Believe it or not: Moving non-biological stimuli believed to have human origin can be represented as human movement

Cognition 146 (2016) 431–438

Contents lists available at ScienceDirect

Cognition journal homepage: www.elsevier.com/locate/COGNIT

Believe it or not: Moving non-biological stimuli believed to have human origin can be represented as human movement E. Gowen a,⇑, E. Bolton a, E. Poliakoff b a b

Faculty of Life Sciences, University of Manchester, UK School of Psychological Sciences, University of Manchester, UK

a r t i c l e

i n f o

Article history: Received 9 December 2014 Revised 28 July 2015 Accepted 13 October 2015

Keywords: Visuomotor priming Automatic imitation Mirror neuron system Stimulus response compatibility Human–robot interaction

a b s t r a c t Does our brain treat non-biological movements (e.g. moving abstract shapes or robots) in the same way as human movements? The current work tested whether the movement of a non-biological rectangular object, believed to be based on a human action is represented within the observer’s motor system. A novel visuomotor priming task was designed to pit true imitative compatibility, due to human action representation against more general stimulus response compatibility that has confounded previous belief experiments. Stimulus response compatibility effects were found for the object. However, imitative compatibility was found when participants repeated the object task with the belief that the object was based on a human finger movement, and when they performed the task viewing a real human hand. These results provide the first demonstration that non-biological stimuli can be represented as a human movement if they are believed to have human agency and have implications for interactions with technology and robots. Crown Copyright Ó 2015 Published by Elsevier B.V. All rights reserved.

1. Introduction It is well known that observation of a human action can influence the observer’s own motor system. For example, observing another person’s action activates brain areas involved in execution of that action (Gazzola & Keysers, 2009; Kilner, Neal, Weiskopf, Friston, & Frith, 2009) and can interfere with or facilitate movement production (Brass, Bekkering, & Prinz, 2001; Sturmer, Aschersleben, & Prinz, 2000). These effects are thought to be due to the mirror neuron system (MNS) present within the premotor cortex and inferior parietal lobe that responds during both observation and execution of an action (Buccino et al., 2001; Rizzolatti & Craighero, 2004; Van Overwalle & Baetens, 2009). Recently, there has been increasing interest in whether non-biological stimuli (e.g. abstract shapes or robots) are processed in a similar way to human actions, leading to non-biological movements being represented within the observer’s motor system (Gowen & Poliakoff, 2012; Press, 2011). Measuring whether non-biological movements are represented in a similar way to human actions could indicate the success of human–robot interaction which is particularly relevant as humanoid robots are likely to increasingly play a role in society, ⇑ Corresponding author at: Carys Bannister Building, Dover Street, Manchester M13 9PL, UK. E-mail address: [email protected] (E. Gowen). http://dx.doi.org/10.1016/j.cognition.2015.10.010 0010-0277/Crown Copyright Ó 2015 Published by Elsevier B.V. All rights reserved.

such as in healthcare, education and entertainment (Andrade et al., 2014; Chaminade & Cheng, 2009; Dautenhahn, 2007; Tapus, Mataric, & Scassellati, 2007). More controversially, representing the action of a non-human agent may suggest the attribution of characteristics associated with humans such as mental states to non-human stimuli (Chaminade & Cheng, 2009). In this work, we address whether belief that a non-biological stimulus is based on a human action produces action representation, using a behavioural visuomotor priming task. In visuomotor priming, also termed Automatic Imitation, observing and performing a compatible action (e.g. lifting one’s index finger while observing another index finger move upwards) facilitates reaction times, whereas reaction times are slowed when observing a movement incompatible with a performed action (e.g. lifting one’s index finger while observing a finger press; Brass et al., 2001). As visuomotor priming is likely to result from activation of the MNS (Catmur, Walsh, & Heyes, 2009; Heyes, 2011), it provides a behavioural measure of whether an action is represented within the observer’s motor system. Although previous studies have compared visuomotor priming for human and nonbiological movements (Gowen, Bradshaw, Galpin, Lawrence, & Poliakoff, 2010; Jansson, Wilson, Williams, & Mon-Williams, 2007; Press, Bird, Flach, & Heyes, 2005) these are confounded by Stimulus Response Compatibility effects, whereby responses are faster to spatially or directionally aligned stimuli (Cho & Proctor,

