Mental rotation (MR) is the ability to mentally rotate two- or three-dimensional objects. Since the original study by Shepard and Metzler (1971), a classical MR paradigm has consisted of indicating whether two stimuli (multiarmed abstract cubes figures) are identical or mirror images. Response times (RTs) are usually impacted by the angular disparity between the two figures. Accordingly, it is believed that participants mentally rotate one stimulus into alignment with the other before giving their answer. Later, several studies introduced the distinction of MR of body parts or MR of the whole body (Parsons, 1987, 1994; Sekiyama, 1982). This kind of MR consists of presenting a body part in different orientations (e.g., a hand) followed by a judgment of its laterality. It is commonly believed that participants tend to imagine their own body part moving toward the stimulus, and therefore, that MR of body parts differs from that of other stimuli, such as characters, numbers, and abstract shapes (Parsons, 1987, 1994). In particular, RT is known to be modulated by the biomechanical limits that constrain the imagined movement of the hand toward the stimuli. In other words, rotating right hands in a clockwise (CW) orientation is biomechanically more difficult than counterclockwise (CCW) rotations, whereas the opposite is true for left hands (Parsons, 1994).

Overall, at least two different strategies are possible. In the egocentric (internal) perspective, participants imagine themselves physically manipulating the stimulus or moving toward it. This strategy is usually used during MR of hand or body parts or when the object is to be mentally grasped and moved (e.g., a screwdriver, hammer, or mug). In the allocentric (external) perspective, participants visualize the consequences of using external forces to make the stimulus move. This strategy may have been called upon for MR of abstract shapes, as this kind of stimulus does not prime moving the hands, and thus does not involve motor processes. Kosslyn, Digirolamo, Thompson, and Alpert (1998) described the neural mechanisms of MR by comparing the MR of hands to that of cubes. MR of cubes activated higher visual areas, whereas MR of hands elicited activation of motor areas involved in movement preparation (e.g., left primary motor area, superior and inferior parietal lobules, primary visual cortex, insula, and left frontal areas). In addition, behavioral studies have reported the use of an egocentric strategy for body part MR (Petit, Pegna, Mayer, & Hauert, 2003; Thayer, Johnson, Corballis, & Hamm, 2001). The MR of body parts is sensitive to proprioceptive information, thus leading to longer RTs when the participants were requested to identify the laterality of stimuli presented in anatomically impossible positions. Furthermore, the time spent for mentally rotating one’s limb in order to find its handedness was shorter when the hand was kept in a congruent position than when it was presented in a more awkward, incompatible position (Conson, Pistoia, Sarà, Grossi, & Trojano, 2010; Conson, Mazzarella, & Trojano, 2011; Funk, Brugger, & Wilkening, 2005; Ionta, Fourkas, Fiorio, & Aglioti, 2007; Sirigu & Duhamel, 2001).

The use of an egocentric (motor) strategy is the reason why the hand laterality task is extensively used in participants with motor coordination disorders and/or impaired motor planning or execution processes. They are usually slower than control participants in hand laterality judgment tasks (de Lange, Roelofs, & Toni, 2008; Helmich, de Lange, Bloem, & Toni, 2007; Nico, Daprati, Rigal, Parsons, & Sirigu, 2004; Steenbergen, van Nimwegen, & Crajé, 2007; Williams, Thomas, Maruff, & Wilson, 2008; Wilson et al., 2004). Despite that, same–different hand MR has been used for persons with Parkinson Disease (Amick, Schendan, Ganis, & Cronin-Golomb, 2006). Unfortunately, items requiring visual allocentric strategies are not appropriate for studying motor impairment, and thus caution should be taken before using them with participants with motor disorders.

Even though an egocentric strategy is usually used during the hand laterality task, MR of body parts can also be performed from the allocentric perspective, especially when stimuli are presented in specific orientations. For instance, hand images displayed in upside-down orientations (near 180º) are automatically oriented within an allocentric frame, whereas hands presented in an upright position with the fingers up (near 0º) are processed from the egocentric perspective (Brady, Maguinness, & Ni Choisdealbha, 2011; Ni Choisdealbha, Brady, & Maguinness, 2011).

