Co-Registering Kinematics and Evoked Related Potentials during Visually Guided Reach-to-Grasp Movements

Background In non-human primates grasp-related sensorimotor transformations are accomplished in a circuit involving the anterior intraparietal sulcus (area AIP) and both the ventral and the dorsal sectors of the premotor cortex (vPMC and dPMC, respectively). Although a human homologue of such a circuit has been identified, the time course of activation of these cortical areas and how such activity relates to specific kinematic events has yet to be investigated. Methodology/Principal Findings We combined kinematic and event-related potential techniques to explicitly test how activity within human grasping-related brain areas is modulated in time. Subjects were requested to reach towards and grasp either a small stimulus using a precision grip (i.e., the opposition of index finger and thumb) or a large stimulus using a whole hand grasp (i.e., the flexion of all digits around the stimulus). Results revealed a time course of activation starting at the level of parietal regions and continuing at the level of premotor regions. More specifically, we show that activity within these regions was tuned for specific grasps well before movement onset and this early tuning was carried over - as evidenced by kinematic analysis - during the preshaping period of the task. Conclusions/Significance Data are discussed in terms of recent findings showing a marked differentiation across different grasps during premovement phases which was carried over into subsequent movement phases. These findings offer a substantial contribution to the current debate about the nature of the sensorimotor transformations underlying grasping. And provide new insights into the detailed movement information contained in the human preparatory activity for specific hand movements.


Introduction
In our everyday life, we interact continually with objects. We reach for them, we grasp them, we manipulate them. All these actions are apparently very simple. Yet, this is not so. The mechanisms that underlie them are complex, and require multiple visuomotor transformations entailing the capacity to transform the visual features of the object in the appropriate hand configuration, and the capacity to execute and control hand and finger movements.
Motion analysis of grasping shows that the motor configuration that is formed by the hand in contact with the object represents the end result of a motor sequence that begins well ahead of the action of grasping itself [1][2][3][4][5]. The fingers begin to shape during transport of the hand towards the object. This process of preshaping first involves a progressive opening of the grip with straightening of the fingers, followed by a closure of the grip until it matches object size. The point in time where grip size is the largest (maximum grip size) is a clearly identifiable landmark that occurs well before the fingers come into contact with the object.
In neural terms, grasping behavior can be dissociated into separate reach and grip components (for review, see [6][7][8][9][10][11]). According to this view, computations regarding the grasp component occurs within a lateral parietofrontal circuit involving the anterior intraparietal area (AIP) and both the dorsal (PMd) and the ventral (PMv) premotor areas [12][13][14][15][16][17]. The general agreement is that the processes occurring in AIP constitute the initial step of the transformation leading from representation of objects to movement aimed at interacting with such objects [18,19]. Evidence supporting this view comes from neurophysiological studies showing that the representation of 3D object features influences both the rostral sector of the ventral premotor cortex (area F5) and the ventro-rostral sector of the dorsal premotor area (area F2vr; [15]). According to this model Area F5 plays a primary role in selecting the most appropriate type of grip on the basis of the object affordances provided by AIP to which it is reciprocally connected, thus activating a motor representation of the object. This motor representation is then supplied to area F2vr which keeps memory of it and combines it with visual information provided by cortical areas of the superior parietal lobe to continuously update the configuration and orientation of the hand as it approaches the to-be-grasped object. These properties suggest that F2vr neurons code the continuous activation of the object representation in motor terms, but that they are more dependent than F5 neurons on the visual information during actual grasp. With respect to the reach component, there is agreement that it is subserved by a more medial parieto-frontal circuit including the medial intraparietal area (mIP) termed as the parietal reach region (PRR), area V6A, and the dorsal premotor area F2 [20][21][22][23][24].
A point worth noting is that such dichotomic view has recently been questioned. Evidence from single-cell data [15,46,47] and lesion studies [48] suggests that areas V6a and F2 are also involved in managing specific aspects of grasping behavior such as grip posture and wrist orientation. Similarly, functional magnetic resonance imaging (fMRI) investigations reported grasping-related parieto-occipital and dorsal premotor cortex activations [38,39,42] which might be considered the possible human homologue for areas V6A and F2, respectively. Moreover, it is noteworthy that a recent neuroimaging study, based on the quantification of the modulation of the effective parieto-frontal connectivity, argues against the existence of dedicated circuits for reaching and grasping [49]. Rather, the authors suggest a differential level of effective connectivity in the AIP-PMv circuit depending on the type of grasped objects. Whereas grasping small objects is characterized by a high degree of on-line control requirement, grasping large objects led to an increased coupling in the so-called reaching circuit (V6A-PMd).
