Top-down attentional factors modulate action priming in reach-to-grasp action

Abstract

Previous studies report that viewing exaggerated, high-lifting reaches (versus direct reaches) primes higher vertical deviation in wrist trajectory in the observer’s subsequent reaches (trajectory priming), but it is unclear to what extent this effect depends upon task instructions relevant to top-down attention. In two experiments, participants were instructed to gaze at a dot presented on a large monitor for a colour-change go signal that cued them to execute a direct reach to a target. In the background, the monitor also displayed life-sized films of a human model. The films were of the model either remaining still or reaching to grasp a target with either a direct trajectory or an exaggerated, high-lifting trajectory. When the dot traced the human model’s wrist throughout her movement, a robust trajectory priming effect emerged. When the dot remained stationary in a central location but the human model reached in the background, the human model’s trajectory did not alter the participants’ trajectories. Finally, when the dot traced exaggerated and direct trajectories and the human model remained stationary, the dot’s movement produced an attenuated, non-significant trajectory priming effect. These findings show that top-down attentional factors modulate trajectory priming. In addition, a moving non-human stimulus does not produce the same degree of action priming when contextual factors make salient its independence of human agency and/or intention.

Keywords

Contextual effects action priming trajectory priming attention top-down factors instruction effects

Introduction

Observing others’ actions affects our own motor output (Brass, Bekkering, Wohlschläger, & Prinz, 2000; Kilner, Paulignan, & Blakemore, 2003 see Heyes, 2011 for review). These action observation effects can manifest online, when observed actions alter parameters of concurrently executed actions (e.g., Hardwick & Edwards, 2011; Kilner et al., 2003), and also offline, when an observed action influences subsequent executed actions (e.g., Bach, Peatfield, & Tipper, 2007; Griffiths & Tipper, 2009, 2012; Hardwick & Edwards, 2011; Sparks, Sidari, Lyons, & Kritikos, 2016). Offline action priming can manifest both in the time taken for the observer to initiate actions (e.g., Bach et al., 2007; Brass, Bekkering, & Prinz, 2001; Cook & Bird, 2011; Gowen, Bradshaw, Galpin, Lawrence, & Poliakoff, 2010) and the executed actions’ unfolding kinematics (Griffiths & Tipper, 2009, 2012; Hardwick & Edwards, 2011; Sparks, Douglas, & Kritikos, 2016). In this article, we focus on the latter, specifically, the priming of trajectory kinematics in reach-to-grasp actions (Griffiths & Tipper, 2009, 2012; Hardwick & Edwards, 2011; Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, Lyons, & Kritikos, 2016).

The bulk of studies on offline action priming focus on intransitive and simple actions such as finger-lifting/tapping (e.g., Brass et al., 2001; Gowen, Bolton, & Poliakoff, 2016; Gowen et al., 2010; Liepelt, Von Cramon, & Brass, 2008) and hand opening/closing (e.g., Press, Bird, Walsh, & Heyes, 2008; Stürmer, Aschersleben, & Prinz, 2000) and use initiation time as the primary measure. Participants execute an action as fast as possible after observing a congruent action (e.g., view hand closing then execute hand closing) or after observing an incongruent action (e.g., view hand opening then execute hand closing). Typically, participants’ initiation times are faster for congruent versus incongruent trials (e.g., Brass et al., 2000; see Heyes, 2011 for review). These studies provide strong evidence that observed actions modulate action preparation and initiation, but do not assess whether these effects extend to the observer’s kinematic output.

In trajectory priming studies participants observe a model execute reach-to-grasp actions that always achieve end states congruent with their own (grasping an object). Trial-by-trial, the model’s reach varies in the trajectory used to achieve this end-state, using either a relatively direct wrist trajectory or a higher-lifting, exaggerated wrist trajectory. Multiple studies have demonstrated that the wrist trajectory of the model’s reach partially transfers to the observer’s reach, both when the model appears in vivo (Griffiths & Tipper, 2009, 2012; Hardwick & Edwards, 2011) and when the model is presented via life-sized films (Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). Specifically, the observer attains a higher maximum wrist height while reaching to a target after observing a higher-lifting, exaggerated reach compared with a relatively direct reach. Importantly, this kinematic transfer occurs both when the model’s higher wrist-lift is goal relevant, such as when the model must reach over an intervening obstruction to grasp their target (Griffiths & Tipper, 2009, 2012) and when there is no observable goal-related need for the model to deviate from a more direct wrist trajectory (Hardwick & Edwards, 2011; Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). In addition, trajectory priming occurs even when participants, under no time pressure, are instructed to disregard how the model moves and reach directly for their target (Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). Similar results occur when participants reach towards but do not manipulate a target (Wild, Poliakoff, Jerrison, & Gowen, 2012; but note this is under instructed imitation conditions). Combined, these findings suggest that even goal-irrelevant kinematics of observed actions can modify executed actions involuntarily and with high kinematic specificity. Thus, these studies complement initiation-time paradigms as a tool for assessing the scope of action observation effects.

Although action observation effects seem involuntary to an extent, top-down factors can modulate them. In offline action priming paradigms, evidence has accumulated showing that the observer’s beliefs and assumptions about the intentionality, agency, and/or humanness of the model’s actions modulate the effects (for review see Gowen & Poliakoff, 2012, but cf. Cracco et al., 2018). For example, viewing a human hand executing finger-lifting actions elicits stronger action priming in initiation time when the action seems intentional than when the action appears to have been caused by a machine lifting the fingers (Liepelt et al., 2008). Furthermore, stimuli with ambiguous causal origins, such as moving geometric shapes, can induce action priming in initiation time when verbal instructions suggest that the stimuli represent an intentional human action (Gowen et al., 2016). There is evidence that these factors extend to priming of specific kinematics. Moving dot stimuli can induce trajectory priming in reach-to-grasp actions when visual context suggests that the dot represents a human hand’s motion, whereas identical dot movements do not affect the observer’s reach trajectories in the absence of such cues (Sparks, Sidari, et al., 2016).

