accuracy (finger), ρ = − 0.47, p = 0.065. Summing up, we found evidence for compatibility effects in the foot lift experiment, as shown by the fact that presentation of an incongruent movement distractor impaired task performance in the symbol cue condition. Degraded performance was revealed by delayed foot lift times and an increased number of APAs. The same pattern was found for the latency of the last APA in trial, while the onset of first (potentially wrong) APA was not affected by the congruency manipulation. Importantly, these findings are not explained by any differences between the conditions in movement performance (foot lift height, lateral trunk movement) or pre-trial load asymmetry. The finger lift experiment replicated previous studies using this paradigm (Brass et al., 2000). Significant or marginally significant correlation between interference scores for different performance measures in the two foot and finger lift paradigms suggests that the two tasks may draw on shared processes. 4. Discussion The present study investigated compatibility effects between observed and to-be-performed movements in a complex motor task with balance constraints: lifting one foot from the floor while standing in response to a symbolic cue. As indicated by reaction times and number of anticipatory postural adjustments (APAs), participants were better at mirroring a visually presented movement than at performing the same movement in response to a symbolic cue. Moreover, the presence of symbolic distractors did not affect performance in the mirroring task, while the presence of congruent/incongruent movement distractors facilitated/degraded performance in the symbolic task, compared to a baseline condition without movement distractors. Our results extend earlier work on perception–action compatibility effects from relatively isolated movements of individual body parts (Brass et al., 2000; Gillmeister et al., 2008; Heyes et al., 2005; Leighton & Heyes, 2010; Stürmer et al., 2000) to a more complex whole-body task with balance constraints. This suggests that the influence of the observed task-irrelevant movements goes beyond priming individual muscles related to the focal movement (such as lifting a finger or opening/closing a hand). Instead, it can involve multiple muscles and – most importantly – preparatory adjustments involving the entire body to satisfy balance constraints. Thus, the neuronal mechanisms solving the “correspondence problem” in imitation paradigms (Brass & Heyes,

2005) appear to take into account the complexity of integrating movements with postural control. Automatic imitation tendencies have been explained in terms of stimulus–response mappings acquired through the experience of visually observing one's self-performed movements (Brass & Heyes, 2005; Brass et al., 2000; Catmur et al., 2008; Hommel et al., 2001). This may be a plausible explanation for hand and finger movements, which we are likely to observe ourselves doing, for instance when manipulating objects or typing on a keyboard. In contrast, we arguably do not often observe ourselves lifting one foot from the floor, at least not from the visual perspective used here, suggesting that direct visuomotor experience from self-performed movements is not a prerequisite for the compatibility effects found in the present study. Our results might partially be explained by spatial compatibility (Bertenthal, Longo, & Kosobud, 2006; Brass et al., 2000; Catmur & Heyes, 2011). This explanation cannot be ruled out based on the present experiment. However, this does not undermine our main result, namely, that automatic response tendencies to task-irrelevant movement stimuli do take balance constraints into account. Delayed foot lifting in symbol cue trials with incongruent movement distractor was accompanied by an increased number of APAs, corresponding to erroneous initial weight shifts. In contrast, the latency of the first APA in a trial was not affected by congruency, indicating that an early motor response was triggered at a time point at which the response conflict between the symbolic cue and the incongruent distractor had not yet been resolved. To further elucidate the mechanisms underlying the increased number of APAs under response conflict, we analyzed APA latencies and amplitudes in incongruent symbol cue trials in more detail. This analysis showed that, in trials with multiple APAs, the first (usually wrong) APA was initiated earlier compared to trials with a single APA, and at a similar latency as the first and single APAs in the movie cue condition. This indicates that observation of an incongruent movement automatically activated the corresponding (wrong) motor response, including the associated APA, providing strong support for the automatic imitation hypothesis. In addition, APA amplitudes were smaller for the first APA in a trial with multiple APAs, compared to the final APA in trials with a single or multiple APAs. This suggests that either, initial erroneous APAs are executed with less vigor (potentially reflecting an underlying response conflict), or that corrective mechanisms come into play early in the execution of the first APA. Taken together, results from the analysis of APA latencies and amplitudes strongly suggest that the observed compatibility effects are not confined to cognitive levels of action control but directly interact with overt motor performance. The increased number of APAs in the incongruent condition reflects initial weight shift in the wrong direction. Thus, our results are in line with previous work on partial errors in manual response paradigms (for review, see Coles, Scheffers, & Fournier, 1995; McBride, Boy, Husain, & Sumner, 2012). In these studies, competing response tendencies lead to increased (sub-threshold) activation of alternative responses, which can be measured at different levels of the neuromotor system by means of force sensors, EMG, or EEG. Partial errors reflected in APAs (as in the present study) have previously been found during gait initiation, in particular for older adults (Cohen et al., 2011). However, to our knowledge, our study is the first to directly address the relation between APA errors and response conflict. As mentioned in the Introduction, previous studies used continuous measures to characterize automatic imitation tendencies in more detail, going beyond reaction time and overt response errors (Bouquet et al., 2007; Kilner et al., 2003; Saby et al., 2011). Our results are in line with these results and extend them to a task requiring the integration of a focal movement with balance constraints. In order to assess to what extent imitation tendencies generalize across effectors and levels of task complexity, participants also completed a simple finger task (Brass et al., 2000). Between-subject correlations of interference effects in the symbol cue condition in the foot lift task and