Complementary to these approaches, Evoked-Related Potentials (ERPs) measured by electroencephalography (EEG) provide a quantitative measure of the whole brain's electrical activity, revealing the time course of brain activity modulations throughout reach-to-grasp movement from planning to execution. Wheaton and colleagues [50,51] reported the involvement of parietal activity preceding that of the frontal areas in praxis hand movements. In this study, the authors compared motor potentials related to the generation of self-paced simple movements (i.e., thumb adduction) with motor potentials related to self-paced tool-use movements (e.g., hammer pantomime). Motor-related potential showed significant greater amplitude and earlier onset for more complex movements. Specifically, they observed that the motor-related potential in the posterior parietal cortex anticipated that in the frontal areas, and continued as the movement onset approached. They postulated that the complexity of the movement per se (e.g., multiple joint coordination) requires higher neural computation demand, which took place in the parietal lobe.
More recently, Bozzacchi et al. [52] defined the spatiotemporal activity of parietal and frontal areas in self-paced object-oriented actions. By examining motor-related potentials in planning reachto-grasp movement, they clearly showed that parietal areas were involved in the early phase of planning. Such parietal activity started long before movement onset and was followed by a classical fronto-central component. The observed timing of parieto-frontal interaction in reach-to-grasp movements further confirmed previous evidence showing that parietal areas provide premotor areas with grasp-related information [49].
Another study [53] considering a precuing task found higher late Contingent Negative Variation (lCNV; [54][55][56][57][58]) amplitude over Cz and FC electrodes when the cue provided information regarding the type of grip to use and/or the level of force required to pull an object. Furthermore, whereas the force-related lCNV was more distributed over fronto-central electrodes, the griprelated lCNV was chiefly restricted to parietal and premotor areas. Aside from outlining the composite nature of the such ERP component in terms of high-and low-level planning processes, these findings confirmed that a functional parietal-premotor network is involved in the planning of grip [53].
To sum up, these studies suggest that in humans, like in monkeys, reach-to-grasp movements involve a large network of interconnected structures in the parietal and frontal lobes [8,9,59]. And, that this cortical network is differentially involved for the control of distinct aspects characterizing the planning and the control of reach-to-grasp movement. Nevertheless, how the neural control systems interact with the complex biomechanics of moving limbs -as to help us to identify the operational principles to look for in reach-to-grasp studies and, more in general, in motor control -remains an open question. In this respect, it is only through the use of converging techniques with different characteristics that we might fully understand how the human brain controls the grasping function [9]. What is so far lacking in the literature on cortical control of grasp in humans is a systematic documentation of the time course of neural activity and kinematical signals during performance of grasp. To fill this gap our study investigated ERPs with kinematical signals in order to provide deeper insights into the neuro-functional basis of grasping in humans. Participants were requested to perform a natural reach-to-grasp movement towards a visually available target object which could be either of a small size, requiring a precision grip movement (i.e., the opposition of the thumb and index finger) or of a larger size requiring a whole hand grasp (i.e., the opposition of the thumb with the other fingers) in order to be grasped. Differently from previous studies [50], we did not investigate ERPs evoked by a cue signaling specific object's intrinsic features, but by the target stimulus itself. Such approach may allow to examine how information about an object's geometric properties is transformed into specific motor programs more directly. We hypothesize that the ERP analysis may reveal the time course of activation of the differential cortical areas related to the planning, initiation and on-line control of reach-to-grasp movements and how such activity varies depending on grasp types. Kinematic analysis will provide an objective standard for parsing hand movements into distinct stages and for determining their temporal occurrence. Hand movements kinematics, acquired by means of a three-dimensional (3D) motion analysis system synchronized with the EEG recording system, will make possible the correlation across neural and kinematical temporal events.

