The activity of single cells was recorded in behaving monkeys while they performed several eye–hand directional motor tasks. The results revealed that in parietal area 7a there exists a directional representation of eye and hand motor space that, contrary to that of superior parietal, premotor and motor cortex, is highly skewed toward the contralateral workspace. In man, the loss of this representation after parietal lesions might explain the emergence of the directional movement disorders of neglect. In fact, although unilateral neglect is consequence of damage to different brain structures, it is more common and enduring after right inferior parietal cortex lesions. Neglect patients ignore and avoid interacting with events occurring in the contralesional part of their physical and mental space. Current theories distinguish perceptual from motor components of neglect. One key feature of the latter is directional hypokinesia, an impaired representation of space for action, evident as difficulty to plan hand movements toward the contralesional part of egocentric space. An impairment of a similar nature is also observed for eye movements. In this study, we offer an interpretation of directional movement disorders of neglect from a physiological perspective, i.e. by focusing on the mechanisms underlying the representation of visuomotor space in parietal cortex.
It was Bálint (1909) who first described a parietal syndrome that included optic ataxia, ‘psychic’ paralysis of gaze, and hemispatial neglect. Bálint's patient had a large bilateral lesion of parieto-occipital cortex. While optic ataxia commonly occurs after superior parietal lesions (Perenin and Vighetto, 1988; for a review, see Battaglia-Mayer and Caminiti, 2002), neglect is mostly reported after damage of the supramarginal gyrus (Brodmann's area 40) of the right inferior parietal lobule (Vallar and Perani, 1986; Leibovitch et al., 1998; Vallar, 2001), although it can also be consequence of frontal and subcortical lesions (Vallar, 2001; Husain and Rorden, 2003). Pure neglect without hemianopia has been reported after superior temporal lesions (Karnath et al., 2001), and in patients suffering from a parieto-frontal disconnection syndrome (Doricchi and Tomaiuolo, 2003), due to damage of the superior longitudinal fasciculus.
The disordered representation of contralesional part of space impairs many of neglect patients' routinely activities, such as shaving, dressing, reading and drawing, so that, for example, they tend to shave only half of their face, read only the right part of a word or sentence, and copy or draw only the right part of a figure or self-portrait. Neglect is a complex syndrome composed by a constellation of symptoms. For instance, neglect patients with normal perceptual awareness can be impaired in encoding space for action. When presented with hierarchical figures (Navon, 1977), consisting of global letters or figures formed by smaller components, they recognize and name the global form, but are unable to cancel out with a pencil in their right hand the local components on the left side of each figure (Marshall and Halligan, 1995). The hallmark of output-related disorders of neglect is directional hypokinesia (Heilman et al., 1985; Mattingley et al., 1992, 1998), which consists of an elongated reaction time and inaccuracy of reaching to visual targets in the contralesional part of space, regardless of the limb used. Neglect patients also suffer from similar disorders in the oculomotor domain (Girotti et al., 1983; Pierrot-Deseilligny et al., 1991; Niemeier and Karnath, 2003; for reviews, see De Renzi, 1982; Husain and Rorden, 2003). Although the nature of eyes deficit is still disputed, lengthening of saccadic reaction time to visual targets in the contralesional space has been reported by several authors (Girotti et al., 1983; Nagel-Leiby et al., 1990; Karnath et al., 1991; Pierrot-Deseilligny et al., 1991). All together, these hand and eye motor disorders reveal a profound impairment in the treatment of directional motor information. Therefore, understanding the representation of visuomotor space in parietal cortex is a necessary step to investigate the mechanisms underlying the motor aspects of neglect from a neurophysiological perspective. To this end, we have studied the dynamic properties of neurons in area 7a of the inferior parietal lobule of monkeys while these performed six different tasks, aimed at assessing the relationships between neural activity and the direction of different forms of visually- and memory-guided eye/hand movements. The behavioural tasks were intended to reproduce items of behaviour that are compromised by inferior parietal lesions in neglect patients. They were aimed at identifying and dissociating the relative influence of eye and hand directional signals on neural activity, at evaluating if and to what extent they were combined at single cell level, at describing the nature of their representation at the population level.
Materials and Methods
Animals, Apparatus and Tasks
Two rhesus monkeys (Macaca mulatta; body weights 4.3 and 6.1 kg) were used in this study. The monkeys sat on a primate chair with head fixed, and the eyes 17 cm in front of a 21″ touch-sensitive (MicroTouch Systems, Wilmington) computer monitor used to display the tasks and control the animals' hand position.