Manipulating stimulus features can also lead to shifting from the ego- to the allocentric strategy. For example, the allocentric strategy may be used when a hand stimulus is embedded within the human silhouette. Accordingly, Conson, Mazzarella, Donnarumma, and Trojano (2012) recently compared RTs of hand MR embedded in a human silhouette to those of hands embedded in a nonhuman silhouette. Participants were slower in the human silhouette condition. Thus, observing the human silhouette interfered with egocentric processing and impaired performance. In a second experiment, new participants were explicitly asked to judge the laterality of hands embedded in a human silhouette from their own egocentric perspective or from the silhouette’s allocentric perspective. Participants from the egocentric group exhibited faster RTs than did those from the allocentric group. Conson et al. (2012) concluded that changing stimulus features could induce a shift from one strategy to another. In the case of concurrent activation between ego- and allocentric frames, performance was impaired, due to conflict between motor and visual processes.

As in the Conson et al. (2012) study, Ionta, Perruchoud, Draganski, and Blanke (2012) compared MR of hands only (one hand on the screen) and of hands on a body (a whole human body with two hands, but one target hand depicted in black). They specifically studied the effects of hand postural change on MR of these two types of stimuli, looking at the effect of actual hand position as a way of inferring the strategy used by the participants. Although hand posture effects are not systematically observed, they remain present in the MR of stimuli representing the body segment whose posture is manipulated (e.g., for hands but not for feet; Ionta & Blanke, 2009; Ionta et al., 2007). In Ionta et al.’s (2012) study, the postural effect was observed in the hands-only condition, but not in the hands-on-body stimuli. Ionta et al. (2012) concluded that the hands-on-body stimuli elicited an allocentric strategy, whereas the hand-only stimuli relied more on kinesthetic and somatosensory representations, and thus elicited an egocentric strategy.

All of the above-mentioned studies on hand or body MR consisted of laterality judgment tasks of one stimulus. According to Zacks, Mires, Tversky, and Hazeltine (2002), a left–right judgment for a single picture of a human body spontaneously elicits an egocentric strategy. On the other hand, asking people to perform a same–different judgment for two pictures of human bodies is mainly an allocentric transformation. Steggemann, Engbert, and Weigelt (2011) recently replicated this finding by demonstrating that experts in rotational movements of sporting activities (gymnasts or trampolinists) performed better than nonexperts in a left–right judgment task (image of a human body with the left or right arm extended), but not in a same–different judgment evoking an allocentric strategy. However, it is difficult to control variables related to individual differences, such as life experience and previous exposure to motor and visual expertise, in order to evidence the use of an egocentric or allocentric strategy. Therefore, examining the effects of hand postural changes on a laterality judgment task relative to a same–different task will be crucial to clarifying previous findings (e.g., Ionta et al., 2012; Steggemann et al., 2011; Zacks et al., 2002) and bringing new knowledge on how different tasks recruit different mental imagery strategies. With this aim, in the present study we requested the same participants to perform a laterality hand MR task and a same–different judgment of two hands while keeping their own hands in a compatible or incompatible position. To the best of our knowledge, this is the first time that hand posture effects have been investigated for one-hand or two-hand MR tasks (Steggemann et al., 2011, and Zacks et al., 2002, used images of complete human bodies). We hypothesized that hand position would influence performance on the hand laterality judgment only. Furthermore, biomechanical effects were expected only for the laterality test, whereas the same–different task would be independent of the biomechanical features of the stimuli but would be impacted by the angular disparity between the two figures.

Method

Participants

A group of 30 healthy participants (15 females, 15 males) 18–30 years of age (M = 23 years, SD = 4) were enrolled in the experiment after giving their informed consent. All participants were right-handed, according to the revised Edinburgh Handedness Inventory (Dragovic, 2004). The local ethics committee approved the experiment, which was in accordance with the Declaration of Helsinki (1964).