Ethics Statement
The experimental procedures were approved by the Institutional Review Board at the University of Padua, and were in accordance with the Declaration of Helsinki (Sixth revision, 2008). All participants gave their informed written consent to participate in the study.

Participants
Twenty-two students, recruited from the Faculty of Psychology at the University of Padua, took part in the study. They had a mean age of 23.68 years (SD = 2.49; range = 19.28; 11 females); they were all right handed, as measured by the Edinburgh Handedness Inventory [60], with normal or corrected to-normal vision, and without neurological or psychiatric pathologies.

Apparatus and Procedures
The participant was seated on a height adjustable chair so that the thorax pressed gently against the front edge of the table and the feet were supported. The position of the head was controlled by means of a head-chin-rest. A pressure sensitive starting switch was positioned 15 cm anterior to the mid-line of the participant's thorax. With the hypothenar eminence of the right hand placed upon this switch, the starting position was slight shoulder flexion and 70-80u of internal rotation, 90u of elbow flexion, semipronation of the forearm, 5-10u wrist extension and opposition between the pads of the index finger and thumb. The experimental stimuli were either a small or a large wooden sphere ( Figure 1). The small sphere was of diameter 3 cm whereas the large sphere was of diameter 7 cm. The stimulus was placed upon the working surface 30 cm directly in front of a pressure sensitive starting switch ( Figure 1). Visual availability of the stimulus was controlled via Plato Liquid-crystal shutter glasses (translucent Technologies, Toronto, ON, Canada) worn by the participant throughout the test (Figure 1). Under computer control, the shutters change from translucent to transparent within 10 ms and return to translucent in 2 ms. All participants naturally adopted a precision grip (PG, opposition between the index finger and thumb) to grasp the small stimulus and whole hand prehension (WHP, all fingers opposing the thumb) to grasp the large stimulus. And they were requested to maintain their gaze fixed towards the stimulus location. There were two experimental conditions, a grasping large (GL) condition in which participants grasped the large stimulus adopting a WHP. And a grasping small (GS) condition in which participants grasped the small stimulus adopting a PG. During a training session task instructions were given to participants. The experimenter explained the task consisting in reaching towards and grasping the presented stimulus. Once the participant was comfortable with the task they performed a total of 80 trials, 40 for the GL and 40 for the GS conditions. The sequence of events was the following. At the start the shutter glasses were in a closed (opaque) state. At the time the shutter glasses opened (i.e., became translucent) the stimulus become visible and the participant was instructed to initiate the reach to grasp movement towards the stimulus. The shutter glasses remained open for the entire duration of the movement. Trials were administered in two blocks presented in a pseudorandom order. All failed trials were reintegrated and presented randomly later in the block. ERPs and kinematical recordings started at the time the shutters glasses became translucent ( Figure 2).
Kinematical recording and data processing. Reflective passive markers (0.25 cm diameter) were attached to the following points of the reaching limb: (a) wrist -radial aspect of the distal styloid process of the radius; (b) index finger -radial side of the nail; and (c) thumb -ulnar side of the nail ( Figure 1). Movements were recorded with the SMART system (BTS, Milan, Italy). This consisted of six infra-red cameras (sampling rate 200 Hz) inclined at an angle of 45 degrees to the vertical, and placed around the table (Figure 1). The calibrated working space was a parallelepiped (length 50 cm, breadth 50 cm, height 50 cm) from which the spatial error measured from stationary and moving stimuli was 0.4 mm. Coordinates of the markers were reconstructed with an accuracy of 1/3000 over the field of view and sent to a host computer. The SD of the reconstruction error was 1/3000 for the vertical (Y) axis and 1.4/3000 for the two horizontal (X and Z) axes. The SMART analyzer software package was used to assess the data. This gave a three-dimensional reconstruction of the marker positions. The data were then filtered using a