Monkeys performed six different tasks in total darkness. All arm movements were made with the hand contralateral to the hemisphere of recording. Arm and/or eye movements originated from a central position (Fig. 1A) and were directed toward eight peripheral targets (subtending 1.5° visual angle) located on a circle of 7.5 cm radius (23.8° visual angle).
All tasks (Fig. 1B–E) started with the presentation of a central red light that the animal had to touch and fixate for a variable control time (CT; 1–1.5 s). In two arm reaction time tasks (Reaching and Reach-Fixation) to visual targets (Fig. 1B,C), a peripheral red target was lit at the end of the CT. In the Reaching task (Fig.1B), the animals moved the eyes (e) and then the hand (h) to the target, within given reaction times (RT) and movement times (MTe, upper limit 600 ms; MTh, upper limit 800 ms). The first reaction time (RTeh, upper limit 400 ms) probably reflects preparation for both eye and hand movement, while the second (RTh, upper limit 500 ms) reflects preparation for hand movement, which in our experiment started with the eyes already on the target, as result of animals' natural behaviour. At the end of the trial, the eye and the hand stayed on the target for a variable target holding time (THT, 1–1.5 s). In the Reach–Fixation task (Fig. 1C) animals only moved the hand to the target and stayed there for a variable THT, while keeping central fixation. Time limits imposed to different epochs were as in the Reaching task. In the three Memory tasks (Fig. 1D) and the No-Go task (Fig. 1E), at the end of the CT, an instruction signal (IS) was presented for 300 ms. After a variable memory delay (M; 1–3 s), the central light went off, as go-signal. Depending on the colour of the IS, within given RT and MT, the animals made coordinated eye–hand movements (Memory Reach), hand movements (Memory Reach–Fix) or eye movements (Memory Eye) to the memorized target locations, and kept them there for a variable target holding time (THT). In the No-Go task, the animals were required to keep both eye and the hand immobile at the centre of the workspace for the entire duration of the trial. Time constraints for eye and/or combined eye–hand movements in these tasks were relaxed of ∼100 ms to facilitate correct performance by the animal, whose reaction and movement times were, however, much shorter than those imposed by the upper limit. All tasks were performed separately, in a block-design fashion, without a specific order in the sequence of their presentation.
Hand position was monitored using the touch screen, with 0.28 × 0.3 mm (1 screen pixel) resolution. Hand accuracy was controlled through 3cm diameter circular windows (10° visual angle). Eye position signals were recorded by using implanted scleral search coils (1° resolution) and sampled at 100 Hz (Remmel Labs, Ashland, MA). Fixation accuracy was controlled through circular windows (3.5° diameter) around the targets. Eye velocity was calculated in off-line analysis. The onset time of the saccade was defined as the time when eye velocity exceeded 50°/s, and 180°/s, respectively, in the two adjacent 10 ms intervals beginning at the onset time of change of eye velocity. The end of the saccade was defined as the time when eye velocity fell below 50°/s.
The activity of single neurons was recorded extracellularly. A seven-channel multielectrode recording system (System-Echkorn, Thomas Recording, Marburg) was used. In combination with seven dual time–amplitude window discriminators (Bak Electronics Inc, Mt. Airy, MD), recording could be obtained from up to 14 cells simultaneously. Electrodes were glass-coated tungsten-platinum fibres (1–2 MOhm impedance at 1 kHz). The eye-coil, recording chamber and head-holder were implanted aseptically under general anaesthesia (sodium pentobarbital, 25 mg/kg i.v.).
Analysis of neuronal activity.
In each task, the average firing rate during different epochs was computed trial by trial. Significant modulation of neural activity relative to the CT and to target direction was assessed by a two-way ANOVA (factor 1: epoch; factor 2: direction). Directional modulation was defined by the significance of either factor 2 or the interaction factor 1 × 2 (P < 0.05). At single cell level, coexistence of eye and hand directional signals was assessed through significant directional modulation of neural activity occurring at once in at least one hand-related (RT, MT, THT of Memory Reach–Fix task) and one eye-related (RT, MT, THT of Memory Eye task) epoch. The temporal span of analysis of some epochs was adjusted: the first 500 ms of CT were excluded from analysis to prevent potential effects of previous eye and/or hand movement; the first 500 ms of memory delay activity after cue presentation were excluded to prevent carry-over effects of visual responses; finally, the first 300 ms and the last 500 ms of THT were not considered, to avoid influence of previous movements or planning of return movements to the centre during inter-trial interval.