432

E. Gowen et al. / Cognition 146 (2016) 431–438

2003). Consequently, visuomotor priming to non-biological stimuli could result purely from stimulus response compatibility effects in the absence of action representation (Jansson et al., 2007). Models of visuomotor priming share the idea that priming occurs along a visuomotor route, transforming visual input into a motor response and that priming produced by stimulus response compatibility and imitative compatibility are dissociated with the latter involving the MNS (Gowen & Poliakoff, 2012; Heyes, 2011; Wang & Hamilton, 2012). This visuomotor route is modulated by top-down factors such as attention, prior knowledge and social cognitive processes which can exert influence at the early sensory input stage or at the later motor, output stage (Gowen & Poliakoff, 2012; Heyes, 2011). One top-down influence, termed belief refers to prior knowledge or assumptions that a person has about the observed stimulus. For example, visuomotor priming is greater if a person believes (having received explicit instruction) that the non-biological stimulus is created from a human movement (Liepelt & Brass, 2010; Shen, Kose-Bagci, Saunders, & Dautenhahn, 2011; Stanley, Gowen, & Miall, 2007), whereas belief that a hand is virtual can reduce priming (Longo & Bertenthal, 2009). A more spontaneous or implicit form of belief could also occur for non-biological stimuli that have human characteristics (e.g. a robot) or for non-biological stimuli that are presented in a similar context to a previous human stimulus (Stanley et al., 2007). On the one hand, these results could suggest that belief produces action representation for non-biological stimuli by activating the MNS at the input stage or enhancing the MNS at the output stage. However, it could be that implicit or explicit belief merely alters attention to the stimulus movement, which either enhances or reduces stimulus response compatibility effects via the input route, without activating the MNS (Heyes, 2011; Press, 2011). Consequently, it is still unknown whether a non-biological movement can produce action representation equivalent to a human movement. The aim of this work was to resolve these issues by separating imitative compatibility, due to action representation, from more basic stimulus response compatibility effects. We used a modified version of the visuomotor priming task where participants observe an index finger or blue rectangular object moving upwards or

downward and must respond when they observe a go signal in the form of a yellow flash (Fig. 1). Participants viewed a right hand rotated 90 degrees counter clockwise (from the participant’s viewpoint), in a ‘‘thumb up” orientation and were required to make a key press response with their left hand. This stimulus orientation and response combination separated three stimulus response compatibility effects from imitative compatibility (Fig. 2). Directional stimulus response compatibility effects were removed by rotating the hand so that up/down index finger movements now became left/right movements. However, rotating the hand introduces two further potential stimulus response compatibility effects (i) leftdown and up-right stimulus response pairings are faster (orthogonal stimulus response compatibility; Weeks & Proctor, 1990); (ii) an advantage when the stimulus and response are on the same side of space (Simon effect; Simon, 1990). By using a ‘‘thumb up” orientation together with a left handed pressing response (Fig. 2) we were able to isolate imitative compatibility from both orthogonal stimulus response compatibility and the Simon effect. Thus, when the finger moves leftward across the screen, this is compatible in terms of the Simon effect (left advantage due to using left hand to response) and orthogonal stimulus response compatibility (down-left advantage when pressing button), but is imitatively incompatible (downward response, observing upward finger movement). However, when the finger moves rightward across the screen, this is incompatible in terms of the Simon effect and orthogonal stimulus response compatibility, but is imitatively compatible (downward response, observing downward finger movement). Participants responded to the go signal under three stimulus conditions. Firstly, they carried out the task while observing the object (object condition). Next, they responded while observing the object following a belief manipulation informing them that the object was based on the movement of a human index finger (belief condition). Lastly, they performed the task with the real human hand (hand condition). We hypothesized that orthogonal spatial compatibility/Simon effects would be present in the initial object condition and that imitative compatibility would be present for the hand stimulus. However, in the belief condition, there were two possibilities: (1) There would be an increase in orthogonal

Fig. 1. Time course of one trial for the hand (top) and object stimuli (bottom). Trial starts at left of picture in neutral position and shows a downward movement for both stimulus types. The yellow go signal is presented for 80 ms at 0, 120 or 280 ms following the start frame. A second end frame is presented for the 280 ms go signal. Pictures to right of dashed line show position of flash on stimuli. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

E. Gowen et al. / Cognition 146 (2016) 431–438

Left

Right

Imitation=incompatible

Imitation=compatible

Simon effect=compatible

Simon effect=incompatible

Orthogonal =compatible

433

Orthogonal =incompatible

Response: press (down) using left hand

Fig. 2. Dissociating imitative and stimulus–response compatibility. The participant’s response (downward key press) shares imitative compatibility with the finger when it moves rightward across the screen (downward finger movement), but is incompatible when the finger moves leftward across the screen (upward finger movement). In contrast, stimulus response compatibility effects are compatible with a leftward movement: the leftward stimulus movement is compatible with the left hand response (Simon effect) and the downward response (orthogonal spatial compatibility; right-up left-down).