Stimuli

We presented two types of stimuli: naturalistic pictures of hands and alphanumeric characters (the letter “R” and number “2”), which were used as experimental and control stimuli, respectively. Both types of stimuli were presented in two kinds of items: one stimulus at a time on the computer screen (laterality judgment for hands and normal–mirror judgment for alphanumeric characters), or two stimuli (two hands, two “R”s, or two “2”s) presented simultaneously on the screen (same–different judgment). Finally, the hand stimuli were either simple (dorsum or palm view from four orientations: 0º, 90º, 180º, or 270º) or complex (palm from finger view, dorsum from wrist view, palm from wrist view, or dorsum from finger view). Our simple hand stimuli were equally rotated either CW or CCW.

Procedure

Participants were tested individually in a quiet room, which was slightly darkened during the session, so as to prevent reflections on the computer screen. Participants sat comfortably at a distance of about 50 cm from the computer screen. For both the hand MR and alphanumeric MR items (one item presented on screen), participants were asked to verbally answer as quickly and as accurately as possible whether the hand was a right or a left hand or whether the alphanumeric character was normal or mirror. Under the two-stimulus condition, participants should decide whether the two hands or the two alphanumeric character items were the same or different. Stimulus presentation was controlled with E-Prime 2 (Psychology Software Tool Inc., Pittsburgh USA). RTs were recorded from a microphone positioned in front of the participants. Thus, RTs were stopped as soon as a response was orally given. The experimenter manually collected response accuracy. The instructions were standardized, and participants could read them out on their own. After one practice block for familiarization, the experimental session consisted of six blocks. Each block contained 32 items of only one type of item (one simple or complex hand; two simple or complex hands; one alphanumeric character; or two alphanumeric characters). All of the hand MR items contained the same number of left and right hands and the same number of CW and CCW rotations (for the simple-stimulus blocks only). At the beginning of each trial, a cross was displayed for 1,000 ms on the screen in order to help participants focus their attention. Then the stimuli appeared and remained visible until participants gave a response. In the middle of each block, a short break allowed the participants to relax their eyes, stretch out their arms, and change hand posture.

In the compatible condition, participants were requested to place their hands on their knees, whereas under the incompatible condition, they held their hands behind their back. Under both conditions, the hands were hidden to participants, who were thus only informed about hand position through tactile and proprioceptive information. The order of the conditions and blocks was counterbalanced across participants.

Data analysis

Following previous studies (Ionta et al., 2012; Parsons, 1987, 1994), we excluded trials with incorrect answers and with RTs shorter than 500 ms and longer than 3,500 ms. Accordingly, the RT was the time calculated between the stimulus onset and the participant’s verbal response. To test the real-hand postural effects on hand laterality, as compared to same–different judgments, we used a four-way repeated measures ANOVA with Hand Posture (compatible, incompatible), Item (one stimulus only, two stimuli presented simultaneously), Difficulty (simple, complex), and Sex (female, male) as the main factors. In order to test the biomechanical-constraint effects on hand laterality relative to same–different judgment, we gathered RTs from CW-rotated (90º) and CCW-rotated (270º) items. RTs from 180º-rotated items were not collected, in accordance with recent studies (Brady et al., 2011; Ni Choisdealbha et al., 2011), that have shown that stimuli presented in upside-down orientations (near 180º) induce an allocentric frame. We subdivided the hand laterality items into biomechanically easy (CW for left hands and CCW for right hands) and biomechanically difficult (CW for right hands and CCW for left hands). The same–different items were also subdivided into biomechanically easy (both stimuli without any biomechanical constraint) and difficult (both stimuli with biomechanical constraints). We used a two-way repeated ANOVA with Item (one stimulus, two stimuli) and Biomechanical Constraint (easy, difficult) as the main factors. An additional one-way repeated ANOVA was carried out for same–different items, to determine whether RTs would increase with increasing angular disparity. Finally, we used a three-way repeated measures ANOVA for the control items (i.e., alphanumeric characters) with Hand Posture (compatible, incompatible), Item (one stimulus only, two stimuli presented simultaneously), and Sex (female, male) as factors. Post-hoc comparisons were carried out using Bonferroni corrections, with the significance threshold set at p < .05.