The relationship between cell activity and direction of movement was described through a modified cosine tuning function (Amirikian and Georgopoulos, 2000; Battaglia-Mayer et al., 2000). In our 2-D experimental set-up, the angular variable of interest is the location of the target, univocally determined by the angle α, varying from 0 to 360°.
The standard cosine function (Georgopoulos et al., 1982) has the general form
An additional analysis was prompted by the observation that several PDs were obtained for each neuron, depending on the number of epochs in which cell activity was (i) directionally modulated (ANOVA, P < 0.05; see above for criteria), and (ii) directionally tuned (R2 ≥ 0.7) across different tasks. Therefore, for each individual cell the Rayleigh test of randomness (P < 0.05) was performed to assess whether the distribution of their PDs had a significant mean vector, i.e. whether the distribution was unimodal. The unimodality of the distribution of PDs defines the existence of a global tuning field (GTF; Battaglia-Mayer et al., 2000) whose mean orientation is given by the direction of a significant mean vector. Furthermore, the uniformity of the distribution of the mean vectors of the population of cells with GTF was assessed through the Rayleigh test.
The polar distributions of PDs reported in Figures 5–7 were obtained by taking into consideration one epoch of a given task or similar behavioural epochs from different tasks, as follows:Pooling of PDs in e, g, i, j, k and l was justified by the common tendency of their distributions to be uniform or skewed toward contralateral space (Rayleigh test; P < 0.05), when considered separately.
(a) Eye Memory: memory delay epoch of Memory Eye task
(b) Hand Memory: memory delay epoch of Memory Reach–Fix task
(c) Eye and Hand Memory: memory delay epoch of Memory Reach task
(d) No Go Delay: delay epoch in the No Go task
(e) Eye RT to memorized target: eye RT in Memory Eye and Memory Reach tasks.
(f) Eye RT to visual target: eye RT in Reaching task.
(g) Eye MT to memorized target: eye MT in Memory Eye and Memory Reach tasks.
(h) Eye MT to visual target: eye MT in Reaching tasks.
(i) Hand RT to memorized target: hand RT in Memory Reach–Fix and Memory Reach tasks.
(j) Hand RT to visual target: hand RT in Reaching and Reach–Fixation tasks.
(k) Hand MT to memorized target: hand MT in Memory Reach and Memory Reach–Fix tasks.
(l) Hand MT to visual target: hand MT in Reaching and Reach–Fixation tasks.
(m) Eye THT: THT in Memory Eye task.
(n) Hand THT: THT in Memory Reach–Fix task.
(o) Eye and hand THT: THT in Memory Reach task.
We recorded the activity of 559 individual neurons in area 7a of two left hemispheres of two monkeys, while these performed the six different tasks illustrated in Figure 1. In monkey 1 penetrations were made in a region (Fig. 2A) of the inferior parietal lobule (IPL) which has been identified as area 7a, on the basis of two main criteria — (i) the histological reconstructions of the microelectrode tracks relative to gross-anatomical landmarks, such as the position of the intraparietal (IPS) and superior temporal (STS) sulci; and (ii) the architectonic features of the area of recording — on the assumption that area 7a of Vogt and Vogt (1919) is coextensive with areas Opt and PG, according to Pandya and Seltzer (1982) and Gregoriou et al. (2003). Monkey 2 has not yet been sacrificed. In both animals, the location of IPS and STS were reconstructed from the readings of the top and bottom of neural activity during recording sessions, also thanks to a computer-aided reconstruction (Fig. 2B,D). This was made by interpolating the available values of the top of neural activity across the area of recording, adjusted for dura mater thickening over time. In monkey 1 (Fig. 2B) this procedure allowed a definition of the location of the IPS that perfectly matched that verified on the animal's brain (Fig. 2A), thus supporting the prediction of the reconstruction made in monkey 2 (Fig. 2D), indicating that microelectrode penetrations were performed in the desired area (Fig. 2C). Furthermore, in monkey 2 neurons attributed to area 7a had functional properties virtually identical to those found in monkey 1. In both animals, penetrations were perpendicular to the cortical surface and the extent of recording was usually confined within 2 mm from the top of neural activity, with an average depth across all penetrations of 853 ± 25 (SE) μm. This confirms that the data obtained in this study come from the flat exposed part of IPL.