spatial compatibility/Simon effect if the belief effect produced increased attention to the stimulus or (2) Imitative compatibility in the opposite direction, similar to the hand would be present if the belief effect is caused by representation of a human movement. In this case, as stimulus response compatibility can be stronger than imitative compatibility (Longo & Bertenthal, 2009; Jimenez et al., 2012) we expected the imitative compatibility effects to be smaller than previously reported, but importantly in the opposite direction to stimulus response compatibility effects. 2. Methods 2.1. Participants 20 right-handed healthy individuals were recruited for the study, but the data of two participants was excluded due to failure to comply with instructions or a high percentage (>20%) of responses made prior to the go signal. The mean age (±SD) of the remaining 18 participants (9 female) was 22.94 (±4.71). As the task was new, a power calculation was not possible. However, an N of 20 was chosen based on earlier versions of the task that showed a significant interaction between stimulus condition (object/ hand/belief) and compatibility, with a large effect size of gp2 = 0.2. The study was approved by the University of Manchester Research Ethics Committee and carried out in accordance with the provisions of the World Medical Association Declaration of Helsinki. 2.2. Stimuli Participants sat 80 cm from a flat screen monitor, upon which sequences of stimuli were displayed using Presentation Software (Neurobehavioral Systems). Stimuli consisted of either a right hand rotated 90 degrees counter clockwise, in a ‘‘thumb up orientation” or a blue rectangular object in an equivalent position (Fig. 1). Participants observed the object or the index finger of the hand moving in a leftward or rightward direction across the screen (i.e. up/down movement of the finger) and were required to make a press response using their left index finger, when they observed a yellow flash appear (Fig. 1). Key presses were recorded on a separate keypad that was positioned centrally in front of the computer screen, so that the participant’s left hand was aligned with the centre of the image on screen. This stimulus orientation and response combination separated the Simon effect (left advantage due to using left hand to response) and orthogonal spatial compatibility effects (down-left advantage when pressing button) from imitative

compatibility (right stimulus movement, representing downward movement, imitatively compatible with downward key press) (Fig. 2). The hand stimuli were created by converting digital .avi files into a sequence of 9 still frames consisting of an initial start frame (finger in a neutral position) and 7 movement frames. The initial start frame was presented for 1600 ms and was identical for upward and downward movements, ensuring it was not possible to predict the movement direction. The 7 movement frames (presented for 40 ms each) depicted the finger making an upward (33 pixels or 12.6 mm) or downward (27 pixels or 10.3 mm) movement. The movement of both the hand and object had a biological profile, accelerating toward the middle and decelerating towards the end. To plot the trajectory in the object movement condition, a blue rectangle, was positioned over the moving index finger in each frame. However, the hand was not visible in the resulting clips in this condition; instead the rectangle was shown alone over a background constructed to resemble that used in the hand clips. The luminance (142 cd/m2) and size (5.5  1.8 cm) of the rectangle were matched to the finger of the hand. The yellow flash 9.8  1.6 cm and was presented for 80 ms. It was positioned over the finger or object so that it covered the full extent of the movement (Fig. 1).

2.3. Procedure The experiment was split into three blocks. Firstly, the object was presented, followed by a questionnaire asking whether participants found that the rectangle made them think of a human finger movement (Table 2, questions 1–2). Secondly, participants were told that the object movement was generated from a human finger movement. This was demonstrated using the experimenter’s own hand, as well as using a printed picture depicting how the hand had been rotated. Therefore, the way the hand was orientated and the direction of movement (upwards = rightwards; downwards = leftward) was made clear to the participants and the object was referred to as ‘the finger’, moving ‘upwards/down wards’. They then repeated the task. Following the belief condition, participants completed a second short questionnaire, asking to what extent they believed that the object movement was based on a human finger movement (Table 2, questions 3–5). Thirdly, participants observed the hand stimulus. Each trial consisted of the 9 frames detailed above depicting the moving finger or object (Fig. 1). The yellow flash go signal appeared at one of three different stimulus onset asynchronies (SOAs) following the start frame (0, 120, 280 ms). The three SOAs were

434

E. Gowen et al. / Cognition 146 (2016) 431–438

chosen to assess the strength of priming at the start, middle and end of the movement. As the 280 ms SOA occurred after the end frame, a second end frame was presented. In the case of no-go trials, the flash did not appear. For baseline trials, no movement frames were presented. Instead, the flash appeared after the start frame, followed again by the start frame for 40 ms. Both the nogo and baseline trials decreased predictability of the flash go signal. Trials were terminated if the participant made a response or if no response occurred within 2000 ms of the appearance of the flash. Between each trial a blank screen was displayed for 2000 ms. Compatibility was determined by whether the direction (press) was compatible or incompatible with the finger stimulus movement (upward/downward). As detailed in Fig. 2, upward finger movements correspond to leftward movements across the screen for all three stimuli, while downward movements correspond to rightward movements. If Reaction Times (RTs) are faster when the stimulus was moving in the rightward (compatible) compared to leftward (incompatible) direction, this will result in positive compatibility effects (Incompatible–Compatible RTs). In this case, movements have been facilitated by representation of an upward moving finger movement and indicate the presence of imitative compatibility. However, if RTs are faster when the stimulus was moving in the leftward (incompatible) compared to rightward (compatible) direction, this will result in negative compatibility effects (Incompatible–Compatible RTs). Here, movements have been facilitated by the direction/position of the stimulus and indicate the presence of stimulus response compatibility effects. Therefore, positive compatibility effects represent imitative compatibility, whereas negative compatibility effects represent stimulus response compatibility (the left/right movement). The experimental conditions were stimulus condition (object, belief, hand)  compatibility (compatible, incompatible)  SOA (0, 120, 280 ms). For each block, the 6 compatibility  SOA conditions were presented 12 times, together with 12 baseline and 12 no go trials giving a total of 96 trials. These trials were split into 2 mini-blocks of 48 trials and a short break was given after each mini-block. A longer break was given after each block. For the entire experiment there were 216 experimental trials, 36 baseline trials and 36 no go trials, totalling 288 trials.