Clearly, high-level top-down factors, such as beliefs about agency, are important in action priming (Gowen & Poliakoff, 2012), but it is less clear how task instructions regarding top-down control of visual attentional influence action priming effects. In initiation-time studies, participants consistently show action observation effects while selectively attending to (and moving in response to) cues appearing near or superimposed on the observed action, suggesting that selective attention to the action itself is not necessary (Brass et al., 2000; Cook & Bird, 2011; Heyes, 2011; Liepelt et al., 2008 but cf. Chong, Cunnington, Williams, & Mattingley, 2009). In contrast to selective attention, Bach et al. (2007) showed that drawing spatial attention away from the relevant effector on the model’s body attenuated action priming in initiation time, which suggests that spatial attention to the relevant body site is important for action priming. It is worth noting, however, that the stimuli Bach et al. used were static depictions of in-progress actions, so it is unclear whether similar constraints will apply to full-motion actions. Full-motion actions may be more salient and contain additional kinematic information, thereby producing stronger motor activation in the observer. In addition, Bach et al.’s static actions appeared at a considerable distance in depth from participants. Action priming effects can be more robust within peripersonal (reachable) space than outside of it (Griffiths & Tipper, 2009).

Reach-to-grasp trajectory priming paradigms offer a means to test the effect of instructions regarding attention in full-motion actions presented close to the observer. Existing trajectory priming studies vary in the relevant instructions. Participants have been instructed to: gaze at their upcoming reach target while observing the model in the periphery to time their own actions (Hardwick & Edwards, 2011); passively watch the model’s reach and close then open their eyes before executing their own action (Griffiths & Tipper, 2009, 2012); carefully watch the model’s hand because their grasp type was a go/no-go signal (Sparks, Douglas, & Kritikos, 2016) or fixate centrally while covertly attending to the spatial region of the model’s hand for a go signal (Sparks, Sidari, et al., 2016). In most of these studies, the instructions specified or implied that participants should spatially and selectively attend to the model’s action (Griffiths & Tipper, 2009, 2012; Hardwick & Edwards, 2011; Sparks, Douglas, & Kritikos, 2016). Sparks, Sidari, et al. (2016) instructed participants to attend selectively to a dot that traced the model’s wrist, rather than to the model’s action itself. Trajectory priming still occurred under these instructions, which suggests that selective attention to a human action may not be necessary for the effect. The instructions in Sparks, Sidari, et al. (2016), however, still required that participants spatially attend to the relevant body site. Thus, it is currently unclear to what extent instructions regarding spatial and selective attention modulate trajectory priming in reach-to-grasp action.

Findings from studies involving coordinative simultaneous execution and observation of sinusoidal arm movements are relevant here (Constable et al., 2017; Kilner et al., 2003; Stanley, Gowen, & Miall, 2007). In these paradigms, observers are instructed to execute vertical or horizontal arm waving motions in time with observed arm waves executed by a model in a compatible (e.g., execute horizontal, view horizontal) or incompatible (e.g., execute horizontal, view vertical) plane. The typical finding is that observation of incompatible arm movements increases variability in observers’ trajectories relative to observation of compatible movement (Constable et al., 2017; Kilner et al., 2003; Stanley et al., 2007). In these studies, participants are usually instructed to watch the model’s hand or finger (Kilner et al., 2003) and would therefore be likely to fixate upon and attend the hand region. Constable et al. (2017) suggested that tracking the model’s hand with the eyes might at least partially account for the motor interference seen in the incompatible conditions in these paradigms. They demonstrated that when participants were instructed to fixate centrally while covertly attending to the model’s action, interference effects were abolished. Constable et al. suggest that interference from executed eye movement trajectory to executed hand trajectory could partially or wholly account for the interference effect typically reported with this paradigm.

These coordinative arm-waving paradigms differ from the offline trajectory priming tasks described earlier in key ways. Because the arm-waving paradigms are online tasks requiring simultaneous action observation and execution, participants would be tracking the model’s hand simultaneously with executing their own actions. This contrasts with offline trajectory priming studies where participants alternate between observing and executing actions (Griffiths & Tipper, 2009, 2012; Hardwick & Edwards, 2011; Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). In some instantiations of the trajectory priming paradigm, participants can shift their covert and overt visual attention away from the model to their target object before or during action execution (Griffiths & Tipper, 2009, 2012; Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). They may also turn their head away from the model towards their target before acting (Griffiths & Tipper, 2009, 2012; Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). In these scenarios, if moving the locus of overt and/or covert attention during action observation contributes to trajectory priming, its influence would need to persist through intervening shifts in the attentional locus and (in some cases) head movements. It is therefore an open question whether instructions about selective attention, spatial attention, and gaze contribute to or constrain offline trajectory priming.

Experiment 1

We modified our reach-to-grasp trajectory priming paradigm to test whether instructions regarding spatial attention, selective attention, and gaze contribute to or constrain trajectory priming effects (Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). In Experiment 1, participants were instructed to fixate on a white shape that either followed the model’s wrist (cue-on-wrist condition) or remained stationary in a central location (central cue condition) on a randomised trial-by-trial basis. This shape changed colour to indicate a “go” or “no-go” trial. Thus, we directed participants’ selective attention to the shape rather than to the model, and the model’s hand and arm were not strictly task-relevant. As in our previous trajectory priming studies, we explicitly informed participants that they should try to ignore human model’s movement and not let it influence their own movement. Thus, in the central cue condition, participants had no need to fixate upon or spatially attend to the model’s movement, whereas instructions in previous studies required or encouraged at least one of these factors. Based on previous evidence that spatial attention to the relevant body site is important for offline action priming in initiation time (Bach et al., 2007), we hypothesised that trajectory priming would be attenuated or abolished in the central cue condition compared with the cue-on-wrist condition in Experiment 1.

The stimuli were displayed on a large monitor such that the model appeared roughly life-sized and seated beside the participant. We chose this side-by-side orientation because it successfully produced trajectory priming in previous studies (Sparks, Douglas, & Kritikos, 2016; Sparks, Sidari, et al., 2016). First-person perspective has been associated with strong sensorimotor activation during action observation (Jackson, Meltzoff, & Decety, 2006), but that perspective would make the vertical differences in the model’s trajectory less visually obvious, so we opted for the side-by-side orientation. In addition, there was always a visible target for the model’s movements. There is evidence that when a clear target is not visible participants attend more to the model’s hand and trace it with their gaze (Wild et al., 2012), and our aim was to minimise elements that would encourage participants to attend (covertly or overtly) a location other than the white circle cue. Finally, we used an offline task where participants have no need to coordinate the timing of their actions with the model’s. The go signal always appears once the model has completed her movement. Romero, Coey, Schmidt, and Richardson (2012) suggest that online coordinative tasks may produce variability in kinematics that relates not to action observation per se but instead to the task of coordinating one’s own movements with another’s. Here, our focus is on action priming, not coordinative action.