A summary of the basic results obtained from cells analyzed in a quantitative way (ANOVA, P < 0.05; see Materials and Methods) across the six different tasks is shown in Table 1. Only data from directional modulation are reported.
|Memorized target||Visual target|
|Hand to extrafoveal targets||48.0||43.2||49.2||51.3||53.5||60.2||52.4|
|Hand to foveal targets||50.6||48.8||60.3||51.5||49.9||63.9||56.3|
|Memorized target||Visual target|
|Hand to extrafoveal targets||48.0||43.2||49.2||51.3||53.5||60.2||52.4|
|Hand to foveal targets||50.6||48.8||60.3||51.5||49.9||63.9||56.3|
Numbers are percentages of cells with significant ANOVA (P < 0.05).
It can be seen that the activity of a large population of neurons was directionally modulated when the animal held in memory the location of the target for future eye (47.9%), hand (48.0%) or coordinated eye–hand movements (50.6%), as well as during their planning (RT) and execution (MT) toward visual or memorized locations (Table 1). In most cells, neural activity was also modulated during holding of the eye (58.8%) on memorized targets. Many cells were also influenced by hand-holding on memorized (51.3%) or visual (52.4%) targets, or by combined eye–hand position on them (51.5% on memorized, 56.3% on visual targets). Coexistence of eye and hand directional information (Fig. 3) influenced the activity of a high proportion of individual directional neurons (45.3%), while influence of eye (20.9%) or hand (24.4%) signals alone was less common. This coexistence was not observed in any particular combination of epochs, but occurred in a rather regular fashions across them. The activity of a remaining small population of neurons (9.4%) was modulated by neither eye nor hand signals. It is worth noticing that the activity of many neurons (52.9%) was also directionally modulated in the No-Go task, where no eye and/or hand movements were performed. The activities of those cells that were significantly modulated both in the No-Go task and in the memory epochs of the Memory Eye, Memory Reach–Fix, and Memory Reach tasks were compared. This analysis showed that in most cells, neural activity in the No-Go task differed significantly (ANOVA, P < 0.05) from that observed during the memory delay preceding eye (83.1%), hand (80.5%) or combined eye–hand movements (79.3%). This suggests that beyond spatial attention (Mountcastle et al., 1981; Robinson et al., 1995; Constantinidis and Steinmetz, 2001), signals concerning planning and execution of eye–hand movement influence neural activity in area 7a (Hyvärinen and Poranen, 1974; Mountcastle et al., 1975; MacKay, 1992; Snyder et al., 1997).
Directional Tuning Properties
In most cells, neural activity was maximal (Fig. 4) when the eye and/or the hand prepared to move, or moved in a given PD, or stayed immobile on peripheral targets. Cell activity decreased in an orderly fashion for directions further and further away from the preferred one. This directional tuning was observed across different behavioural epochs and tasks conditions, suggesting that encoding direction of eye and hand motor behaviour is a basic property of neurons in area 7a. In most instances, the PD of individual cells pointed toward the contralateral space (Fig. 4), and this was true for many of the activity types analyzed. As an example, Figure 4 shows the neural activity of three different cells in the form of rasters and spike-density function, as studied in three typical task conditions. In Figure 4A, the activity recorded during the Reaching task is aligned to the RTh epoch. This cell is directionally modulated during preparation of hand movement to foveated target. However, activity in this epoch could be dependent from an eye position signal, as in fact emerged from its significant (P < 0.01) modulation during THT of the Memory Eye task (not shown in Fig. 4). Since neural firing during the latter epoch was significantly (P < 0.01) different from that shown in Figure 4A, one can conclude that this cell's activity is influenced not only by eye position on the peripheral target, but also by preparation of hand movement towards it, i.e. to the fixation point. In Figure 4B, neural activity was recorded during the Reach–Fixation task, and it is aligned to the RTh epoch, that in this case accounts for signals related to hand movement preparation without the influence of directional information about eye position and/or movement. Thus, in this cell neural activity reflects a genuine directional hand reaction- and movement-time signal. Finally, Figure 4C shows the case of another cell with neural activity directionally modulated during eye holding time (THT) in the Memory Eye task, thus encoding an eye position signal. Common to these cells is the significant increase of firing rate for planning or execution of eye and/or hand actions in the contralateral space.