3. Results RTs for trials were removed if the participant made an incorrect response, did not respond, if the response was longer than 1000 ms or shorter than 150 ms or lay outside of 2.365 standard deviations of the participant’s mean RT (Van Selst & Jolicoeur, 1994). This resulted in a loss of 0.39% of trials. A repeated-measures ANOVA with factors of stimulus condition (object, belief, hand), SOA and compatibility was conducted on mean RTs (Table 1). There was a main effect of SOA (F(2,34) = 51.026, p < .0005, gp2 = .75), indicating that while RTs were significantly slower at 0 ms compared to 120 ms (t(17) = 11.202, p < .0005) and 280 ms (t(17) = 5.193, p < .0005), there was no significant difference between RTs at 120 ms and 280 ms (t(17) = 1.017, p = .323) (Bonferroni correction = 0.02). Importantly, there was a significant interaction between stimulus condition and compatibility (F(2,34) = 5.461, p = .009, gp2 = .24) (Fig. 3). Paired t-tests conducted between mean compatibility effects in each stimulus condition (averaged across SOA) indicated that compatibility effects were significantly smaller for the object ( 10.04 ms) than both the belief (6.20 ms) (t(17) = 3.024, p = .008; d = 0.6) and the hand condition (4.09 ms) (t(17) = 2.535, p = .021; d = 0.48), while there was no significant difference between the belief and hand conditions (t(17) = .416, p = .628; d = 0.06) (Bonferroni correction = 0.02). Simple main effects (paired t tests) comparing

Table 1 Reaction times for the different conditions. Condition Object

SOA (ms) 0 120 280

Belief

0 120 280

Hand

0 120 280

Reaction time (ms) [95% confidence intervals]

Compatibility effect (ms) [95% confidence intervals]

355.03 349.78 331.19 317.77 322.98 311.58

[334.7, 375.4] [328.9, 370.6] [308.2, 354.2] [298.6, 336.8] [307.5, 338.5] [294.4, 328.8]

5.25 [ 16.89, 6.38]

389.64 390.63 334.33 346.63 332.21 337.53

[353.0, 426.3] [346.3, 435.0] [301.8, 366.9] [305.7, 387.6] [305.1, 359.3] [303.5, 371.5]

0.99 [ 16.69, 18.67]

378.24 373.76 319.93 327.53 316.01 325.16

[343.7, 412.8] [345.4, 402.1] [299.2, 340.6] [292.4, 339.6] [292.4, 339.6] [301.5, 348.8]

13.48 [ 24.31,

2.64]

11.34 [ 24.86, 2.06]

12.3 [ 0.12, 24.72] 5.32 [ 9.8, 20.46] 4.48 [ 20.18, 11.22] 7.6 [ 11.1, 26.3] 9.15 [ 9.18, 27.48]

Note: Bold = compatible responses.

Table 2 Questionnaire results. Question

Mean score [95% CI]

One sample t test

1

0.12 [ 0.1, 0.3]

t = 1.46; p = .16

0.43 [ 0.1, 1.0]

t = 1.58; p = .13

7.59 [6.5, 8.6]

t = 15.4; p < .001

7.0 [5.8, 8.2]

t = 12.5; p < .001

7.43 [6.3, 8.6]

t = 13.4; p < .001

2

3

4

5

During the first half of the experiment, did you at any time think that the moving rectangle might represent a human finger movement? To what extent, if at all did the movement of the rectangle remind you of a human finger movement? Did you think of the rectangle as more of a ‘block representation’ of a moving finger, compared to when you had seen the rectangle for the first time? To what extent did you think of the moving rectangle as being a human finger compared to when you had seen the block for the first time? When you were told the rectangle represented a finger movement, how much did you believe that this was true?

Note: Questions 1–2 were asked following the object condition, questions 3–5 following the belief condition. For questions 1, 2, 4, 5, a score of 0 represents ‘‘not at all” and a score of 10 represents ‘‘very much”. For question 3, a score of 0 represents ‘‘disagree” and 10 represents ‘‘agree”.

compatible and incompatible reaction times (across SOA) revealed that a significant negative compatibility effect was present for the object condition (t(53) = 3.077, p = 0.003; d = 0.24), but that compatibility effects were not significant for the belief or hand condition (t(53) < 1.5, p > 0.14; d < 0.08) (Bonferroni correction = 0.02). These results show that while negative compatibility effects, representing stimulus response compatibility and Simon effects were observed for the object, positive (non significant) compatibility effects, representing imitation of a human action were observed for both the belief and hand condition. Finally, there was a significant interaction between stimulus condition and SOA (F(4,72) = 2.850, p = .049, gp2 = 0.13). Paired t tests comparing differences in RTs between the stimulus pairings at each SOA indicated that the difference between belief and hand (14.4 ms) was significantly smaller than the difference between the object and hand conditions (23.6 ms) for the 0 ms SOA only (t(35) = 3.044, p = 0.004, Bonferroni correction = 0.005; d = 0.71). Simple main effects comparisons at 0 ms SOA also revealed significantly faster responses for the object (352.4 ms) than both the