Method

The local institutional ethics committee approved all experimental stimuli and procedures reported in this article.

Participants

Twenty female undergraduate students aged 18-28 years (M = 20.80, standard deviation [SD] = 2.31) participated in exchange for course credit. Eighteen were right-handed and two were left-handed.

Apparatus

Participants sat at a table (60 x 100 x 70 cm³) covered by a black tablecloth. Between trials, participants rested their right hand on the table with thumb and index finger gently opposed atop a white spot aligned with their midsagittal axis. Participants’ reach target was the lower end of a cylindrical wooden dowel (1 cm diameter, 20 cm height) suspended vertically 5 cm above the table surface and 40 cm directly in front of the hand’s resting spot. Stimuli appeared on a 55-inch monitor (Samsung UA55D6600; length 1,247 mm, height 723 mm, diagonal 1,442 mm) at a resolution of 1,920 × 1,080 pixels in 32-bit colour at 60 Hz refresh rate. This monitor was approximately 40 cm from participants’ right shoulder and ran parallel to their midsagittal axis (see Figure 1).

Figure 1.

Representative photograph of apparatus setup for both experiments. Please note that the image on the monitor has colour distortion due to photographing the monitor and showed the colours displayed in Figure 2 during the experiment.

Behind the table (i.e., in front of the participant) were three Qualisys ProReflex infrared motion-capture cameras calibrated with respect to the table surface. Participants wore passive reflective markers (12 mm diameter) above the distal ends of the radius and ulna and on the index fingernail and thumbnail. The cameras recorded the markers’ locations at 100 Hz and generated a representation of each marker’s movement through three-dimensional Cartesian space.

We used E-Prime 2.0 (Psychology Software Tools, Pittsburgh, PA) to present stimuli and synchronise the ProReflex camera recordings with each trial.

Stimuli and procedure

First, participants were introduced to the motion capture cameras and had reflective markers attached to their hands. Participants then completed a familiarisation task, the main task, and a verbal debriefing.

Familiarisation task

This task consisted of 20 reach-to-grasp actions cued by the onset of a white rectangle on a black screen. The purpose was to reduce extraneous kinematic variability in participants’ reaches in the main task by allowing them to become familiar with the apparatus and accustomed to the reach-to-grasp action. Each trial consisted of a 1,000 ms white fixation cross on a black background, followed by a 1,000 ms white rectangle, and last a 6,500 ms black screen during which participants could resume the resting position. The experimenter (S.S.) instructed participants to reach directly towards the target at their natural speed when they saw the white rectangle, grip it for 2 s between thumb and index finger, then return their hand to the marked starting spot.

Main task

Films

Stimuli for the main task consisted of side-on films of a female model reaching to her own target dowel (identical in appearance to participants’). On each trial, she reached with one of two trajectories, direct or exaggerated (high-lifting)—see Figure 2. The films appeared on the 55-inch display such that the model appeared approximately life-sized, sitting beside the participant, and reaching parallel to them. Superimposed on this film was a semitransparent white dot. In cue-on-wrist trials, this dot traced the model’s wrist as she reached towards her target. On central cue trials, the dot remained motionless in a central location approximately midway between the peaks of her direct and exaggerated wrist trajectories. Simultaneously with the model gripping the target, this dot changed to either green or blue, which corresponded to go or no-go (colours counterbalanced across participants). Thus, there were go and no-go trials for each combination of reach trajectory (direct versus exaggerated) and cue location (cue-on-wrist vs. central cue).

Figure 2.

Example frames from exaggerated (left) and direct (right) reach trajectory films in Experiment 1, central cue condition. In the cue-on-wrist condition, the white dot traced the wrist of the model throughout the movement.

In all films, the model remained motionless for 167 ms after onset before initiating her reach. For direct reach trials, the model’s grip and the go signal occurred at 1,293 ms after film onset, and for exaggerated reach trials, 1,502 ms after film onset. After this, the model gripped the target and the go/no-go signal remained until 6,500 ms after film onset. A 1,000 ms black screen separated each trial.

Instructions

Participants were instructed to keep their eyes fixed on the white dot until it changed colour. If it changed to the go colour (blue or green, counterbalanced across participants), they were to execute a reach-to-grasp action the same as in the familiarisation block. Participants were permitted to look towards their target after they received the go signal. They were told that the person in the video may reach differently, but regardless of how she reached, they should always reach directly to their target at a natural pace. We informed participants that the model might reach differently to avoid confusion and violation of expectation for participants. In pilot studies, we found that unless we informed participants that the model would sometimes reach differently to them, a large proportion would begin intentionally imitating her reach or ask whether they should imitate it during the practice block. Conscious, intentional imitation is a different phenomenon from involuntary trajectory priming, and we wanted to avoid inadvertently conflating the two.

Block structure

Participants first completed a 16-trial practice block (3 go and 1 no-go trial[s] of each type) to confirm they understood the task, followed by two 36-trial test blocks (7 go and 2 no-go trials of each type). Trial order within each block was fully randomised. Participants could take rest breaks between blocks when desired.

Measures

Participants’ reaches were processed offline using MATLAB (Mathworks, Natick, MA). Maximum vertical deviation was calculated as the maximum positive z-plane deviation attained by the wrist above a straight-line path from its initial position to its position at grip. Grip was defined as the first frame after movement initiation that satisfied these criteria: (a) less than 1 mm change in the distance between the thumb and index markers compared with the previous frame (the amount of thumb and index separation should remain steady once the dowel is gripped), (b) less than 1 mm change in radius marker position from the previous frame (the wrist should remain steady at grip), and (c) radius marker at least 3 mm above resting position.

Kinematic data preparation and screening

Only go trials with reliable motion capture recordings of a complete reach-to-grasp action were included in the analysis. Initially, we excluded one participant who had fewer than four reliable motion-capture recordings per cell per block, and in one cell, only one usable trial. We excluded an additional participant due to recording problems on 23.21% of trials, but including this participant in the remainder of data screening and analysis did not alter the pattern of results. For the remaining 18 participants, an average of 2.58% of go trials were excluded due to recording problems (range 0%-10.71%). For these participants, we calculated maximum vertical deviation means for each observed trajectory within cue type condition within each block. To screen for within-subjects outliers, we computed difference scores based on these means for each cue type condition within each block by subtracting the mean for direct trials from the mean for exaggerated trials. We standardised these difference scores and screened for potential outliers (Z > 3.291, p < .001) to ensure that participants with extreme observed trajectory effects did not bias the results. This revealed one participant with extreme scores who was excluded from further analysis. This participant showed observed trajectory effects of greater than 81 mm in both cue conditions, all other participants showed observed trajectory effects of less than 10 mm. Thus, the analyses reported below use a final sample of 17 participants.