J. Verrel et al. / Acta Psychologica 150 (2014) 129–135

the finger lift task were significant or marginally significant in several instances, providing preliminary evidence that there are shared sensorimotor-cognitive mechanisms underlying successful performance of these tasks, presumably related to inhibition of automatic imitation tendencies. Moreover, the directionality of the effects in the finger and foot lift task was compatible, that is, better performance in one task was associated with better performance in the other. However, the link between performance in the two tasks established by the correlation analysis is not entirely straightforward: while response times did not correlate significantly between the two tasks, a significant correlation was found, for instance, between foot lift time and finger lift accuracy. This may be partly explained by differences in the temporal characteristics of the two tasks. For instance, the need for postural preparation in the foot lift task may leave more time for executive control to influence the ongoing response, increasing the likelihood of partial errors (i.e., multiple APAs). Further work is needed to clarify the underlying mechanisms and how they may be affected by task complexity or biomechanical characteristics, for instance induced by balance constraints. The movie stimuli used in our present study showed a person facing the participant, rather than a person seen from behind. Our results confirm that this “mirror-perspective” does induce mirror-symmetric automatic imitation tendencies, as previously shown for instance for finger movements (Brass et al., 2001). Perspective has been shown to influence the ease of direct imitation and imitation learning, with seeing the model from behind facilitating anatomical matching while facing the model facilitating mirror-symmetric matching (Ishikura & Inomata, 1995; Press, Ray, & Heyes, 2009). The movie sequences were generated from mirror-symmetric versions of a single foot which either remained on the ground or were lifted. It was therefore impossible to detect any postural adjustments from the visual stimuli. Future studies should address the question to what extent automatic imitation tendencies in whole-body movement might be influenced by different perspectives as well as the presence or absence of postural adjustments in the visual movement stimuli. 5. Conclusions We demonstrated that automatic imitation tendencies exist in a motor task with whole-body balance constraints: lifting one foot from the floor while standing. This extends previous work on compatibility effects for relatively isolated movements of body parts (e.g., lifting a finger, opening/closing a hand), allowing two main conclusions. First, automatic imitation tendencies go beyond activation of individual muscles and take into account the complexity of integrating movements with postural control. Second, analysis of anticipatory postural adjustments showed that the response conflict induced by incongruent movement distractors is not resolved at a purely cognitive level but directly permeates into motor control. Acknowledgments We would like to thank Gabriele Faust, Stefan Haydn, Ilka Popp, Katharina Voigt, and Manuel Zellhöfer for their help with this study and Ulman Lindenberger for the helpful discussions of an earlier version of the manuscript. References Bakeman, R. (2005). Recommended effect size statistics for repeated measures designs. Behavior Research Methods, 37(3), 379–384. Belot, M., Crawford, V. P., & Heyes, C. (2013). Players of matching pennies automatically imitate opponents' gestures against strong incentives. Proceedings of the National Academy of Sciences, 110(8), 2763–2768, http://dx.doi.org/10.1073/pnas.1209981110. Bertenthal, B. I., Longo, M. R., & Kosobud, A. (2006). Imitative response tendencies following observation of intransitive actions. Journal of Experimental Psychology: Human Perception and Performance, 32(2), 210–225, http://dx.doi.org/10.1037/0096-1523.32.2.210.