E. Gowen et al. / Cognition 146 (2016) 431–438

435

Fig. 3. Compatibility effects for each of the three stimulus conditions. Positive compatibility effects indicate imitative compatibility, whereas negative compatibility effects indicate stimulus response compatibility. Standard error bars are shown (Morey, 2008).

belief (390.14 ms; t(35) = 3.044, p = .004; d = 0.62) and the hand (376.0 ms; t(35) = 2.98, p = .005; d = 0.46), but no difference between the belief and hand condition (t(35) = 1.46, p = .15; d = 0.2) (Bonferroni correction = 0.02). There were no significant interactions between compatibility and SOA (F(2,34) = 0.37, p = .69, gp2 = .02) and condition, compatibility and SOA (F(4,68) = 1.02, p = .40, gp2 = .0.06). Prior to the belief instruction, participants did not associate the object with a finger movement (Table 2). However, following the belief instruction, scores were significantly different from 0, indicating that participants considered that the object represented a human movement. 3.1. Control conditions 3.1.1. Control Experiment 1: Order effects To test for possible practice/order effects when the object was viewed for a second time in the belief condition, compatibility effects for the first and last six trials of the object condition were calculated and averaged across participants. Compatibility effects were negative for both first (0 ms = 2.65, 120 ms = 10.7, 280 ms = 8.52) and last (0 ms = 8.42, 120 ms = 16.8, 280 ms = 17.7) six trials. This suggests that the positive compatibility found in the belief condition was not due to a practice effect. This was further confirmed by presenting two separate blocks of the object condition to a separate group of naive participants (n = 22, 15 females, mean age (±SD) = 21.7 ± 3.8 years). Compatibility effects for the second object block (mean across SOA = 4.59 ms, standard error = 2.89) were significantly lower than the belief condition in the main experiment (mean = 6.2 ms, standard error = 4.13; t = 2.8, p = 0.04). 3.1.2. Control Experiment 2: Weaker positive compatibility effects Although the negative compatibility effects for the object were significantly smaller than the positive compatibility effects found for the belief and hand conditions, these positive compatibility effects were non-significant. This is likely to be due to the require-

ment to pit orthogonal spatial compatibility and Simon effects against imitative compatibility. To test this we presented a new group of 19 participants (9 female, mean age (±SD): 22.1 (±3.01) with identical hand stimuli to previously described except that the hand was a left rather than right hand in thumb down position (Fig. 4a). Participants responded to the appearance of the yellow flash by pressing down their right index finger. Using this setup, when the finger moves downward (rightwards), the participants pressing response has both imitative and stimulus response (Simon effect) compatibility, whereas when the finger moves upward (leftward), the pressing response is incompatible for both imitation and the Simon effect. As imitative and stimulus response compatibility effects are combined, we would expect larger, significant positive compatibility effects. In line with this, a compatibility  SOA ANOVA revealed a main effect of compatibility (F(1,18) = 11.87, p = .0003, gp2 = 0.4) and an interaction between compatibility and SOA (F(2,36) = 4.08, p = .03, gp2 = 0.19) (Fig. 4b). Simple main effects (paired t tests) revealed significant positive compatibility effects for the 280 ms SOA only (t = 4.6, p < 0.001; d = 0.67, Bonferroni correction = 0.02). This compatibility effect was twice as large as the compatibility effects for the finger in our main experiment (Fig. 3), highlighting that imitative and stimulus response compatibility effects are likely to be combined, resulting in strong compatibility effects when in the same direction (Control Experiment 2), but weaker compatibility effects when in opposite directions (Main Experiment). 4. Discussion Our results provide the first behavioural demonstration that observation of a non-biological stimulus, believed to have human agency, leads to imitative compatibility as opposed to stimulus response compatibility. More generally, by separating imitative compatibility from stimulus response compatibility, this is the first evidence to show that priming caused by non-biological stimuli can be generated by imitative compatibility. When participants observed the object, prior to the belief manipulation, negative

436

E. Gowen et al. / Cognition 146 (2016) 431–438

(a)

Le

(b)

Right

Imitaon=compable

Imitaon=incompable

Simon effect=incompable

Simon effect=compable

Orthogonal =compable

Orthogonal =incompable

Response: press (down) using right hand Fig. 4. (a) Left hand in thumb down position, showing co-occurrence of imitation and Simon effect compatibility when responding with right finger press. (b) Compatibility effects across the three different SOA’s, showing large compatibility effects at the 280 ms SOA. Standard error bars are shown.