Results

Effects of actual hand posture

We first observed main effects of posture [F(1, 196) = 4.28, p = .03, η 2 = .02], item [F(1, 196) = 142.49, p < .001, η 2 = .44], task difficulty [F(1, 196) = 9.67, p = .002, η 2 = .05], and sex [F(1, 196) = 7.8, p = .01; η 2 = .22]. We also observed a significant two-way interaction between posture and item [F(1, 196) = 8.45, p = .004, η 2 = .04]. Post-hoc tests with Bonferroni corrections did not show any RT difference between compatible (M = 2,106 ms, SD = 709) and incompatible (M = 2,062 ms, SD = 695) body postures when the MR task involved two hands (p > .05). Conversely, RTs were significantly shortened when the MR task was performed with a compatible body posture and involved only one hand (p < .001). Accordingly, the mean RTs were 1,324 ms (SD = 382) and 1,586 ms (SD = 574) for compatible and incompatible hand postures, respectively. Figure 1 shows the effects of hand posture on RT differences for both one-hand and two-hand items (i.e., laterality and same–different judgments).

Fig. 1
figure 1

Effects of compatible versus incompatible hand position on response times (RTs, in milliseconds) for hand laterality judgments and same–different judgments. A posture effect was only observed for hand laterality judgments. Accordingly, RTs were significantly lower when the hands were held in a compatible position. *** p < .001; NS, nonsignificant

Full size image

No three- or four-way interaction reached the statistical threshold for significance.

Effects of biomechanical constraints

We first observed a main effect of item [F(1, 87) = 21.15, p < .001, η 2 = .24]. We also observed a significant two-way interaction between item and biomechanical constraints [F(1, 87) = 4.9, p = .02, η 2 = .05]. Post-hoc investigation with Bonferroni corrections did not show any RT difference between the biomechanically easy (M = 1,716 ms, SD = 520) and difficult (M = 1,667 ms, SD = 715) conditions for same–different items (p > .05). However, RTs were significantly shortened when the hand laterality judgment involved biomechanically easy stimuli (p = .001). Accordingly, the respective mean RTs were 1,261 ms (SD = 459) and 1,507 ms (SD = 610) for biomechanically easy and difficult items. Figure 2 illustrates the biomechanical effects on hand laterality, relative to same–different hand judgment.

Fig. 2
figure 2

Biomechanical effects on response times (RTs, in milliseconds) for hand laterality judgments versus same–different judgments. A biomechanical-constraint effect was only observed for hand laterality judgments. Accordingly, RTs were significantly lower when stimuli were biomechanically easy (i.e., clockwise rotation for the left hand and counterclockwise rotation for the right hand). *** p < .001; NS, nonsignificant

Full size image

Effects of angular disparity on two-hand items

We observed a significant effect of angle on RTs in the two-hand items [F(2, 58) = 18.22, p < .001, η 2 = 0.61]. Post-hoc tests with Bonferroni corrections showed significant differences between 0º and 90º (p = .001), 90º and 180º (p < .001), and 0º and 180º (p < .001). Accordingly, the mean RTs were 1,682 ms (SD = 537), 1,871 ms (SD = 565), and 2,179 ms (SD = 799) for 0º, 90º, and 180º, respectively. Figure 3 shows the increased RTs as a function of angular disparity.

Fig. 3
figure 3

Effect of angular disparity on same–different items. Response times (RTs, in milliseconds) progressively increased as a function of the angular disparity between two hand stimuli. Accordingly, participants were faster for 0º (1,682 ms) than for 90º items (1,871 ms) and 180º items (2,179 ms). *** p < .001

Full size image

Alphanumeric MR task (control task)

General main effects of item [F(1, 84) = 59.65, p < .01, η 2 = .42] and sex [F(1, 28) = 3.85, p = .05, η 2 = .12] were observed. Participants were faster when processing one alphanumeric character (M = 1,268 ms, SD = 332) rather than two stimuli on the screen (M = 1,675 ms, SD = 543). Furthermore, men were faster (M = 1,344 ms, SD = 361) than women (M = 1,599 ms, SD = 571). The latter result is in line with several previous results (e.g., Peters, 2005) and will not be discussed later on in this article.