Results and discussion

Go-no-go task performance

No-go trials

A no-go trial counted as an error if the participant initiated movement at any time during the trial. Seven participants made between one and two no-go errors each; the remaining participants made no errors on no-go trials. The error rate therefore ranged from 0% to 14.29%, averaging 4.62%.

Go trials

A go trial counted as an error if participants either: (a) initiated movement less than 150 ms after the cue appeared (which indicates an anticipatory response rather than a cue-related response), (b) initiated movement but failed to grip the target during the allotted time, or (c) failed to initiate movement at all. No participants made errors on go trials.

Trials used for wrist trajectory analysis

After accounting for both unreliable motion capture recordings and participant errors on go trials in the final sample of 17 participants, an average of 97.69% of go trials per participant were eligible for the kinematic analysis (range 89.26%-100%).

Wrist trajectory analysis

We conducted within-groups 2 (Block: First Block, Second Block) x 2 (Cue Type: Cue-on-Wrist, Central Cue) x 2 (Observed Trajectory: Direct, Exaggerated) analysis of variance (ANOVA) with maximum vertical deviation as the dependent measure. The Block factor was included to capture any variance in trajectory related to fatigue or practice. Holm-Bonferroni corrections were applied to p values for follow-up t-tests. All reported standard errors of the mean (SEM) are within-subjects SEMs calculated according to the Morey-Cousineau method (Morey, 2008).

The ANOVA revealed a significant main effect of Observed Trajectory, F(1, 16) = 7.96, p = .012, η_p² = .33, qualified by a significant Cue Type by Observed Trajectory interaction, F(1, 16) = 5.13, p = .038, η_p² = .24, which is graphed in Figure 3. All other main effects and interactions were nonsignificant (see Table 1). Follow-up t-tests revealed a significant trajectory priming effect within the Cue-on-Wrist condition: Maximum vertical deviation was significantly higher, t(16) = 3.60, p = .010, SE_diff = 0.73, d_z = 0.87, following observation of an exaggerated trajectory (M = 24.90, within-subjects SEM = 0.54) compared with a direct (M = 22.53, within-subjects SEM = 0.67) trajectory t(16) = 3.60, p = .010, SEdiff = 0.73, dz = 0.87. In the Central Cue condition, there was no significant difference between viewing exaggerated (M = 23.59, within-subjects SEM = 0.57) and direct (M = 23.14, within-subjects SEM = 0.50) trajectories t(16) = 0.78, p = .446, SEdiff = 0.58, dz = 0.19.

Figure 3.

Maximum vertical deviation (mm) in Experiment 1 as a function of observed trajectory and cue type. Error bars denote one within-subjects SEM for each condition calculated according to the Morey-Cousineau method (Morey, 2008).

Table 1.

ANOVA effects for Experiment 1 analysis.

Effect	df	F	p	η_p²
Block	1, 16	0.98	.338	.06
Cue type	1, 16	0.39	.539	.02
Observed trajectory	1, 16	7.96	.012	.33
Block * Cue type	1, 16	2.67	.122	.14
Block * Observed trajectory	1, 16	1.14	.302	.07
Cue Type * Observed trajectory	1, 16	5.13	.038	.24
Block * Eye movement * Cue type	1, 16	0.03	.867	<.01

In summary, participants only showed evidence of trajectory priming in the cue-on-wrist condition, that is, when the white dot they were instructed to gaze at followed the model’s hand during her reach. In contrast, when the white dot remained stationary in the centre of the action space, the model’s trajectory did not affect participants’ trajectories.

These results suggest that a model’s action occurring outside of participants’ fixated area and locus of spatial attention is insufficient to evoke trajectory priming. The absence of detectable trajectory priming in the central-cue condition is consistent with previous data from initiation-time studies where motor priming was attenuated when spatial attention was diverted away from the relevant effector (Bach et al., 2007). The results are also consistent with the possibility that instructions to track the moving cue partially or wholly accounted for the trajectory priming effect in the cue-on-wrist condition. That is, the act of moving the eyes to follow either a relatively direct or exaggerated arc in cue-on-wrist trials may have biased wrist trajectory in participants’ subsequent reaches. If this wholly accounts for trajectory priming in the cue-on-wrist condition, then tracking a white dot should produce the same effects regardless of whether there is a human movement in the background or not.

Experiment 2

Introduction

In Experiment 2, the white dot cue always followed a direct or exaggerated trajectory, but we manipulated whether or not the human model moved along with it. In the human moves condition, the human model reached towards the target, always congruent with the dot’s motion such that the dot appeared to track her wrist smoothly. This condition was identical to the cue-on-wrist condition in Experiment 1. In the human still condition, the dot moved alone along the same exaggerated or direct trajectories while the human model remained still. We hypothesised that the trajectory priming effect would be weaker in the human still condition compared with the human moves condition. Gowen and Poliakoff (2012) indicate that when the task instructions or immediate context suggest that some movements, but not others, correspond to human agency or intention, action priming is weaker for seemingly non-human movements relative to seemingly human movements. We presented an equal number of human still and human moves trials within each block in random order, and it would therefore quickly become salient to participants that the human model and the dot moved independently. Thus, the dot did not appear linked to the intentions of the only visible human agent, and trajectory priming should be weaker in the human still condition (where only the dot moves) relative to the human moves condition (where the human also moves). We favoured this prediction because we have shown previously that trajectory priming effects are sensitive to visual cues to human agency (Sparks, Sidari, et al., 2016). Alternatively, if the trajectory priming effects are entirely due to shifting the focus of attention to track a moving object, and there is no unique contribution of the moving stimulus itself, the human moves and human still condition should produce equivalent trajectory priming effects because the white dot go cue follows the same path in both conditions.

Method

Participants

Twenty-five female undergraduate students aged 17-51 years (M = 23.76, SD = 9.20) participated in exchange for course credit. Twenty-four were right-handed and one left-handed.

Apparatus

The apparatus was identical to Experiment 1.