compatibility effects were present due to stimulus response compatibility produced by the direction and position of the stimulus. However, following the belief manipulation, compatibility effects became numerically positive and of a similar magnitude to the subsequently viewed hand stimulus. Importantly, compatibility effects for both the belief and hand condition were numerically positive and significantly different to the object condition. If the belief effect was simply caused by increased attention to the stimulus enhancing bottom-up stimulus response compatibility effects, the belief condition should have produced larger negative compatibility effects, similar to the object condition. Consequently, our results indicate that belief that a non-biological stimulus is human-generated causes action representation, validating previous belief studies unable to conclusively rule out the influence of attention (Liepelt & Brass, 2010; Shen et al., 2011; Stanley et al., 2007). More generally, our hand condition adds to accumulating data that imitative compatibility cannot be explained by stimulus response compatibility effects (Catmur & Heyes, 2011; Jimenez et al., 2012). While significant negative compatibility effects were observed for the object, it should be noted that the positive compatibility effects for the belief and hand condition were non-significant. However, it is unsurprising that we obtained smaller imitative compatibility effects since imitative compatibility was pitted against a number of stronger stimulus response compatibility effects (Longo & Bertenthal, 2009; Jimenez et al., 2012). Indeed, when the direction of stimulus response compatibility and imitative compatibility were complementary as in our Control Experiment 2, positive compatibility effects were significant and more than twice as large as the finger stimulus in the main experiment. As with previous work, this highlights that stimulus response compatibility can mask imitative compatibility when pitted against one another (Longo & Bertenthal, 2009; Jimenez et al., 2012) and that significant positive compatibility effects are more likely when imitative compatibility is confounded with stimulus response compatibility. Furthermore, the overall pattern of our results supports the theory that belief causes action representation as opposed to increased attention to stimulus properties. Firstly, the direction of the compatibility effects was positive for both the belief and hand condition and differed significantly from the negative compatibility effects for the object. In addition, reaction times were similarly slower in the belief and hand conditions compared to the object conditions, raising the possibility that although they were not instructed to do so participants were mentally rotating the belief and hand stimuli to match their own posture. Previous work indicates that making explicit 90° mental hand rotations produces longer reaction times (Parsons, 1994) than the difference between the belief and hand compared to object conditions. However, these smaller differences may result from the rotation not being necessary for the task, as well as viewing a consistent hand

rotation on every trial. As the belief condition closely follows the pattern of the hand condition, this supports the notion that the action is represented in the belief condition. Secondly, the finding of similar compatibility effects for the first and second half of the trials in the object condition argues against practice effects or decreased attention to spatial/directional (stimulus response compatibility) properties leading to a reduction in stimulus response compatibility effects in the belief condition. This conclusion is further supported by significantly different compatibility effects between the belief condition and the second object block of Control Experiment 1. Previous belief experiments that have not separated stimulus response compatibility and imitative compatibility (Liepelt & Brass, 2010; Shen et al., 2011; Stanley et al., 2007) also indicate that this possibility is unlikely. These studies show increased compatibility effects for the belief condition, whereas a reduction in the influence of stimulus response compatibility effects would have resulted in smaller compatibility effects. Our results advance existing models detailing how priming is influenced by top-down factors (Gowen & Poliakoff, 2012; Heyes, 2011; Wang & Hamilton, 2012). These models share the idea that priming occurs along a visuomotor route, involving brain areas such as the Superior Temporal Sulcus (STS), parietal and premotor areas such as the IFG. The visuomotor route transforms visual input into a motor response and priming produced by stimulus response compatibility and imitative compatibility are dissociated (Heyes, 2011), with the latter involving the MNS. Priming within the visuomotor route is modulated by top-down factors such as attention, prior knowledge and social cognitive processes, which can exert influence at the early sensory input stage or at the later motor, output stage (Heyes, 2011). There is increasing evidence that one brain area, the Medial Prefrontal Cortex (MPFC) is likely to modulate priming (Cross, Torrisi, Losin, & Iacoboni, 2013; Spengler, von Cramon, & Brass, 2009; Wang, Ramsey, & Hamilton, 2011, 2010), including during belief manipulations (Stanley, Gowen, & Miall, 2010). As priming changed from stimulus response compatibility during the object condition to imitative compatibility for the belief condition, mechanisms modulating the level of motor output are unlikely to be responsible for the effect of belief on priming. Instead, it is more likely that belief causes the visual input to be processed by the MNS pathway, potentially through connections between the MPFC and STS, similar to a gating mechanism (Liepelt & Brass, 2010). In contrast, when belief is drawn to inanimacy (Longo & Bertenthal, 2009) priming may be channelled along non-mirror, stimulus response compatibility routes only. Our behavioural results complement fMRI studies showing that non-biological stimuli activate brain areas associated with the MNS (Cross, Hamilton, Kraemer, Kelley, & Grafton, 2009; Engel, Burke, Fiehler, Bien, & Rosler, 2008; Gazzola, Rizzolatti, Wicker, & Keysers, 2007; Press et al., 2012). However, these studies are