finite impulse response (FIR) linear filter -transition band of 1 Hz (sharpening variable = 2; cutoff frequency 10 Hz). The reach component was assessed by analysing the trajectory and the velocity profile of the wrist marker. The manipulation component was assessed by analysing the trajectory of each of the hand markers, and the distance between these two markers. Reaction time was defined as the time interval between the opening of the crystal liquid lenses and the release of the start button upon which the hand was resting. Movement duration was calculated as the time between movement onset (defined as the time at which the wrist first began to move) and the end of the action (defined as the time when the fingers closed on the target and there were no further changes in the distance between the index finger and thumb). The period following this, whereby the stimulus was lifted, was not assessed. AFz. Impedance of all electrodes was kept below 5 kV. The EEG signal were digitized at a sampling rate of 512 Hz (16 bit AD converter), and high-pass filtered at 0.15 Hz. Data processing was performed by BrainVision Analyzer 2 software (Brain Products GmbH, Gilching, Germany). Continuous EEG was off-line lowpass filtered at 30 Hz. Epochs were extracted separately for each of the two type of object stimuli (small, large), time-locked at the time the glasses were opened (i.e., stimulus appearance) and lasted 2000 ms. The considered time window encompassed the time at which the shutter glasses opened and the time at which the object was grasped (see Figure 2). Artifacts were corrected by means of Independent Component Analysis (ICA) applied on all epochs together, regardless of stimulus size. The ICA correction was performed by using a toolbox in the EEGLAB software (9.0.3.4b version; [59]). The ICA allows for the identification of the independent components in the segmented EEG signal by taking simultaneously into account frequency, timing and location on the scalp. This procedure helps in isolating artifactual components, such as blinks and head muscles' contraction [61][62][63]. In addition, epochs containing amplitude deflection greater than 675 mV was rejected for all the recorded channels prior to further analysis. The signal was then baseline-corrected against the mean voltage during the 200 ms prior to object appearance. Epochs containing erroneous movements were discarded. A total of 38.31 epochs (SD = 1.84) for each size condition were included within the statistical analyses. Based on visual inspection of grand average waveforms and amplitude scalp maps, the following ERP components were statistically analyzed: amplitude and latency of P300, namely the positive peak evoked 200-400 ms following stimulus appearance at occipital (O1, O2) and parietal (Pz) sites; amplitude and latency of N400, namely the negative peak occurring at 300-500 ms after object appearance at frontal (F3,

Data Analysis
Mean values for reaction time, movement duration and each kinematical measure were compared between grasping conditions (small, large) by means of paired t-test. ERP components were analyzed by means of separate repeated measure ANOVAs (see 'results' section). The alpha level of significance was fixed at 0.05. Before running the analyses, we checked for all the main assumptions behind this statistical parametric model (i.e., normal-ity and sphericity). Kolmogorov-Smirnov test revealed that the normality assumption was satisfied. In all ANOVAs, Mauchly test showed that the sphericity assumption was not violated. The effect size of ANOVA results was quantified by means of partial etasquare values (g 2 p ). P-values of t-test and correlation results were corrected for multiple comparisons using a false discovery rate (FDR). Post-hoc comparisons of ANOVA were corrected by Bonferroni method. Correlation analyses by means of Pearson's r coefficient were performed between kinematical and ERP measures as well as between movement duration and ERPs events.

Reaction Time and Movement Duration
Reaction time did not differ between the GL and the GS conditions (5176137 vs 5006123 ms; p.0.05). However, movements towards the smaller stimulus had a longer duration than

Kinematics
The manipulation of object size had predictable effects on the reaching and the grasping component, respectively. In particular, the reach component was characterized by a bell-shaped wrist velocity profile with a single peak. The latency of this peak did not differ significantly with stimulus size (4756123 vs 4766130 ms).