Stimuli and procedure

As in Experiment 1, participants were introduced to the motion capture cameras, had reflective markers attached to their hands, completed a familiarisation task, completed the main task, and received a verbal debriefing at the end of the session.

Familiarisation task

The familiarisation task was identical to Experiment 1 except that the duration of the intertrial interval (i.e., blank black screen) was 5,500 ms.

Main task

There were two types of trials: human moves and human still. Human moves trials were identical to the cue-on-wrist trials in Experiment 1, that is, the semitransparent white shape tracked the model’s wrist throughout its movement. In human still trials, the white circle followed the same direct or exaggerated trajectories used in human moves trials, whereas the human model remained motionless throughout. Thus, in human still trials, the white circle moved from its starting position towards the target independently of the model. In both trial types, the white circle turned green at the end of its trajectory to signal a go trial, and remained white on no-go trials. Instructions were the same as in Experiment 1 except that participants were told that the no-go signal was when the circle remained white.

Participants first completed a 16-trial practice block (3 go and 1 no-go trial[s] of each type) to confirm they understood the task, followed by two 40-trial test blocks (8 go and 2 no-go trials of each combination of observed trajectory and stimulus type). Trial order was randomised within blocks (participants encountered both the human moves and human still trials within each block). Participants could take rest breaks between blocks when desired.

Data preparation and screening

Motion capture recording screening

Only go trials with reliable motion capture recordings were included in the analysis. All participants had sufficient test trials to be included in the analysis. The proportion of unreliable motion capture recordings on go trials per participant ranged from 0% to 3.13%, averaging 0.25%. One participant had two go trials with unreliable motion capture recordings, two participants had one go trial with unreliable motion capture recording, and the remainder had no recording problems.

Outlier screening

We tested for potential outliers in the same manner as for Experiment 1. For each participant, we calculated mean differences for each observed trajectory effect (exaggerated trajectory minus direct trajectory) within the design and standardised these scores. This revealed one participant with extreme observed trajectory effects (Z > 3.291, p < .001) who was excluded from all analyses reported below. Thus, the analyses reported below use a final sample of 24 participants.

Results and discussion

Holm–Bonferroni corrections were applied to p values for follow-up t-tests reported below. In the ANOVAs, we included a Block factor to capture any variance in trajectory related to fatigue or practice. All reported SEMs are within-subjects SEMs calculated according to the Morey-Cousineau method (Morey, 2008).

Go-no-go task performance

The criteria for identifying errors on go and no-go trials were the same as for Experiment 1.

No-go trials

Participants’ error rates for no-go trials ranged from 0% to 18.75% and averaged at 2.34%. Six (of 24) participants made between one and three errors on no-go trials. The remainder made none.

Go trials

Participants’ error rates on go trials ranged from 0% to 4.69%, averaging 0.61%. Seven (of 24) participants made between one and three errors on go trials. The remainder made no errors on go trials.

Trials used for wrist trajectory analysis

After accounting for both unreliable motion capture recordings and participant errors on go trials in the final sample of 24 participants, an average of 98.96% of go trials per participant were eligible for the kinematic analysis (range 93.75%-100%).

Wrist trajectory analysis

We conducted within-groups 2 (Block: First Block, Second Block) x 2 (Stimulus Type: human moves, human still) x 2 (Observed Trajectory: Direct, Exaggerated) ANOVA with maximum vertical deviation as the dependent measure. The ANOVA revealed significant main effects of Observed Trajectory, F(1, 23) = 8.08, p = .009, η_p² = .26, qualified by a significant Stimulus Type by Observed Trajectory interaction, F(1, 23) = 4.47, p = .045, η_p² = .16, which is graphed in Figure 4. All other main effects and interactions were nonsignificant (see Table 2).

Figure 4.

Maximum vertical deviation (mm) in Experiment 2 plotted with respect to observed trajectory and human movement condition. Error bars denote one within-subjects SEM for each condition calculated according to the Morey-Cousineau method (Morey, 2008).

Table 2.

ANOVA effects for Experiment 2 analysis.

Effect	df	F	p	η_p²
Block	1, 23	<.01	.990	<.01
Stimulus type	1, 23	4.23	.051	.16
Observed trajectory	1, 23	8.08	.009	.26
Block * Stimulus type	1, 23	0.23	.635	.01
Block * Observed trajectory	1, 23	0.12	.728	.01
Stimulus type * Observed trajectory	1, 23	4.47	.045	.16
Block * Stimulus type * Observed trajectory	1, 23	0.563	.461	.02

Follow-up t-tests revealed a significant trajectory priming effect within the human moves condition: Maximum vertical deviation was significantly higher, following observation of exaggerated (M = 26.70, within-subjects SEM = 0.72) compared with direct (M = 22.96, within-subjects SEM = 0.83) trajectories t(24) = 3.33, p = .006, SE_diff = 1.12, d_z = 0.68. In the human still condition, this difference was smaller and non-significant, t(24) = 2.06, p = .051, SE_diff = 1.10, d_z = 0.42; exaggerated: M = 25.19, within-subjects SEM = 0.85; direct: M = 22.92, within-subjects SEM = 0.72.

The results from Experiment 2 suggest that instructions to gaze at a moving object cannot account fully for the observed effects. Participants showed a significant trajectory priming effect in the human moves condition but an attenuated, non-significant trajectory priming effect in the human still condition. If instructions to gaze at and attend to a moving stimulus were solely responsible for trajectory priming in this paradigm, we would not expect this interaction because the instructed gaze paths are the same in both conditions. In addition, when considered in the context of previous studies (Sparks, Sidari, et al., 2016), these results suggest that whether or not a non-human moving object evokes trajectory priming depends on the immediate context, a point we consider further in the General Discussion below.

General discussion

In Experiment 1, participants attempted to execute direct reach-to-grasp actions to a dowel target after observing a film of a model executing a direct or exaggerated (high-lifting) reach to her own target. We instructed participants to fixate on (and spatially attend to) a white dot and monitor it for their go signal. The dot either remained stationary or followed the model’s wrist along its trajectory. Participants showed significant wrist trajectory priming when the dot tracked the model’s wrist (cue-on-wrist condition), but not when it remained stationary in a central location (central cue condition).

