Exp Brain Res (2006) 173: 395–406 DOI 10.1007/s00221-006-0378-x
R ES E AR C H A RT I C L E
Lari Vainio Æ Rob Ellis Æ Mike Tucker Ed Symes
Manual asymmetries in visually primed grasping
Received: 8 March 2005 / Accepted: 24 January 2006 / Published online: 18 February 2006 Springer-Verlag 2006
Abstract Previous research has shown that the task irrelevant size of familiar objects facilitates compatible precision and power grip responses. The present study examined whether the task irrelevant size of novel objects produces the same compatibility effect. However, the main objective of the study was to investigate whether visually primed precision and power grips are manually asymmetric. Experiment 1 showed that the size of a novel prime object does facilitate compatible precision and power grips, even when both the object itself and the grasp type are irrelevant to the current task. However, this effect was only found when the precision grip was made with the right hand (RH) and the power grip was made with the left hand (LH). When these grips were made with the opposite hands, the effect was absent. Experiment 2 replicated the LH bias for large objects and the RH bias for small objects when power and precision grip responses were replaced with simple LH and RH button-press responses. It appears that the two hemispheres are specialised with regard to precision and power compatible objects. Keywords Asymmetries Æ Precision Æ Power Æ Grasp Æ Affordance
Introduction Recently there has been a great deal of interest in the mechanisms underlying the planning and control of visually guided movements. Researchers have shown manual asymmetries in visually guided movements. Furthermore, the reach and grasp components of reachto-grasp movements may be controlled in part by different central structures (Jeannerod 1981). Despite
L. Vainio (&) Æ R. Ellis Æ M. Tucker Æ E. Symes School of Psychology, University of Plymouth, Drake Circus, PL4 7AA, Plymouth, England E-mail:
[email protected] Tel.: +44-1752-233146
evidence of manual asymmetries in visually guided reaching, manual asymmetries in grasping have not been studied to the same extent. The principal goal of this paper, therefore, was to examine manual asymmetries in grasping. More specifically, this paper investigated whether the programming of precision and power grips (as engineered through visual priming rather than visual guidance), might be lateralised in the brain. Virtually all people prefer one hand to the other in making skilled movements. A majority of the population are more proficient with their right hand (RH) than their left hand (LH). The laterality of manual movements has been thought to be the product of the specialisation of each hemisphere for different cognitive, visual and/or motor information processing functions (e.g. Goodale 1990). A goal-directed manual aiming task (Woodworth 1899) has been one of the most common methods in research on manual asymmetries in visually guided movements. This has demonstrated faster and more accurate aiming movements of the RH (e.g. Fisk and Goodale 1985; Elliott et al. 1993), and a RH superiority in making small adjustments to the movement trajectory as the hand approaches the target location (e.g. Mieschke et al. 2001). This is often attributed to a greater ability of the left hemisphere in processing the perceptual and/or motor information required for motor control during ongoing movements (e.g. Annett et al. 1979). Alternatively, it has been suggested that the RH system may be more proficient at the utilisation of kinesthetic feedback (Woodworth 1899). Neurophysiological and neuropsychological research suggests that the left hemisphere is associated with the computation of many cognitive-motor processes such as the selection of motor programs for sequential movements (e.g. Kimura and Archibald 1974). However, a dominant arm advantage in reaching accuracy is not evident during ‘‘ballistic’’ (low-precision, high-speed) movements and could be observed only when the precision requirements of a task are increased (e.g. Todor and Cisneros 1985). In a manual aiming task, participants point at a target. However, more typical manual actions involve
396
reaching and grasping. Jeannerod (1981) separated human prehension into two independent motor programs, reaching and grasping, that involve separate brain regions. More recently, Jeannerod et al. (1995) have suggested that planning the reach component in prehension movements is largely based on analysing the spatial attributes of the target object such as distance and direction. In contrast, planning the grasp itself is largely based on an analysis of the object’s intrinsic properties such as size (see also Roy et al. 2002 for an argument favouring partially independent and inter-related visuomotor channels). Support for the distinct visuo-motor channel hypothesis comes from evidence that separate circuits transform visual information into motor codes in reaching and grasping. In monkeys, the connections between the anterior intraparietal area (AIP) in the posterior parietal cortex (PPC) and the F5 neurons in the premotor cortex, code information relevant to grasping, namely intrinsic object properties such as size and shape (e.g. Jeannerod et al. 1995). In sharp contrast, the ventral intraparietal area (VIP) in the PPC codes object position and orientations in peripersonal space. This information about a target is passed to the F4 neurons in the premotor cortex, which represents the arm’s goal position and is then responsible for setting up the initial reach program (Colby et al. 1993). Thus, the connections between area the VIP and the F4 form a circuit that transforms information relevant to reaching, namely visual information about an object’s position (see Rizzolati et al. 1998 for a review). Human AIP is a likely homologue of macaque AIP, an area with neurons that are activated by the viewing and grasping of specific shapes. Like macaque AIP, human AIP demonstrates activation during both the visual and somatomotor phases of a delayed grasping trial (Culham 2004). In light of this kind of evidence, it is possible that when people plan and execute actions towards objects of different positions, sizes, orientations and shapes, the spatial attributes of the target are primarily analysed for programming reaches, whereas the intrinsic attributes of the target are primarily analysed for programming grasps. Napier (1956) divided grips into precision and power grips from a functional and a phylogenetic perspective. The precision grip (the use of a thumb–index grip) has developed in primates for manipulation of small objects whereas the power grip has developed for holding and grasping larger objects with high stability. There is some evidence that a precision grip engages neural circuits that are different from those engaged during power grips (e.g. Ehrsson et al. 2000). Interestingly, some research in monkeys suggests manual asymmetries in computing precision and power grips. For example, Hopkins et al. (2002) showed that in chimpanzees the RH is more frequently used in making precision grip. Furthermore, the dominant and non-dominant hands have special roles in bi-manual manipulation movements (the stabilizing function of the non-dominant hand and the manipulative function of the dominant hand). The