Since directional modulation was observed during different behavioural epochs across the various tasks, it became crucial to study the entire continuum of directional motor activity related to hand and eye movements. For this analysis, the distributions of the PDs of each neuron were studied (Figs 5–7) to determine whether or not, in defined task epochs, each reflecting a particular form of visuomotor behaviour, the cell PDs were uniformly (Rayleigh test; P < 0.05) distributed or not. In the memory epochs (Fig. 5) preceding eye (Fig. 5a), hand (Fig. 5b) or coordinated eye–hand movement (Fig. 5c), as well as in the No-Go task (Fig. 5d), the cell PDs were uniformly distributed across the workspace. A different picture (Fig. 6) emerged during the RT and MT, when the intention to move the eye or the hand evolved into real eye (Fig.6a–d) or hand (Fig. 6 e–h) movement. In most instances, the distribution of PDs of the population of cells studied was not uniform anymore, but displayed a marked anisotropy toward the contralateral part of space. The emergence of this anisotropy was task-dependent, since it was observed not during eye RT preceding movement to memorized targets (Fig. 6a), but before eye movement to visual targets (Fig. 6b), as well as during saccades to both memorized (Fig. 6c) and visual targets (Fig. 6d). The distribution of PDs during the different task epochs that defined the full repertoire of hand motor behaviour (Fig. 6e–h) displayed a marked over-representation of contralateral directional space. This anisotropy was observed during hand RT and MT to memorized targets (Fig. 6e,g), as well as to visual ones (Fig. 6f,h).
Concerning positional signals (Fig. 7), a distribution of PDs skewed toward contralateral space was found during eye holding on memorized targets (Fig. 7a), an epoch that reflects encoding of an eye position signal, with the hand at the centre of the workspace. A different picture emerged for hand position signals (Fig. 7b), since their distribution showed a marked anisotropy that favoured the representation of ipsilateral space. In such instance, the eyes were at the centre of the workspace. When the eye and the hand were both on the same peripheral target (Fig. 7c), these opposite anisotropies tended to cancel each other out, and the resulting distribution of PDs was not significantly different from uniformity.
The data presented above raise the question of a possible evolution in time of the directional properties of 7a cells across the different behavioural epochs of each task. In other words, does cell activity remain directionally modulated during the time course of each task? To answer this question, Figure 8 shows, in form of Venn's diagrams, the number of cells directionally modulated in the different epochs across the task used. The overlap regions, indicating the number of cells which are, at once, directional in two consecutive behavioural epochs, are scarce, since the subpopulation of cells directional in a given epoch is generally different from that recruited in the subsequent one. These results indicate that only ∼20% of the cells remain directionally modulated in two contiguous epochs. This proportion increases to ∼35% during preparation and execution of movement, as well as static holding of arm in space. It is worth noticing that this analysis takes into account only the comparison of two consecutive epochs, and does not compare those that are non adjacent in time.
Concerning directional relationships across task epochs, for every cell many PDs were obtained, one for each epoch in which the directional model fitted cell activity. In order to see, at the single cell level, which were the directional relationships among PDs belonging to different epochs, a quantitative analysis was attempted to evaluate if the distribution of these PDs across epochs and task conditions had any order. Figure 9 shows that this was the case, thus revealing a further manifestation of directional anisotropy in area 7a. In fact, in 106/441 (24.0%) cells studied, the PDs computed during different forms of visuomotor behaviour across tasks, clustered within a restricted part of the workspace. This property, referred to as global tuning field (GTF), has first been observed (Battaglia-Mayer et al., 2000, 2001) in different areas of the superior parietal lobule (SPL). It is worth stressing, that the proportion of cells with GTF is very similar to that predicted by the study of the temporal evolution of directional tuning across task epochs, shown in Figure 8. In area 7a, different cells had different GTFs (Fig. 9), each with particular orientation in space. The mean vector (Fig. 9) provided a quantitative measure of the orientation of the GTF of each cell. It is remarkable that the population analysis revealed that most mean vectors pointed toward contralateral space (Fig. 10a), contrary to what was described in the SPL (Fig. 10b), where they are uniformly distributed across space (Battaglia-Mayer et al., 2000, 2001).