In Experiment 2, in the human moves condition, the dot again tracked the model’s wrist—this was identical to the cue-on-wrist condition in Experiment 1. In the human still condition, the model remained stationary while the dot moved alone through the same two trajectories (direct or exaggerated) that it followed in the human moves trials. Thus, the only difference between human moves and human still trials was whether the human model moved in the background. There was a significant stimulus type (human moves vs. still) by observed trajectory (direct vs. exaggerated) interaction. Follow-up tests showed that a significant wrist trajectory priming effect occurred in the human moves condition where the model herself moved, whereas this effect was smaller in magnitude and nonsignificant in the human still condition.

Before interpreting the data further, we would like to note that our data should not be taken to show that no trajectory priming occurs in the central cue condition of Experiment 1 and model still condition of Experiment 2. The significant interactions support our argument that both of these conditions dampen trajectory priming relative to when the model herself moves and the dot traces her wrist. Our experiments, however, lacked the statistical power to detect smaller trajectory priming effects in these conditions. For example, in Experiment 2, the effect size for the human still condition was 2.27 mm (d_z = 0.42) versus 3.74 mm (d_z = 0.68) in the human moves condition. Supposing that the human still condition does in fact produce a smaller trajectory priming effect (i.e., d_z = 0.42), a sample size of 75 would be required to detect this effect reliably (i.e., achieve 95% experimental power), whereas our final sample size was 24 (yielding 37% experimental power). Our main aim with the current experiments was to identify factors that modulate trajectory priming rather than to determine whether the effect can be completely abolished. In future, larger samples will be necessary to distinguish whether such factors partially attenuate or fully abolish the effect.

The attenuation of trajectory priming in the central cue condition of Experiment 1 demonstrates that instructions regarding attention can constrain trajectory priming. There are two broad explanations for this instruction effect. The first possibility is that spatial attention constrains wrist trajectory priming. In both the central cue condition and the cue on wrist condition, participants were instructed to attend selectively and spatially to a white dot and not to be influenced by the human model’s actions. Thus, in both conditions, participants were instructed to selectively attend to something other than the human model. In the central cue condition, the dot appeared and remained stationary in a central location so there was no need for participants to attend spatially to the area of the screen where the human model’s wrist moved. In the cue-on-wrist condition, the white dot spatially overlapped the model’s wrist. Thus, one interpretation of these data is that trajectory priming diminishes when the observer directs spatial attention away from the relevant body site, in this case, the wrist. We cannot rule out, however, that participants spatially attended to the model’s action to some extent even in the central cue condition. In particular, the onset of the model’s action could have captured spatial attention (Abrams & Christ, 2003; Guo, Abrams, Moscovitch, & Pratt, 2010), although we cannot verify if this was the case. If participants spatially attended the model’s movement on some central cue trials, this was insufficient to produce a statistically detectable trajectory priming effect.

Another possible explanation for these data is the task relevance of the human model’s movement. In Experiment 1, trial type was randomised, but at the start of each trial participants could observe the starting location of the white dot and use this to infer whether it would follow the model’s wrist. If the dot started in the central location rather than on her wrist, it would always remain stationary in that position and therefore the model’s movement would be entirely irrelevant to the dot’s location throughout the remainder of the trial. If the dot started on the model’s wrist, it would always follow her wrist, and therefore the location of the model’s wrist would correlate with the location of the dot for the remainder of the trial. Thus, it is possible that participants did treat the model’s movements as task-relevant in the cue-on-wrist condition and selectively attended to it to some extent. In both the spatial-attention account and the task-relevancy-via-selective-attention account, it seems clear that top-down allocation of attention modulates trajectory priming in reach to grasp actions, and this is a novel finding.

Our findings here are similar to those of Constable et al. (2017) who reported that instructing participants to fixate on a central location (rather than the model’s hand) abolished motor contagion in a modification of Kilner et al.’s (2003) paradigm. Constable et al.’s task required participants to execute arm waving movements in time with observed arm waving movements, which had either a congruent or incongruent trajectory to their own. It is not fully clear whether mechanisms responsible for online motor contagion in that paradigm are similar or distinct from those responsible for offline trajectory priming in our paradigm. Romero et al. (2012) suggest that rather than being caused by motor contagion per se, the kinematic interference in such motor contagion tasks may arise from mechanisms responsible for executing actions entrained to a rhythmic motion (i.e., the observed arm waves). Also relevant is that Constable et al. (2017) did not have visible objects as end points whereas our human movement stimuli ended in grasping a target. Wild et al. (2012) showed that participants instructed to imitate a model’s actions were more likely to track pointing hand movements with their eyes when there was no clear end-point for the movement compared with the same movements with a visible end-point. This suggests that there may be differences in the way people naturally attend to such actions. It would be interesting to clarify further the relationship between these paradigms in future studies. If evidence accumulates showing attentional instructions and constraints affect both paradigms in similar ways, it would be consistent with a common mechanism supporting both effects.

It is possible that shifting the locus of attention to track a moving object explains the trajectory priming effect if we consider the data of Experiment 1 alone. In Experiment 1, participants only had to track a moving object in the cue-on-wrist condition, which was where significant trajectory priming occurred. We suggest, however, that the results of Experiment 2 are difficult to explain if tracking a moving object entirely accounts for trajectory priming in this paradigm. If this were the case, it is unclear why trajectory priming in the human still condition of Experiment 2 was attenuated compared with the human moves condition. In Experiment 2, participants were instructed to gaze at the white dot, which moved along identical trajectories in the human movement condition and the human still movement condition. Without eye tracking data, we cannot definitively state that participants obeyed our instruction to fixate on the white dot in both conditions. Regardless, there is no clear reason that they would obey this instruction in the human moves condition and not the human still condition, or vice versa. In both the human moves and human still conditions, the go signal always occurred at the very end of the dot’s trajectory. If participants chose to ignore the gaze instructions, the easiest strategy to complete the task would be simply to gaze at and attend to the area of the screen where the dot always finished its movement and changed colour. This path-of-least-resistance strategy would not involve shifting the locus of attention at all. Future studies can confirm this more definitively using eye tracking equipment.

The attenuation of trajectory priming in human still condition of Experiment 2 (relative to human moves condition) suggests that observing an unambiguous human movement has a stronger effect on participants’ motor output than observing a non-human stimulus moving along a very similar trajectory. This is particularly interesting given the possibility that observing the human model reaching in tandem with the dot on human moves trials could prime participants to simulate a human reach even on the human still trials when the dot moved independently of the model’s hand. Gowen and Poliakoff (2012) propose that if contextual cues (such as these) suggest that a non-human stimulus represents a human action, participants are more likely to show an action priming effect. We cannot determine if such a priming effect did in fact occur, but if so, it would work against our experiment detecting the significant interaction between observed trajectory and the human model’s movement.