Discussion

The main objective of this study was to determine whether hand laterality judgment and same–different hand MR share the same cognitive strategies (egocentric or allocentric strategies). First, we studied the influence of actual hand posture on RTs in two types of MR items. The data showed no effect of hand posture when participants had to decide whether two hand stimuli were the same or different. Thus, the anatomical joint constraints of the hand did not influence MR of a target hand stimulus in order to compare it to another hand stimulus. This suggests that an allocentric visual strategy was certainly used for same–different hand judgment. Conversely, in line with Ionta et al.’s (2012) study, hand posture affected the MR of a single hand, as demonstrated by a performance decrease when the participants kept their hands in an incompatible position. This suggests that anatomical constraints interfered with single-hand MR, and therefore that the participants referred to an egocentric strategy. Up to now, no data have described the effects of hand posture on a same–different hand MR task. The present results show that same–different hand judgment is less dependent on kinesthetic information and anatomical constraints regarding body posture than is one-hand laterality judgment.

Second, we studied how RTs are modulated by the biomechanical limits that constrain the imagined movement of the hand toward the stimuli. Our data revealed a significant difference in RTs between biomechanically easy and difficult items, but only for laterality judgment. The lack of a difference in the same–different items confirms that an allocentric strategy may be preferred for such items. To further validate our hypothesis, we looked at the relationship between RTs and angular disparity in same–different items. These data showed a significant increase of RTs as a function of the angular disparity between the two stimuli. This finding supports the idea that participants mentally rotated one hand stimulus into alignment with the other before giving their answer. We may conclude that, as with the abstract objects used by Shepard and Metzler (1971), an allocentric strategy is used when we compare two hand stimuli that are simultaneously presented.

Our results replicate and further complete Steggemann et al.’s (2011) findings. Their study recently showed that an egocentric and an allocentric strategy, respectively, were used for laterality judgment items and same–different judgment items. Our findings are in line with Steggemann et al.’s results regarding the use of egocentric or allocentric strategies for laterality versus same–different judgment items. However, some methodological differences in the two experimental paradigms can account for some subtle nuances. First, Steggemann et al. compared the RTs of experts in rotational movements (e.g., gymnasts) with those of nonexpert athletes (e.g., participants in handball, soccer, track and field, or swimming). The researchers assumed that the egocentric strategy would lead motor experts to faster RTs during hand laterality judgments. However, their results were not very conclusive. Accordingly, experts and nonexperts performed equally, except for stimuli presented upside down (at 135º and 180º), on which experts were faster. This result is at odds with previous studies (Brady et al., 2011; Ni Choisdealbha et al., 2011) that have stated that stimuli presented in upside-down orientations (near 180º) are automatically visualized in an allocentric frame. We assume that studying both postural interference and biomechanical constraint effects is a more suitable way of proving the use of an egocentric strategy than is comparing the performance of experts and nonexperts. Especially, studies dealing with MR performance in athletes have already shown discrepant results (see Jansen, Lehmann, & Van Doren, 2012; Jola & Mast, 2005; Ozel, Larue, & Molinaro, 2004). The second methodological difference is related to the type of stimuli used. Accordingly, Steggemann et al. used whole bodies with an extended left or right arm as their stimuli, whereas we used pictures of hands. We assume that these different stimuli would engender different cognitive processes, notably because a whole body with an extended arm is more salient than a hand stimulus. In other words, a 90º-extended arm is considered a salient visual cue that is much easier to encode than is a hand stimulus, in which only the thumb is slightly salient. A stimulus with a salient visual cue may thus require more of an allocentric than an egocentric strategy.