Dynamic Properties of Neurons in Area 7a
This study focuses on the directional tuning properties of cells in area 7a of the monkey. There are three main results to be discussed. First, in area 7a most neurons combine eye and hand signals in their neural activity. Secondly, the overwhelming majority of cells are directionally tuned, and in a consistent population of them the PDs computed across epochs and task conditions cluster within a restricted part of the workspace (GTF). Thirdly, the distribution of PDs during epochs that reflect different eye and/or hand signals is highly skewed toward the contralateral part of space.
The first conclusion of our study is that eye and hand motor signals influence neural activity in area 7a. Early neurophysiological studies (Hyvärinen and Poranen, 1974; Mountcastle et al., 1975; MacKay, 1992) had described the properties of cells in area 7. In these experiments, the influence of eye and visual signals on neural activity was not dissociated from that of hand signals. By adopting memory tasks, Snyder et al. (1997) described the relative preponderance of activity-types reflecting the intention to perform eye versus hand movements in the IPL. This last study did not explicitly refer to properties of cells in 7a, and the proportion of ‘non-specific’ cells combining eye–hand signals across the reach zone studied in posterior parietal cortex (PPC) was observed to be minimal. In contrast, the results of our study show that the majority of cells in 7a are influenced by combined eye–hand information. However, in Snyder et al. (1997), only the memory epochs across two conditions were analyzed, while in our experiment the co-occurrence of eye–hand influences was assessed across a multiplicity of behavioural epochs, thanks to the multi-task approach adopted. Thus, the first main conclusion of our study is that in area 7a most cells encode eye and hand directional visuomotor signals. In our experimental set-up, the origin of eye and/or hand movement was not varied in space. Therefore, each target was attained through movements along a single direction. This implies that we cannot decide whether neural activity is correlated to abstract direction of movement, final target location or a combination of both. However, studies of cell activity in motor, premotor and superior parietal cortex have consistently shown that neural activity is highly correlated to the direction of hand movement within a high-order reference frame, although in a way that depends on a variety of factors, including the starting hand position (Caminiti et al., 1990, 1991; Lacquaniti et al., 1995).
A second result of this study concerns the question of the relationships among different directional signals at single cell level. In fact, in about 25% of directionally tuned cells, the many preferred directions computed during different task epochs clustered within a restricted part of space, the GTF. This visuomotor domain can be considered as the neural substrate whereby parietal area 7a contributes to the distributed system subserving eye–hand coordination. The relative invariance of PDs in front of varying eye–hand behaviour suggests a coding mechanism for coordinated eye–hand movements in head or allocentric reference frames, since cells are tuned to eye–hand actions toward particular spatial locations independently of the specific spatial and temporal combination of effectors involved. The GTF was originally found in the superior parietal lobule (Battaglia-Mayer et al., 2000, 2001), where, however, it accounts for the activity of a majority (∼70%) of the cells studied.
The third, crucial observation obtained of this study concerns the analysis of the tuning properties of parietal neurons at the population level. This analysis revealed a dynamic over-representation of contralateral motor space. In fact, when the animal held in memory the target location for a future eye, hand or coordinated eye–hand movement, as well as in the No-Go task, the corresponding preferred directions were represented in a uniform fashion across the workspace. When these plans evolved into real movement, a dynamic update of the visuomotor representation occurred, characterized by a magnification of contralateral space. This anisotropy was also evident for some special properties of oculo-manual cells, such as the GTF, whose mean vectors pointed to contralateral space.
The dynamic change of configuration of these distributions, when evolving from uniform to anisotropic, could result from at least two different mechanisms: each cell might change its preferred direction during two consecutive epochs, or different populations of cells might be recruited during the temporal evolution of the task. Our results favour the latter hypothesis, for two different reasons. First, the cells that remain directional across adjacent temporal epochs are a small proportion, that can hardly contribute to the change in the configuration of the overall distribution; secondly, the orientation of the PDs of these cells does not change dramatically over time, as also indicated by the existence of a GTF in many of them.