135

Bouquet, C. A., Gaurier, V., Shipley, T., Toussaint, L., & Blandin, Y. (2007). Influence of the perception of biological or non-biological motion on movement execution. Journal of Sports Sciences, 25(5), 519–530, http://dx.doi.org/10.1080/02640410600946803. Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10(4), 433–436. Brass, M., Bekkering, H., & Prinz, W. (2001). Movement observation affects movement execution in a simple response task. Acta Psychologica, 106(1), 3–22. Brass, M., Bekkering, H., Wohlschläger, A., & Prinz, W. (2000). Compatibility between observed and executed finger movements: Comparing symbolic, spatial, and imitative cues. Brain and Cognition, 44(2), 124–143. Brass, M., & Heyes, C. (2005). Imitation: Is cognitive neuroscience solving the correspondence problem? Trends in Cognitive Sciences, 9(10), 489–495. Catmur, C., Gillmeister, H., Bird, G., Liepelt, R., Brass, M., & Heyes, C. (2008). Through the looking glass: Counter-mirror activation following incompatible sensorimotor learning. European Journal of Neuroscience, 28(6), 1208–1215, http://dx.doi.org/10.1111/j. 1460-9568.2008.06419.x. Catmur, C., & Heyes, C. (2011). Time course analyses confirm independence of imitative and spatial compatibility. Journal of Experimental Psychology: Human Perception and Performance, 37(2), 409. Cohen, R. G., Nutt, J. G., & Horak, F. B. (2011). Errors in postural preparation lead to increased choice reaction times for step initiation in older adults. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 66(6), 705–713. Coles, M. G. H., Scheffers, M. K., & Fournier, L. (1995). Where did you go wrong? Errors, partial errors, and the nature of human information processing. Acta Psychologica, 90(1–3), 129–144, http://dx.doi.org/10.1016/0001-6918(95)00020-U. Cook, R., Bird, G., Lünser, G., Huck, S., & Heyes, C. (2012). Automatic imitation in a strategic context: Players of rock–paper–scissors imitate opponents' gestures. Proceedings of the Royal Society B: Biological Sciences, 279(1729), 780–786. R Development Core Team (2008). R: A language and environment for statistical computing. Vienna, Austria. (Retrieved from http://www.R-project.org). Elsner, B. (2007). Infants' imitation of goal-directed actions: The role of movements and action effects. Acta Psychologica, 124(1), 44–59. Gillmeister, H., Catmur, C., Liepelt, R., Brass, M., & Heyes, C. (2008). Experience-based priming of body parts: A study of action imitation. Brain Research, 1217, 157–170. Greenwald, A. G. (1970). Sensory feedback mechanisms in performance control: With special reference to the ideo-motor mechanism. Psychological Review, 77(2), 73. Heyes, C. (2001). Causes and consequences of imitation. Trends in Cognitive Sciences, 5(6), 253–261. Heyes, C. (2011). Automatic imitation. Psychological Bulletin, 137(3), 463. Heyes, C., Bird, G., Johnson, H., & Haggard, P. (2005). Experience modulates automatic imitation. Cognitive Brain Research, 22(2), 233–240. Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 65–70. Hommel, B., Musseler, J., Aschersleben, G., & Prinz, W. (2001). The theory of event coding (TEC): A framework for perception and action planning. Behavioral and Brain Sciences, 24(5), 849–877. Ishikura, T., & Inomata, K. (1995). Effects of angle of model-demonstration on learning of motor skill. Perceptual and Motor Skills, 80(2), 651–658. James, W. (1890). The principles of psychology. New York: Holt. Kilner, J. M., Paulignan, Y., & Blakemore, S. J. (2003). An interference effect of observed biological movement on action. Current Biology, 13(6), 522–525. Kleiner, M., Brainard, D., Pelli, D., Ingling, A., Murray, R., & Broussard, C. (2007). What's new in Psychtoolbox-3. Perception, 36(14), 14. Kühn, S., Keizer, A. W., Rombouts, S. A., & Hommel, B. (2011). The functional and neural mechanism of action preparation: Roles of EBA and FFA in voluntary action control. Journal of Cognitive Neuroscience, 23(1), 214–220. Kühn, S., Seurinck, R., Fias, W., & Waszak, F. (2010). The internal anticipation of sensory action effects: When action induces FFA and PPA activity. Frontiers in Human Neuroscience, 4, http://dx.doi.org/10.3389/fnhum.2010.00054. Leighton, J., & Heyes, C. (2010). Hand to mouth: Automatic imitation across effector systems. Journal of Experimental Psychology: Human Perception and Performance, 36(5), 1174. Lotze, H. (1852). Medicinische Psychologie oder Physiologie der Seele. Leipzig, Germany: Weidmann. McBride, J., Boy, F., Husain, M., & Sumner, P. (2012). Automatic motor activation in the executive control of action. Frontiers in Human Neuroscience, 6. Meltzoff, A. N., & Moore, M. K. (1974). Imitation of facial and manual gestures by human neonates. Brain Research, 19, 124. Miller, N. E., & Dollard, J. (1941). Social learning and imitation. Retrieved from. http:// psycnet.apa.org/psycinfo/1942-00109-000 Press, C., Ray, E., & Heyes, C. (2009). Imitation of lateralised body movements: Doing it the hard way. Laterality, 14(5), 515–527, http://dx.doi.org/10.1080/13576500802607990. Prinz, W. (1990). A common coding approach to perception and action. Springer (Retrieved from http://link.springer.com/chapter/10.1007/978–3–642–75348–0_7). Saby, J. N., Marshall, P. J., Smythe, R., Bouquet, C. A., & Comalli, C. E. (2011). An investigation of the determinants of motor contagion in preschool children. Acta Psychologica, 138(1), 231–236, http://dx.doi.org/10.1016/j.actpsy.2011.06.008. Stürmer, B., Aschersleben, G., & Prinz, W. (2000). Correspondence effects with manual gestures and postures: A study of imitation. Journal of Experimental Psychology: Human Perception and Performance, 26(6), 1746. Tomasello, M., Kruger, A. C., & Ratner, H. H. (1993). Cultural learning. Behavioral and Brain Sciences, 16, 495–511.