E. Gowen et al. / Cognition 146 (2016) 431–438

unable to identify whether non-biological movement activates the MNS as they have either not separated stimulus response compatibility from imitative compatibility or do not use both observation and execution conditions to identify the MNS. Furthermore, some have used training paradigms to associate non-biological stimuli with actions, whereas our study shows that action representation of non-biological movement can occur without prior training. Combining our task with fMRI could be used to separate brain activity relating to MNS and stimulus response compatibility. As imitation can relate to social affiliation (Chartrand & Lakin, 2013), it is possible that imitation could be used to measure the success of human–robot interaction, assessing the effect of modifications in robot design aimed at optimising these interactions (Hofree, Ruvolo, Bartlett, & Winkielman, 2014). Our findings indicate that providing stimulus response compatibility factors are removed, measuring the level of imitative compatibility would be a useful and possible marker for whether a robot elicits action representation. In addition, instructions and prior knowledge may be a useful strategy for encouraging representation of robot actions and facilitating human acceptance of robots (Chaminade & Cheng, 2009). Supporting this, visuomotor priming has been found during observation of a robot movement only if participants are informed that the robot is observing their actions and consequently is ‘‘engaged” in the interaction (Shen et al., 2011). Importantly, the present results suggest that this belief effect is likely to be due to representation of the robot as a human-like agent as opposed to increased attention to the robot. In our experiment, as the object moved using a human trajectory, it is unclear whether belief allows access of this movement to the MNS or whether the object is processed as human. Although this question requires testing using a non-human trajectory, the results of Shen et al. (2011) support the latter possibility as their robot stimulus moved with a nonhuman trajectory. Together, these results suggest that a robot can trigger human associated imitative behaviours that could lead to higher level social representations (Chaminade & Cheng, 2009). These implications would also apply to the situation where a robot triggers implicit belief that they are human-like. A crucial next step is to assess whether those human–robot interactions that produce imitation lead to better human–robot acceptance. Furthermore, as our object stimulus moved with a human trajectory, it will be important to understand whether the belief effect is apparent for non-biological stimuli moving with a non-human trajectory. In summary, the current study has demonstrated that abstract, non-biological stimuli can elicit imitative compatibility effects equivalent to those produced for a human movement provided that the non-biological movement is believed to be humangenerated. These findings imply that non-human movement can be processed by the MNS, as if it were a human movement. This has implications for understanding the control of imitation, as well as developing successful human–robot interaction. Acknowledgements The authors would like to thank Shannon Atkinson and Robyn Dowlen for their help with data collection. References Andrade, A. O., Pereira, A. A., Walter, S., Almeida, R., Loureiro, R., Compagna, D., et al. (2014). Bridging the gap between robotic technology and health care. Biomedical Signal Processing and Control, 10, 65–78. Brass, M., Bekkering, H., & Prinz, W. (2001). Movement observation affects movement execution in a simple response task. Acta Psychologica (Amst), 106 (1–2), 3–22. Buccino, G., Binkofski, F., Fink, G. R., Fadiga, L., Fogassi, L., Gallese, V., et al. (2001). Action observation activates premotor and parietal areas in a somatotopic manner: An fMRI study. European Journal of Neuroscience, 13(2), 400–404.