Our results are in line with previous qualitative studies reporting that in area 7 reach-neurons are active preferentially during movements toward contralateral space (Hyvärinen and Poranen, 1974; MacKay, 1992), and also with parietal lesion studies in monkeys. Although the literature on this matter is exuberant (for a review, see Rizzolatti et al., 2000), only selected studies will be discussed here — those that may be relevant to the issue of directional motor components of neglect. Inactivation (Stein, 1978) and unilateral lesions (Faugier-Grimaud et al., 1985) of area 7 leads to inaccuracy of reaching and to elongation of reaction time for movements toward contralateral visual targets. In the former study, this impairment was reported for both arms. In the latter, all monkeys showed a significant increase of reaction time when using the contralesional arm (more severe for movements toward contra- than ipsilateral space). In one animal, an increase of hand RT was observed also for the ipsilesional arm, especially for movements towards contralateral space. However, it is worth noting that in this study the two arms were tested each in only one direction of movement, i.e. the right hand for leftward movements and the left hand for rightward ones. Therefore, it is difficult to conclude whether the elongation of reaction time was related to the arm used (ipsi- or contraletaral) or to the direction of movement (toward or away from the side of the lesion). In any case, the movements impaired were mainly those directed to target locations in the contralesional space. An additional parietal lesion study (LaMotte and Acuña, 1978) shows a directional impairment of reaches to visual targets performed with the contralesional arm, in either the presence or absence of visual guidance of movement. In fact, reaches towards targets in contralesional space were consistently hypometric, since they were systematically misdirected toward the midline, as if the contralateral space was somehow ‘compressed’ or under-represented. In this experiment, the lesion included both superior and inferior parietal lobules. Finally, monkeys with unilateral lesions confined to area 7a (Deuel and Farrar, 1993) are reluctant, slow and inaccurate when reaching to moving targets only in the contralesional space, although able to detect and glance at them.
In humans performing reaches to visual targets in the ispi- and contralateral visual spaces (Kertzman et al., 1997), PET activation of the inferior parietal lobule was predominantly observed in the hemisphere contralateral to the visual space where targets were located, contrary to what observed in the superior parietal lobule, where activation was mainly bilateral.
In our study, over-representation of contralateral space in area 7a was found also for eye position signals, as well as for information concerning eye movement to visual targets. Contralateral anisotropy of gaze fields in monkey's area 7 has been described by Lynch et al. (1977). Concerning eye movement, our results conform to observations made in area 7 (Lynch et al., 1977) and in the lateral intraparietal area (LIP; Barash et al., 1991), where saccadic activity is mostly tuned toward contralateral space. Interestingly, after unilateral IPL lesions, latencies of saccadic eye movements toward contralesional visual targets are significantly increased (Lynch and McLaren, 1989). Moreover, inactivation of LIP results in impairments occurring in the contralateral visual space, although according to one study (Li et al., 1999) the deficit consists of hypometric memory-guided saccades, while other studies report elongation in eye search time during visual target selection (Wardak et al., 2002), and deficits in covert attention (Wardak et al., 2004).
In humans the position of the eye modulates neural activity during pointing movements in parietal cortex. An fMRI study (DeSouza et al., 2000) reported a significant increase of activation in the right and left hemispheres when subjects fixate leftward or rightward to the pointing location, respectively, as would be predicted by the over-representation of contralateral eye position space observed in this study.
It is interesting that anisotropic distributions of directional properties of parietal neurons have also been described during higher-order mental processes. In the case of covert maze solving, the distribution of preferred path directions in area 7a is skewed toward the contralateral workspace (Crowe et al., 2004).
Neural Mechanisms Underlying the Directional Motor Components of Neglect
Although, many single-unit recording studies of PPC have shown the existence of different types of cells with a rich variety of functional properties, it often remains difficult to understand how the loss of these populations contribute to the emergence of the symptoms of neglect. Studies of IPL have either stressed the relationships between neural activity and attention orienting processes (for a review, see Colby and Goldberg, 1999) or the role of parietal cortex in intentional mechanisms leading to movement initiation (for a review, see Andersen and Buneo, 2002). So far, no neurophysiological study has directly addressed the problem of the neural basis of directional motor components of neglect. Neglect might be associated to a predominant, although not exclusive, representation of the contralateral space (Pouget and Driver, 2000; Rizzolatti et al., 2000; Behrmann and Geng, 2002), so that the loss of this representation could lead to an asymmetry in space coding, the contralesional sites being less well represented then the ipsilesional ones. This hypothesis is based on scrutiny of the functional properties of parietal neurons. When attempting to explain the motor deficit of neglect, this idea rests on the observation that parietal neural activity relates to the animal's intention to make purposeful eye or hand movement (Snyder et al., 1997). The failure of these intentional mechanisms would lead to a reluctance to initiate movement in the contralateral space. This hypothesis, although fully acceptable, does not capture the main feature of the motor disorder of neglect: its directional nature. Our results, while supporting the contention of neglect as consequence of under-representation of contralateral space due to cortical damage, in addition might explain a variety of directional-selective alterations associated to parietal neglect, concerning both eye and arm movements. In other words, the results of this study might help understanding directional motor components of neglect from a physiological perspective.