optimal hand organisation in bi-manual movements might contribute to pressure for an anatomical separation of the two grip types. This present paper asks whether, similar to reaching, grasping also has manual asymmetries in humans. Specifically, this paper investigates whether or not our programming of power and precision grips is manually asymmetric. In normal reach-to-grasp paradigms, it is difficult to measure the respective roles of reaching and grasping in manual asymmetries. However, because the present investigation concerns manual asymmetry of precision and power grip programming, it was essential that reach programming could be ruled out. In requiring an experimental task that only required participants to plan and execute power or precision grips, with no reaching involved, this study used the stimulus-response (S-R) compatibility paradigm presented by Tucker and Ellis (2001). In this paradigm, the size of the viewed object facilitates precision and power grip responses that are made to some object property, even though the object size is irrelevant and participants are not grasping the target object. This paradigm is discussed in more detail below. Object affordances Kinematic studies (Jeannerod 1988; Jakobson and Goodale 1991; Goodale et al. 1994; Gentilucci 2002) show that object affordances (i.e. action-relevant attributes of objects) influence both the selection of the type of grip and the grasp kinematic implementation. More recently, Tucker and Ellis (1998, 2001, 2004) and Ellis and Tucker (2000) have demonstrated in a series of S-R compatibility studies that a viewed object (or photograph of an object) automatically facilitates responses that are compatible with the action-relevant attributes of that object (such as size and orientation). This occurs, despite there being no intention to act on the object itself. Of most relevance to present purposes is a study by Tucker and Ellis (2001), which examined precision and power grip activation initiated by the size of a seen object. In this study, participants viewed graspable objects that belonged to ‘‘manufactured’’ or ‘‘natural’’ categories. Half of the objects in both categories were small and would normally be grasped with a precision grip (e.g. grape, screw) and half were large and would normally be grasped with a power grip (e.g. cucumber, hammer). Participants held two response devices, each equipped with an inlaid micro-switch, simultaneously in their dominant hand. The precision grip device was small and square, and was held between the index finger and the thumb. The power grip device was larger and cylindrical, and was held in the palm of the hand by wrapping the remaining fingers around it. Participants were asked to judge whether the object belonged to a manufactured or natural category by squeezing the precision or power device. It was found that those responses that were compatible with the presented object size were significantly faster and more accurate. For
397
convenience, we shall refer to this finding as the ‘‘object-size effect’’. Tucker and Ellis (2001) also observed the object-size effect in a bi-manual task when one response device was held in one hand and the other device was held in the opposite hand. The object-size effect has also been observed when participants categorized names (i.e. written words) of the objects that could be normally grasped with precision grip or power grip (Tucker and Ellis 2004) suggesting that the effect operates at the level of grip planning. We suggest that the object-size effect is consistent with the possibility that action-relevant object properties (in this case size) activate already existing motor representations of an object’s afforded actions. This, in turn, suggests that the motor system has a central role in visual object representation, whether or not an action that is primed by the viewed object is actually executed. Murata et al. (1997) have presented electrophysiological evidence supporting this view. They demonstrated that neurons in the F5 and the AIP, which were shown to form a parieto-frontal circuit that transforms visual information for grasp computing (as mentioned above), fire during the grasping and additionally discharge to the presentation of graspable objects, even when no immediate action upon the object is allowed. Fagg and Arbib (1998) have developed a neural network model, in which the connections between these same areas in the PPC and premotor cortex play a principal role in the generation of object affordances. Furthermore, the view that actions are encoded as a part of object representation is also supported by brain-imaging studies (e.g. Martin et al. 1995; Chao and Martin 2000; Grezes and Decety 2002; Handy et al. 2003). Finally, it should be mentioned that, in the object-size effect, the grip is not ‘‘visually guided’’ in the same sense that movements are guided in traditional visually guided motor tasks because, in the object-size task, participants are not required to grasp the target object. Rather the size of the target object facilitates precision and power grip responses that are performed with separate devices. However, we presume that, in the object-size effect, the influence of the object size on grasp reflects the same underlying action programming mechanisms, which are operating in grasp programming of traditional visually guided reach-to-grasp tasks. Consistent with this view, human AIP is activated in the object-size effect task (Grezes et al. 2003). Thus, we assume that these neurons (at least partly) underlie the grasp potentiation in the object-size effect, and consequently the effect reflects a grasp plan in the reach-to-grasp movement. Furthermore, we propose that the object-size effect reflects the planning of the optimal, final arm posture of grasping in prehension. Therefore, given that the two hands show difference in actions requiring precise motor control, we predict that object size may afford asymmetrically different hands. Thus, we expect precision grip objects to afford the grip type and the hand (the precision grip and the RH, respectively).