437

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), 409–421. Catmur, C., Walsh, V., & Heyes, C. (2009). Associative sequence learning: The role of experience in the development of imitation and the mirror system. Philosophical Transactions of the Royal Society B – Biological Sciences, 364(1528), 2369–2380. Chaminade, T., & Cheng, G. (2009). Social cognitive neuroscience and humanoid robotics. Journal of Physiology Paris, 103(3–5), 286–295. Chartrand, T. L., & Lakin, J. L. (2013). The antecedents and consequences of human behavioral mimicry. Annual Review of Psychology, 64(64), 285–308. Cho, Y. S., & Proctor, R. W. (2003). Stimulus and response representations underlying orthogonal stimulus–response compatibility effects. Psychonomic Bulletin & Review, 10(1), 45–73. Cross, E. S., Hamilton, A. F., Kraemer, D. J., Kelley, W. M., & Grafton, S. T. (2009). Dissociable substrates for body motion and physical experience in the human action observation network. European Journal of Neuroscience, 30(7), 1383–1392. Cross, K. A., Torrisi, S., Losin, E. A. R., & Iacoboni, M. (2013). Controlling automatic imitative tendencies: Interactions between mirror neuron and cognitive control systems. Neuroimage, 83, 493–504. Dautenhahn, K. (2007). Socially intelligent robots: Dimensions of human–robot interaction. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 362(1480), 679–704. Engel, A., Burke, M., Fiehler, K., Bien, S., & Rosler, F. (2008). Motor learning affects visual movement perception. European Journal of Neuroscience, 27(9), 2294–2302. Gazzola, V., & Keysers, C. (2009). The observation and execution of actions share motor and somatosensory voxels in all tested subjects: Single-subject analyses of unsmoothed fMRI data. Cerebral Cortex, 19(6), 1239–1255. Gazzola, V., Rizzolatti, G., Wicker, B., & Keysers, C. (2007). The anthropomorphic brain: The mirror neuron system responds to human and robotic actions. Neuroimage, 35(4), 1674–1684. Gowen, E., Bradshaw, C., Galpin, A., Lawrence, A., & Poliakoff, E. (2010). Exploring visuomotor priming following biological and non-biological stimuli. Brain and Cognition, 74(3), 288–297. Gowen, E., & Poliakoff, E. (2012). How does visuomotor priming differ for biological and non-biological stimuli? A review of the evidence. Psychological Research, 76 (4), 407–420. Heyes, C. (2011). Automatic imitation. Psychological Bulletin, 137(3), 463–483. Hofree, G., Ruvolo, P., Bartlett, M. S., & Winkielman, P. (2014). Bridging the mechanical and the human mind: Spontaneous mimicry of a physically present android. PLoS ONE, 9(7), e99934. Jansson, E., Wilson, A. D., Williams, J. H., & Mon-Williams, M. (2007). Methodological problems undermine tests of the ideo-motor conjecture. Experimental Brain Research, 182(4), 549–558. Jimenez, L., Recio, S., Mendez, A., Lorda, M. J., Permuy, B., & Mendez, C. (2012). Automatic imitation and spatial compatibility in a key-pressing task. Acta Psychologica, 141(1), 96–103. Kilner, J. M., Neal, A., Weiskopf, N., Friston, K. J., & Frith, C. D. (2009). Evidence of mirror neurons in human inferior frontal gyrus. Journal of Neuroscience, 29(32), 10153–10159. Liepelt, R., & Brass, M. (2010). Top-down modulation of motor priming by belief about animacy. Experimental Psychology, 57(3), 221–227. Longo, M. R., & Bertenthal, B. I. (2009). Attention modulates the specificity of automatic imitation to human actors. Experimental Brain Research, 192(4), 739–744. Morey, R. D. (2008). Confidence intervals from normalized data: A correction to Cousineau (2005). Tutorial in Quantitative Methods for Psychology, 4(2), 61–64. Parsons, L. M. (1994). Temporal and kinematic properties of motor behavior reflected in mentally simulated action. Journal of Experimental Psychology: Human Perception and Performance, 20(4), 709. Press, C. (2011). Action observation and robotic agents: Learning and anthropomorphism. Neuroscience and Biobehavioral Reviews, 35(6), 1410–1418. Press, C., Bird, G., Flach, R., & Heyes, C. (2005). Robotic movement elicits automatic imitation. Brain Research. Cognitive Brain Research, 25(3), 632–640. Press, C., Catmur, C., Cook, R., Widmann, H., Heyes, C., & Bird, G. (2012). FMRI evidence of ’mirror’ responses to geometric shapes. PLoS ONE, 7(12). Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron system. Annual Review of Neuroscience, 27, 169–192. Shen, Q. M., Kose-Bagci, H., Saunders, J., & Dautenhahn, K. (2011). The impact of participants’ beliefs on motor interference and motor coordination in human– humanoid interactions. IEEE Transactions on Autonomous Mental Development, 3 (1), 6–16. Simon, J. R., & Proctor, R. W. (1990). The effects of an irrelevant directional cue on human information processing. Stimulus–response compatibility: An integrated perspective. In R. W. Proctor, T. G. Reeve (Eds.), Advances in psychology (Vol. 65, pp. 31–86). Amsterdam, North-Holland. Spengler, S., von Cramon, D. Y., & Brass, M. (2009). Control of shared representations relies on key processes involved in mental state attribution. Human Brain Mapping, 30(11), 3704–3718. Spengler, S., von Cramon, D. Y., & Brass, M. (2010). Resisting motor mimicry: Control of imitation involves processes central to social cognition in patients with frontal and temporo-parietal lesions. Social Neuroscience, 5(4), 401–416. Stanley, J., Gowen, E., & 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(4), 915–926.

438

E. Gowen et al. / Cognition 146 (2016) 431–438

Stanley, J., Gowen, E., & Miall, R. C. (2010). How instructions modify perception: An fMRI study investigating brain areas involved in attributing human agency. Neuroimage, 52(1), 389–400. Sturmer, 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), 1746–1759. Tapus, A., Mataric, M. J., & Scassellati, B. (2007). Socially assistive robotics – The grand challenges in helping humans through social interaction. IEEE Robotics & Automation Magazine, 14(1), 35–42. Van Overwalle, F., & Baetens, K. (2009). Understanding others’ actions and goals by mirror and mentalizing systems: A meta-analysis. Neuroimage, 48(3), 564–584.

Van Selst, M., & Jolicoeur, P. (1994). A solution to the effect of sample size on outlier elimination. The Quarterly Journal of Experimental Psychology Section A: Human Experimental Psychology, 47A(3), 631–650. Wang, Y., & Hamilton, A. F. D. (2012). Social top-down response modulation (STORM): A model of the control of mimicry in social interaction. Frontiers in Human Neuroscience, 6. Wang, Y., Ramsey, R., & Hamilton, A. F. D. (2011). The control of mimicry by eye contact is mediated by medial prefrontal cortex. Journal of Neuroscience, 31(33), 12001–12010. Weeks, D. J., & Proctor, R. W. (1990). Salient-features coding in the translation between orthogonal stimulus and response dimensions. Journal of Experimental Psychology-General, 119(4), 355–366.