Evoked Related Potentials
ERP waveforms of grand-average, locked to glasses opening (i.e., object appearance), were characterized by an early negative peak at around 100 ms, more marked at parietal and central electrode sites, which showed similar amplitude and latency for the two grasping conditions. Then, differences in ERP amplitude between the two conditions become evident. Specifically, a positive peak at around 300 ms (P300), maximally expressed at parietal electrode sites, showed higher amplitude for the GL than for the GS conditions. Subsequently, a negative electrical activity, peaking at around 400 ms, evident at central and frontal electrode sites, and sustained for a time-window lasting from 400 to 800 ms, clearly showed higher amplitude for the GS than the GL conditions. The polarity, the temporal trend and the scalp distribution for such component suggests that this is likely linked to the motor component of action planning and to premotor areas, therefore we termed this component as motor-related N400 (m-N400). From 800 to about 1200 ms after object visual availability, a slow ERP deflection from negative to positive values at all electrode sites was found, which was characterized by a similar pattern for the two conditions (see Figures 3 and 4). Then, a sustained positivity was evident from 1200 to 2000 ms, which was higher for the GS than for the GL condition. A time window corresponding to the time at which the object was approached and contact points have to be optimized. P300. Figure 3 depicts grand-average waveforms for the two grasping conditions at parietal sites P3, Pz, and P4, in which a P300 was evident. The amplitude and the latency of Post-hoc comparisons revealed that P300 peak reached maximal amplitude earlier for the left and midline sites (P3 and Pz, respectively) compared to the right site P4 (p = 0.024 and p = 0.047, respectively). No significant differences between P3 and PZ were detected. The scalp map in Figure 3, showing the topography of differential amplitude (GL -GS), confirms that within the 300-350 ms time window ERPs were higher for the GL compared to the GS condition. m-N400. Figure 4 illustrates grand-average waveforms for the two grasping movements at the following electrode positions: F3,

Correlations between Kinematic and ERP Events
Mean amplitude and latency of P300 and m-N400 components were averaged at electrode sites in which they were maximally expressed. Specifically, at P3 and Pz for the P300 and at Fz and FCz for the m-N400. Then these values were correlated with movement duration and the considered kinematic measures, namely time to peak velocity and the time of maximum grip aperture. No significant correlations were detected when considering the relationship between kinematical and ERP events.
However, as depicted in Figure 5

Discussion
We set out to investigate kinematics and ERP activity during reach-to-grasp movements performed towards either a large or a small stimulus. Overall the results indicate that the two grasping conditions determine a modulation in timing and amplitude of specific kinematic landmarks and ERP components.
In terms of behavioural performance, our results are in line with previous kinematical studies [2,3,5,[64][65][66]. Literature findings consistently indicate that, with respect to whole hand grips, precision ones are characterized by a longer movement duration, and an anticipated and lowered amplitude of maximum grip aperture (e.g. [2][3][4][5]64]). Customarily, no differences in the times to peak wrist velocity regardless of the type of grasp are usually found. In the same way, here the reaching component was characterized by a bell-shaped wrist velocity profiles with single peaks with no differences in the latency of these peaks depending on grasp type. The total duration of reach-to-grasp movements was longer and the time and amplitude of maximum grip aperture was earlier and smaller for the GS than for the GL condition. These findings indicate that the size of the stimulus influenced selectively the planning and the execution of the reach-to-grasp movement. This is an important aspect of the present study because in order to ascertain the effects that such differential processing might have on ERPs, it is necessary to demonstrate that the participants' movement show classic kinematic signatures depending on grasp conditions.
For an efficient grasp visual information regarding an object's physical properties (e.g., size) must be transformed and used to select an appropriate motor command. For both humans and monkeys the key cortical circuit involved in this transformation involves the anterior intraparietal area (AIP), the ventral and dorsal premotor cortices (PMv and PMd, respectively), and the primary motor cortex (M1) [6,34,38,49,59,[66][67][68][69][70]. AIP contains neurons that discharge in relation to specific object properties [68], whereas many grasp-related ''canonical'' neurons [59] are found in PMv [67,69,70]. Experiments in monkey [67,68] and humans (for review see [9]) appear to show that object properties are encoded as a gradient along the AIP-PMv-M1 axis, with the object being first represented in visual attributes and then in terms of an appropriate grasp.