The results of Experiment 2 should be interpreted in the context of our previous study that manipulated visual cues to human agency (Sparks, Sidari, et al., 2016). In Sparks, Sidari, et al. (2016), we used the same direct and exaggerated human reaches and dot trajectories as the current study. In the first block, we showed that observing the dot’s motion produced no significant trajectory priming in the presence of only a scrambled, non-meaningful image of the human model. Critically, participants had never viewed the model’s unscrambled image or viewed her reaching before this block. In the second block where the human image was unscrambled but remained completely motionless, the same moving dots produced significant trajectory priming when even though the human model remained still. The dot-movement trajectories and still human model in this block of Sparks, Sidari, et al. (2016) were identical in the human still condition of the current study, but in the current study, we did not detect significant trajectory priming with these stimuli (but see earlier note regarding experimental power). In a final stage of Sparks, Sidari, et al. (2016), the model herself reached to the target and the dot always tracked her wrist in an identical manner to the human moves condition in the current study. In contrast to these stimuli in the current study, which produced a significantly larger trajectory priming effect than the human still stimuli, Sparks, Sidari, et al.’s (2016) equivalent condition did not significantly increase trajectory priming compared with the human still stimuli. The key difference between these studies is that in the human still condition of the Sparks, Sidari, et al. (2016) experiment, the addition of a previously absent human image would make the possible human agency of the dot highly salient compared with the previous block where there was no visual cue suggestive of human intent or agency. In contrast, in Experiment 2 of the present study, we presented in random order trials in which the dot moved of its own accord and trials in which the human model also moved within the same block. In line with Gowen and Poliakoff (2012), we suggest that this trial-by-trial contrast between human moves and human still trials would make salient the independence of the dot from the human model. Thus, this within-block contrast would serve as a contextual cue that the dot’s motion did not represent the human agent’s agency or intentions. This could explain why in the current experiment the dot moving while the human remained still failed to produce a trajectory priming effect compared with the robust, significant effect a moving dot with a still human image produced in Sparks, Sidari, et al. (2016). Together, the results of Experiment 2 and Sparks, Sidari, et al. (2016) fit very well with the notion that contextual cues to human agency and/or intention are a key factor determining whether action priming occurs for ambiguous stimuli.

The results from both experiments can be situated within Gowen and Poliakoff’s (2012) two-route model of motor priming effects (within which they include reach trajectory priming). They postulate that action priming arises from a combination of a visuomotor route and a modulating route. The visuomotor route includes action goals, low-level resonance, stimulus-response compatibility, and processing of kinematics. The modulating route comprises factors such as beliefs about agency, similarity with the model, prior knowledge, and top-down attentional control. Our findings inform our understanding of attentional factors in the modulating route. In at least some contexts, such as the cue-on-wrist condition of Experiment 1 and the human moves condition of Experiment 2, top-down application of selective attention and inhibition are not sufficient to prevent observed kinematics influencing executed actions. The lack of trajectory priming in the central cue condition of Experiment 1 suggests that some level of attention to the key body sites is probably necessary for the emergence of reach trajectory priming. Perhaps when actions occur outside the locus of attention, processing of the action is dampened to the point where observers can successfully inhibit the influence of observed actions on their own actions. In addition, as mentioned earlier, trajectory priming is weakened if top-down contextual factors suggest the moving stimulus does not reflect human agency.

It is important to note that in both experiments we warned participants that (a) sometimes the model (or dot) would move differently to the way they did and (b) they should not copy the model, that is, they should reach directly for the target regardless of how she moved. As mentioned earlier, we arrived at this approach after a pilot run where we did not warn participants. In this no-warning pilot run, a substantial number of participants either (a) spontaneously began intentionally imitating the model’s exaggerated reaches despite the experimenter clearly asking them to reach directly before the task began or (b) stopped doing the task and asked the experimenter if they were supposed to copy the model or not. Our final instructions with an explicit warning prevented these problems and ensured that we were not biasing the data by excluding the subgroups of participants mentioned above. It is possible, however, that this warning acted as a top-down attentional influence in its own right. We cannot rule out that it increased the overall salience of the model’s actions, and therefore increased the tendency of participants to show involuntary trajectory priming in both experiments. Consistent with this possibility, in an intentional imitation task, Bek, Poliakoff, Marshall, Trueman and Gowen (2016) reported increased imitation of hand trajectory when participants were verbally instructed to attend to the nature of an observed movement. It would be interesting to attempt a similar instructed attention manipulation in an involuntary trajectory priming paradigm, if this can be done without confusing participants. If the warning did increase the propensity for trajectory priming, the lack of trajectory priming in the central cue condition of Experiment 1 still suggests that instructing participants to attend spatially and selectively away from the observed action can dampen or abolish involuntary imitation of the trajectory. In Experiment 2, the warning mentioned the dot and the human model and it is unclear whether it could have differentially influenced the salience of the dot in the human-still condition versus the hand/arm in the human moves condition.

One caveat to our study is that we have only tested female participants with a female model. This choice was deliberate. There is evidence that females may show higher levels of motor resonance (Cheng et al., 2008) and are more sensitive to social attentional cues (Bayliss, Pellegrino, & Tipper, 2005). In addition, this avoided dissimilarity between observer and model gender, which may have affected trajectory priming levels. It will be important to test whether these effects emerge in males. Other individual differences may also be relevant, for example, differences in impulsivity and inhibitory control.

In conclusion, our results show that top-down attentional factors can modulate trajectory priming and confirm the importance of instructions regarding gaze and attention in these paradigms. A human model’s actions occurring within the field of view did not produce wrist trajectory priming when participants did not need to spatially or selectively attend the model’s wrist. In addition, attending to a moving stimulus with a similar trajectory to the model’s wrist produced, at best, an attenuated trajectory priming effect relative to a human reach when the context made salient that the dot moved independently of the human.

Footnotes

Acknowledgements

We thank the reviewers and editor for their insightful comments that greatly contributed to this manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by ARC DP130100253 to AK.