Experiment 1 Given that the RH is normally more accurate in visually guided reaching, and given that this accuracy decreases if the precision requirements of the task are lowered, we decided to examine whether similar asymmetries could be observed with visually primed grasping. It was hypothesised that the RH would show superiority over the LH for precision grips, but not for power grips. In order to test this hypothesis, the following response assignments were made: half of the participants held a precision grip device in their LH and a power grip device in their RH (mapping 1) and the other half held a precision grip device in their RH and a power grip device in their LH (mapping 2). While the primary aim of this experiment was to explore manual asymmetries relating to power and precision grips, there were several secondary aims. 1. Object priming: the object-size effect of Tucker and Ellis (2001) was observed when participants were required to categorise a viewed object. Such categorisation required participants to focus attention on the object. However, it is not clear whether the same effect would be observed when the allocation of endogenous attention to the object is minimal or absent. Therefore, it is particularly interesting to examine whether an action-relevant property of an object (in this case size) could influence responses when the viewed object is task irrelevant (i.e. it does not require endogenous attention). The current experiment examines this aspect of attention by presenting participants with task-irrelevant prime objects, and asking them to respond to a target arrow that is superimposed over the prime. 2. Response dimensions: in Tucker and Ellis (2001) participants were asked to respond with a precision or power grip to the object category. The grasp type was therefore a task-relevant response dimension, and consequently participants were likely to code their responses explicitly as precision and power grips. The current experiment investigates whether viewed objects facilitate the precision and power grip responses even though participants are instructed to respond with their LH or RH. Thus, the grasp type is a task-irrelevant response dimension, and participants are not likely to code their required responses explicitly as precision and power grips. This task arrangement minimizes the influence of cognitive factors related to grip selection. 3. Novel objects: in Tucker and Ellis (2001) participants had to categorise familiar objects. Familiar objects have semantic associations. Indeed, as the Tucker and Ellis (2004) study suggests, even categorizing an object name (rather than a photograph of that object) can produce the object-size effect (presumably because participants have semantically derived the size of the object). It has not been established whether the
398
purely visual size-related information of an object can facilitate compatible grasps. We therefore investigated whether the object-size effect would be forthcoming using novel objects that did not carry semantic connotations. The object set consisted of realistic three-dimensional (3D) objects never previously seen by the viewer (see Fig. 1). 4. Time course: Finally, the time course of response activation in the object-size effect has not been reported in relation to two responding hands. The LH and RH might show differential time courses for response activation. We therefore examined the time course of response activation by varying the onset time between the prime object and target.