With this in mind, our EEG recording revealed differences between grip types in the ERPs evoked by stimulus appearance. Such differences were concerned to, both visuo-spatial processing and motor planning. At first, differences in amplitude between conditions become evident over parietal sites following object appearance, at around 300 ms (P300), that is during the planning phase. Such difference remained significantly distinct during the execution phase up to 800 ms. In particular, the peak amplitude for the P300 component was higher for the GL than for the GS condition. This finding might reflect the greater amount of visuospatial information to be extracted from larger objects. In this view, object metric properties, such as size, are processed at parietal level. Although we cannot firmly determine the brain source of such activation, it is likely that it reflects AIP activity concerned with the amount of visual information related to the  object, and it might be seen as the initial step of the transformation leading from representation of objects to movement. These findings fit with those reported in a recent study on the role of anterior intraparietal sulcus in sensorimotor integration of visually guided hand movements [71]. Here it was shown that the suppression of alpha oscillation over the parieto-occipital electrodes occurred at 220-240 ms following object presentation. Furthermore, the evidence of a parietal involvement is in agreement with neurophysiological evidence showing that parieto-occipital neurons are sensitive to grip type [46,47]. Assuming that our present results reflect this kind of activity, they might provide a further confirmation that motor plans requiring hand preshaping extend farther anteriorly into both the precuneus and the middle intraparietal sulcus [30]. In the light of previous ERPs studies looking at brain sources of motor-related potentials [52] it is likely to suggest the involvement of additional areas, such as the superior parietal lobe.
Subsequent to the parietal activation we found a negative electrical activity, peaking at around 400 ms following object appearance (m-N400), which was evident over central and frontal electrode sites. The amplitude of this component was higher for the GS than the GL conditions and such difference was significant within a time-window lasting from 400 ms up to 800 ms following object presentation. The polarity, the slow temporal trend and the scalp distribution suggest that such component reflects motor planning and that it is linked to premotor activity [72].
Unlike previous ERP studies on motor planning (e.g. [51][52][53][73][74][75]), we did not analyze self-paced movements and we did not adopt a precuing task, but we examined EEG deflections evoked by stimulus appearance which prompted a spontaneous grip movement. Nevertheless, the spatio-temporal characteristics of the m-N400 might be assimilated to an index of motor planning and it is strongly influenced by motor variables. Interestingly, stimulus size significantly affected the latency and the topographical location of such component. The m-N400 peak had a later onset and a wider fronto-central distribution for the GS than for the GL condition. This suggests that the planning phase needed for a precision grip movement takes longer and involves more (dorsal) areas. Taken together, these findings indicate that the parietal' visual information, encoded in an ''object'' reference frame, is subsequently multiplexed into a ''grasp'' reference frame within a premotor network possibly involving both the PMd and the PMv. Beyond forming a critical node in the visuomotor planning circuit underlying grasping, recent evidence suggests that different premotor areas (e.g., PMd and PMv) might have dissociable processes. Experiments in humans and monkeys indicate that PMv is more involved in the distal components of the action, such as hand preshaping and specific grip responses (e.g. [41]). PMd, instead, appears to be more involved in the online control of movement (e.g. [15,38]). Given that most of these previous descriptions are based on the characterization of activity stemming during the movement itself, the decoding of different planned hand movements shown here provides a significant additional dimension to such descriptions, which fits with previous functional imaging reports (e.g. [30]).
The difference in amplitude between the GS and the GL condition may reflect the need for additional sensory-motor control mechanisms for the more accurate GS condition. A result that is in keeping with the evidence that accuracy has the ability to affect readiness potentials [72]. In humans, evidence from developmental, psychophysical, neuropsychological and neuroimaging studies seems to suggest that precision grips (as for our GS condition) are characterised by a greater degree of complexity. Firstly, the ability to perform independent finger movements and grasp with the precision grip is not present when voluntary grasping emerges (e.g. [76]). Secondly, consistent results within the adult reach-to-grasp behavioral literature [6], and those obtained in the present study, indicate that the performance of a precision grip is characterized by the need for additional time. Berthier et al. [77] also showed that as visual information and object size decreased, subjects had longer movement times, slower speeds, and more asymmetrical hand-speed profiles. This kinematic characterization reflects the adoption of a strategy following the principles of the Fitt's Law [78], implying that the difficulty of the task is reflected in movement kinematics. Thirdly, in macaques, it has been revealed that of the premotor area F5 neurons active during grasping, the most frequent were those involved in precision grips [16] whereas whole hand neurons were encountered much less frequently [16,66]. Finally, neuroimaging studies indicate that premotor activity increase more during the execution of a movement toward a small object than toward a large one reflecting the increased planning and on-line control required by grasping small objects [39,49]. An alternative possibility related to the modulation of the N-400 activity might be concerned with inhibition. Previous studies from Kok et al [79], demonstrated a frontal N400 elicited by a No-Go stimulus in a Go\No-Go paradigm. In this view the greater N-400 activity for the condition in which a precision grip is performed would stem from inhibiting the opening of the whole hands in order to specify index finger and thumb when a precise grasping is requested. This idea would imply that the 'simpler' whole hand grasp would be prepared by default and then precision grip would be specified. Evidence that this process might be in place comes from reach-to-grasp perturbation studies in which the passage from whole hand to precision grip movements has been measured [80].