References

Abrams

R. A.

Christ

S. E.

(2003). Motion onset captures attention. Psychological Science, 14, 427–432. doi:10.1111/1467-9280.01458

Bach

Peatfield

N. A.

Tipper

S. P.

(2007). Focusing on body sites: The role of spatial attention in action perception. Experimental Brain Research, 178, 509–517. doi:10.1007/s00221-006-0756-4

Bayliss

A. P.

Pellegrino

G. D.

Tipper

S. P.

(2005). Sex differences in eye gaze and symbolic cueing of attention. Quarterly Journal of Experimental Psychology Section A: Human Experimental Psychology, 58, 631–650. doi:10.1080/02724980443000124

Bek

Poliakoff

Marshall

Trueman

Gowen

(2016). Enhancing voluntary imitation through attention and motor imagery. Experimental Brain Research, 234, 1819–1828.

Brass

Bekkering

Prinz

(2001). Movement observation affects movement execution in a simple response task. Acta Psychologica, 106, 3–22. doi:10.1016/S0001-6918(00)00024-X

Brass

Bekkering

Wohlschläger

Prinz

(2000). Compatibility between observed and executed finger movements: Comparing symbolic, spatial, and imitative cues. Brain and Cognition, 44, 124–143. doi:10.1006/brcg.2000.1225

Cheng

Lee

P. L.

Yang

C. Y.

Lin

C. P.

Hung

Decety

(2008). Gender differences in the mu rhythm of the human mirror-neuron system. PLoS ONE, 3, e2113. doi:10.1371/journal.pone.0002113

Chong

T. T. J.

Cunnington

Williams

M. A.

Mattingley

J. B.

(2009). The role of selective attention in matching observed and executed actions. Neuropsychologia, 47, 786–795. doi:10.1016/j.neuropsychologia.2008.12.008

Constable

M. D.

de Grosbois

Lung

Tremblay

Pratt

Welsh

T. N.

(2017). Eye movements may cause motor contagion effects. Psychonomic Bulletin & Review, 24, 835–841. doi:10.3758/s13423-016-1177-4

10.

Cook

Bird

(2011). Social attitudes differentially modulate imitation in adolescents and adults. Experimental Brain Research, 211, 601–612. doi:10.1007/s00221-011-2584-4

11.

Cracco

Bardi

Desmet

Genschow

Rigoni

De Coster

, . . . Brass

(2018). Automatic Imitation: A Meta-Analysis. Psychological Bulletin, 144, 453–500. doi:10.1037/bul0000143

12.

Gowen

Bolton

Poliakoff

(2016). Believe it or not: Moving non-biological stimuli believed to have human origin can be represented as human movement. Cognition, 146, 431–438. doi:10.1016/j.cognition.2015.10.010

13.

Gowen

Bradshaw

Galpin

Lawrence

Poliakoff

(2010). Exploring visuomotor priming following biological and non-biological stimuli. Brain and Cognition, 74, 288–297. doi:10.1016/j.bandc.2010.08.010

14.

Gowen

Poliakoff

(2012). How does visuomotor priming differ for biological and non-biological stimuli? A review of the evidence. Psychological Research, 76, 407–420. doi:10.1007/s00426-011-0389-5

15.

Griffiths

Tipper

S. P.

(2009). Priming of reach trajectory when observing actions: Hand-centred effects. Quarterly Journal of Experimental Psychology, 62, 2450–2470. doi:10.1080/17470210903103059

16.

Griffiths

Tipper

S. P.

(2012). When far becomes near: Shared environments activate action simulation. Quarterly Journal of Experimental Psychology, 65, 1241–1249. doi:10.1080/17470218.2012.688978

17.

Guo

R. M.

Abrams

R. A.

Moscovitch

Pratt

(2010). Isoluminant motion onset captures attention. Attention, Perception & Psychophysics, 72, 1311–1316. doi:10.3758/APP.72.5.1311

18.

Hardwick

R. M.

Edwards

M. G.

(2011). Observed reach trajectory influences executed reach kinematics in prehension. Quarterly Journal of Experimental Psychology, 64, 1082–1093. doi:10.1080/17470218.2010.538068

19.

Heyes

(2011). Automatic imitation. Psychological Bulletin, 137, 463–483. doi:10.1037/a0022288

20.

Jackson

P. L.

Meltzoff

A. N.

Decety

(2006). Neural circuits involved in imitation and perspective-taking. Neuroimage, 31, 429–439. doi:10.1016/j.neuroimage.2005.11.026

21.

Kilner

J. M.

Paulignan

Blakemore

S. J.

(2003). An interference effect of observed biological movement on action. Current Biology, 13, 522–525. doi:10.1016/S0960-9822(03)00165-9

22.

Liepelt

Von Cramon

D. Y.

Brass

(2008). What is matched in direct matching? Intention attribution modulates motor priming. Journal of Experimental Psychology: Human Perception and Performance, 34, 578–591. doi:10.1037/0096-1523.34.3.578

23.

Morey

R. D.

(2008). Confidence intervals from normalized data: A correction to Cousineau (2005). Tutorials in Quantitative Methods for Psychology, 4, 61–64.

24.

Press

Bird

Walsh

Heyes

(2008). Automatic imitation of intransitive actions. Brain and Cognition, 67, 44–50. doi:10.1016/j.bandc.2007.11.001

25.

Romero

Coey

Schmidt

R. C.

Richardson

M. J.

(2012). Movement coordination or movement interference: Visual tracking and spontaneous coordination modulate rhythmic movement interference. PLOS ONE, 7, e44761. doi:10.1371/journal.pone.0044761

26.

Sparks

Douglas

Kritikos

(2016). Verbal social primes alter motor contagion during action observation. Quarterly Journal of Experimental Psychology, 69, 1041–1048. doi:10.1080/17470218.2015.1113304

27.

Sparks

Sidari

Lyons

Kritikos

(2016). Pictures of you: Dot stimuli cause motor contagion in presence of a still human form. Consciousness and Cognition, 45, 135–145. doi:10.1016/j.concog.2016.08.004

28.

Stanley

Gowen

Miall

R. C.

(2007). Effects of agency on movement interference during observation of a moving dot stimulus. Journal of Experimental Psychology: Human Perception and Performance, 33, 915–926. doi:10.1037/0096-1523.33.4.915

29.

Stürmer

Aschersleben

Prinz

(2000). Correspondence effects with manual gestures and postures: A study of imitation. Journal of Experimental Psychology: Human Perception and Performance, 26, 1746–1759. doi:10.1037//0096-1523.26.6.1746

30.

Wild

K. S.

Poliakoff

Jerrison

Gowen

(2012). Goal-directed and goal-less imitation in autism spectrum disorder. Journal of Autism and Developmental Disorders, 42 1739–1749.