Method Participants Forty-four participants took part in the experiment and were each run in individual 25 min sessions. All were students at the University of Plymouth and received course credit for their participation. Informed consent was obtained from each subject prior to commencing the task. All participants reported having normal or corrected-to-normal vision and were naive as to the purpose of the experiment. Only the data of the right-handed participants was included in the analysis. However, participants were not told that only the right-handers’ data would be analysed. All but two participants signed the participation form using their RH. The participants’ explicit report about their handedness (asked after
signing the participation form) was consistent with the observed writing hand. The two participants who signed the form using their LH also completed the experiment and were credited for their participation but their data were not used in the analysis. This arrangement was assumed to minimise the chance that participants would falsely report their handedness for receiving the credit. Apparatus and stimuli The display and timing was controlled by a RMAccelerator-Intel: Pentium 2 processor computer, interfaced to a Mitsubishi Diamond Pro900u 19in. colour monitor. There were two response devices, each equipped with an inlaid micro-switch. The precision grip device was small and square (1.3·1.3·0.7 cm3) and the power grip device was larger and cylindrical (11 cm long, 1.8 cm diameter). As the switches depressed in each device, there was noticeable tactile and auditory feedback. The prime stimuli consisted of twenty-four computer generated 3D objects (see example in Fig. 1). Each object had a slightly different ‘‘wood’’ texture and each object had a slightly different variation of a natural brown wood colour. Half of the objects were small and therefore more suitable to be grasped with a precision grip (they subtended a visual angle of approximately 2.3 vertically and 2.9 horizontally). Small objects were in the shape of a ball, cone or cylinder. Half of the objects were large and would normally be grasped with a power grip (they subtended a visual angle of approximately 17.1 vertically and 4 horizontally) and conformed to a grasp-appropriate shape. Large objects were in the shape of a cylinder or capsule shape. Objects had different shapes and surface textures to increase the likelihood of automatic attentional capture by a novel attribute (see Ruz and Lupia´nˇez 2002 for a review of attentional capture). In addition to the prime object stimuli, there was a centrally located black fixation cross (1·1), and two centrally located black target arrows (1·1), one pointing left and the other right (these were interchanged in a randomised order). All stimuli were presented against a white background and presented on the monitor at a resolution of 1,024·768 pixels. Design and procedure
Fig. 1 Example prime object stimuli used in experiments 1 and 2
Participants sat in front of a monitor in a dimly illuminated room with their eyes 50 cm from the centre of the monitor. The height of the monitor was adjusted so that each participant was looking directly at the centre of the display. Participants held the power grip device in their RH and the precision grip device in their LH (mapping 1; 21 participants were run in this mapping), or vice versa (mapping 2; 21 participants were run in this mapping). Participants were familiarized with the switches. They were shown how to squeeze the precision grip device with the index finger and thumb, and the power grip device with a whole hand grasp. Participants
399
were instructed to keep both their arms on the table on which the monitor was placed (40 cm apart and 25 cm in front of the monitor). The leads of both response devices were attached to the table so that the participant’s arm placement was consistent. Each trial was initiated by a black fixation cross. After 1,500 ms the fixation cross was replaced by a randomly selected prime object, which was presented in exactly the same central location as the fixation cross. Prime objects appeared standing vertically and therefore they were equally compatible with a RH or LH grasp. Three stimulus-onset asynchrony (SOA) conditions (150, 300, 600 ms) determined the duration of the prime object presentation. After the SOA period the target (the left or right pointing arrow) was displayed over the prime object in the same location that the fixation cross had previously occupied. The target arrow changed back into the fixation cross after 180 ms. The prime object and the fixation cross were presented until the participant responded. Participants were instructed to respond as quickly as possible with their RH when they saw the right-pointing arrow. Similarly, participants were instructed to respond with their LH when they saw the left-pointing arrow. The participant was asked to focus upon the central point through the whole experiment. Participants understood that maintaining fixation at the central locus was the most efficient strategy when attempting to detect a brief target. In addition, participants were told that the objects that were displayed before the appearance of the target were absolutely irrelevant to the task and therefore could be ignored. Error responses were immediately followed by a short ‘‘beep’’-tone from the computer. Participants were timed out if they did not respond within 3,000 ms. A half minute break divided the experiment into three blocks. Each block consisted of a different set of object stimuli. The objects were randomly assigned to one of the three blocks. During the break, the monitor displayed text that indicated the length of the break and instructions for carrying on with the experiment. The experimental design is illustrated in Fig. 2.
Fig. 2 A schematic of the design for experiments 1 and 2. Note that the design of experiment 2 included one more SOA condition (1,000 ms)
Results Response times The experiment consisted of 432 trials. Reaction times (RTs) were cropped for each participant’s data. RTs two standard deviations from each participant’s overall mean were discarded (2.6%). Condition means for the remaining data were computed. Error trials were excluded from the analysis. Condition means were subjected to a repeated measures ANOVA with the within participants factors of prime object size (small or large), SOA (150, 300 or 600 ms), grip type (precision or power), and the between participants factor of mapping rule (M1: RH/power grip, LH/precision grip and M2: RH/precision grip, LH/power grip). The analysis revealed two significant main effects and two significant two-way interactions that were of secondary interest to this study. There was a main effect of grip type, F(1,40)=12.59, P=0.001, MSE=6,113.64; the power grip responses were made faster (M=278 ms) than the precision grip responses (M=285 ms). Additionally, there was main effect of SOA, F(1.7,67.4)g= 20.79, P