Altogether, the present findings confirm a parietal processing related to the vision of a particular graspable object which provides premotor cortices with grasp-related information that allows neurons in these areas to be tuned to the upcoming grasp and on-line control. Importantly, they provide an addition to current literature by revealing the time course of the visuomotor transformation process starting from the 'parietal' visual object discrimination activity to the 'premotor' activity concerned with the assignment of specific hand configurations depending on object's size. Furthermore, they show that once such differential process, depending on grasp type/stimulus size ensemble within the 'parietal' and the 'frontal' component of the grasping circuit is started before movement initiation, it remains sustained throughout the entire action. This indicates that these areas participate in a sensorimotor network involved in grasp planning, prediction of sensory stimulation, and monitoring of appropriate execution of the desired actions. And also suggest the role of the parietal (possibly the AIP region) and premotor cortices (possibly PMv and PMd) not only during the execution of reaching-to-grasp movements as previously reported [81][82], but also during the planning phase. A result which is in line with neurophysiological evidence showing that the discharge of F5 neurons is tuned for specific grasps well before movement onset and this early tuning was carried over in the preshaping period of the task. In line with this evidence we found a marked differentiation across different grasps during the premovement phase which was carried over into early grasp phases (as witnessed by kinematical analysis) characterized by a premotor kind of activity. Altogether, these properties are consistent with the notion that premotor areas play a role in translating visual information about an object's physical properties into the appropriate motor plans to interact with the same object [16,67,69,83].
Another aspect of the present findings is concerned with some relationship between kinematical and ERPs events. Of interest is that the ERPs differences noticed during the planning phase at both parietal and premotor level depending on grasp conditions persisted all along the unfolding of the action. And remained statistically different at the time key kinematic landmarks such as the time of maximum grip aperture occurred. A greater peak of maximum grip aperture and a modulation of the time occurrence for this peak corresponded to a significantly different level of activity for ERPs components. This signifies that when the stimulus become visually available sensory and motor processes specifically tailored to process the stimulus were established and maintained active as to organize the kinematical unfolding of the movement.
Although we did not find any significant correlation between the times at which peak ERPs components and the considered kinematical landmarks occurred, we found that for both the GL and the GS conditions the individual mean latency for the premotor m-N400 component significantly correlated with the individual mean for movement time. This might indicate that at the time the 'parietal' information regarding the visual aspect of the object are integrated within the premotor area (possibly PMv) with the motor prototypes adequate to successfully grasp it, the time to perform the action is kept into account. Similarly, an estimate of movement time, possibly performed at the level of PMd, might serve to plan the amount of on-line control required by the movement. This mode of programming might keep the timing of the commands independent from the spatial parameters of the movement. In other words, selection of the muscles needing to be activated to carry out a given task can be modified, or the kinematics can be modulated within a centrally generated temporal template that determines the co-ordination of a given action. This might appear to be the easiest and most readily chosen organizational option of the neural system to compensate for the postural and joint kinematic variability characterizing reach-to-grasp actions.

Conclusions
In the present study, we have explored the kinematic and ERP dynamics during a reach-to-grasp task. Together, kinematical and ERPs data confirm that the object size/type of grasp ensemble has the ability to modulate both the behavioural and the neural components underlying this kind of action. Analysis of the changes at the level of the ERPs components revealed that the parietofrontal network is modulated differently by prehension movements towards differently sized objects at both planning and execution level. The correlation between movement time and ERPs components is suggestive of a mode of programming relying on a centrally generate template within which dynamic aspects of the movement are coordinated. In a broader perspective, this work underlines the use of EEG for the investigation of movements with unique cortical motor processes such as reach-to-grasp movements.