Hand posture did not influence MR of alphanumeric characters, whether stimuli were presented alone or in a same–different comparison task. This expected result supports the finding that such stimuli engender visual strategies and are not affected by anatomical joint constraints. Furthermore, the participants responded faster during MR of one stimulus than during the comparison of two stimuli. This result shows that comparing two stimuli takes more time because participants mentally rotate a target stimulus to align it with the other one before giving their answer. In other words, participants encode both stimuli and make visual saccades from one to the other before making the final decision. Several studies had already demonstrated such visual strategies using an eyetracker (e.g., Khooshabeh & Hegarty, 2010). On the other hand, when dealing with one alphanumeric character, RTs are slower because participants encode and mentally rotate the stimulus in order to recognize whether it is a normal or a mirror image. In this case, the final comparison process relies on comparing the mentally rotated image of the stimulus to the internal, memorized and learned representation of the given alphanumeric character. This suggests that participants with impaired internal representations of letters and numbers (e.g., dyslexic people) would have more difficulties in the MR of one alphanumeric character than in a same–different judgment. Accordingly, Lachmann, Schumacher and van Leeuwen (2009) stated that dyslexics do not have a problem with MR per se, but rather with ultimately deciding whether a letter is presented in a normal or mirrored position.

This same analysis may be used to understand hand MR. A visual strategy is used to compare two stimuli if they are presented simultaneously. On the other hand, during a hand laterality task, the stimulus is usually compared to an egocentric internal representation of one’s own hand. Regarding the use of hand MR in participants with motor coordination disorders and/or impaired motor planning or execution processes (Amick et al., 2006; de Lange et al., 2008; Helmich et al., 2007; Nico et al., 2004; Steenbergen et al., 2007; Williams et al., 2008; Wilson et al., 2004), we assume that three possible explanations for this impaired performance could be put to the forth. First, participants might try to imagine their own hand rotation into alignment with the stimulus before comparing and deciding whether it is a left or a right hand. In that case, they would show slower RTs due to their impaired motor imagery ability (the ability to imagine an action without any concomitant body movement; see Guillot & Collet, 2008). Second, they might instead visually turn the stimulus (allocentric strategy) and finally compare its position with their own hand. In that case, the impairment would be due to their difficulty with judging laterality, which implies recalling their internal representation of a left or a right hand. Third, participants might use a visual strategy during all subprocesses. Accordingly, Vannuscorps, Pillon, and Andres (2012) recently provided evidence that only an allocentric perspective is used in the case of congenital absence of the upper limbs, and furthermore that it is influenced by biomechanical constraints. Thus, during a hand laterality judgment task, participants may use one of three possible approaches: (1) egocentric perspective alone, considered an implicit motor imagery; (2) a mixture between egocentric and allocentric strategies—allocentric during MR (depending on the angle; see Brady et al. 2011; Ni Choisdealbha et al. 2011), and a switch to egocentric during judgment subprocess; (3) an allocentric perspective alone.

Finally, our results showed an influence of hand posture and biomechanical constraints in healthy people during hand laterality judgment only. Except in the case of congenital absence of the upper limbs (Vannuscorps et al., 2012), during hand laterality judgment participants use an egocentric perspective, whether during the rotation process and the judgment, or just during the judgment process. Our study confirms the importance of using a hand laterality task with motor disorder participants. However, caution should be taken before using same–different hand MR with participants with motor disorders (see Amick et al., 2006). This kind of task requires allocentric visual strategies and would not be appropriate for studying cognitive motor impairment.

Our experimental design was not designed to disentangle such processes. This issue will be addressed in a future study. Furthermore, even though our main concern was to study the effects of hand posture and biomechanical constraint, we did not look at RTs for right versus left hands. Apart from this limitation, we will address the following topic in future research: Whether the egocentric strategy is used during stimulus encoding, during the rotation process, or during the final handedness discrimination has to be studied. By defining the exact process in which an egocentric strategy is used, we can propose better tests for participants with motor impairments.

More caution should be exercised when considering hand MR as an implicit motor imagery task, especially when using same–different hand MR tasks in participants with motor impairment. Even the hand laterality task may rely on motor strategies in some, but not all, of its subprocesses. Future research will hopefully disentangle this issue.