In fact, parietal patients with neglect suffer from hand and eye motor disorders. Among the former ones, the most commonly reported is directional hypokinesia (Heilman et al., 1985; Mattingley et al., 1992, 1998). This deficit is often described as characterized by a constellation of direction-specific disorders for hand movement toward contralateral space, including bradikinesia and hypometria. However, there is wide consensus that the main feature of directional hypokinesia consists of a significant elongation of hand reaction time to visual targets in the contralesional space, a disturbance reflecting difficulties in movement planning. Although directional hypokinesia has been described in patients with both frontal and parietal lesions (for a review, see Vallar, 2001), the latter seem to play a crucial role in the genesis of this disorder. The behaviour of neglect patients with inferior frontal and parietal lesions in the right hemisphere (Mattingley et al., 1992, 1998) has been contrasted during hand reaches within a task where perceptual components remained invariant. Hand reaches were performed to left visual targets from right and left starting positions relative to the movement endpoint. Both frontal and parietal patients displayed an elongation of reaction time for reaches toward contralesional targets. However, only parietal patients had a specific directional motor impairment, since they were much slower in initiating leftwards, as opposed to rightward, hand movements to targets in the left side of space.
Concerning disturbances of eye movements in parietal patients, the literature offers conflicting results. Lengthening of eye reaction time to targets in the contralesional space has been described in many studies (Girotti et al., 1983; Nagel-Leiby et al., 1990; Karnath et al., 1991; Pierrot-Deseilligny et al., 1991) based on visual reflexive saccades. In such studies, errors in amplitude, i.e. hypometric ‘staircase’ saccades, in contralesional space have been reported only by Girotti et al. (1983). On the contrary, during free exploration (Niemeier and Karnath, 2000) and visual search (Karnath et al., 1996), only non-direction-specific hypometric impairments of eye movements were found. Right parietal patients with neglect, when tested in the intact hemifield, display a direction specific hypometric impairment of leftward saccades only when these are stimulus-driven and not when they occur within a visual search task, where subjects make ‘voluntary’ saccades to potential targets (Niemeier and Karnath, 2003). Task-dependent emergence of temporal disorders of eye movements toward contralateral space in parietal patients had been described by Braun et al. (1992). Our observation of a magnification of contralateral directional space during reaction time of visually- but not memory-guided saccades is in line with this view, since it implies the existence of context-dependent representations of visuomotor space in parietal cortex.
The present study also revealed anisotropy in the representation of hand position information. However, contrary to what seen for other hand and eye signals, this representation favoured the ipsilateral, rather than the contralateral space. We have no easy explanation for this result, which can hardly be due to sampling bias, since it was observed in both monkeys from the analysis of the same populations of cells from which we derived preferred directions with marked contralateral anisotropy of their orientation during other epochs. Furthermore, this ipsilateral anisotropy of hand position signals was observed while the eyes were at the centre of the workspace. On the contrary, contralateral anisotropy was observed for eye-position signals, while the hand was at the centre. When eye and hand were both on peripheral targets, their opposite anisotropies tended to cancel out, and the resulting representation of the eye–hand directional space was not significantly different from uniformity. Thus, the apparent conflict brought about by the ipsilateral skew of hand position signals could be reconciled by noticing that, in both anisotropic distributions, the most represented situation is the one where the eye is on the right (i.e. contralateral to the hemisphere of recording) of the hand. This result therefore becomes a further expression of contralateral anisotropy, under the assumption that the observed signals refer to eye position relative to a hand-centred reference frame.
The disruption of a directional visuomotor representation favouring the contralateral space might explain the difficulties in planning and performing directional eye and hand movements observed in neglect patients. This representation seems unique to inferior parietal cortex, since in motor cortex the distribution of PDs of reach-neurons is uniform (Schwartz et al., 1988; Caminiti et al., 1990). Similarly, the areas of the parieto-occipital junction (Battaglia-Mayer et al., 2000, 2001) and dorsal premotor cortex (Caminiti et al., 1991), to which they are linked by association connections (Marconi et al., 2001), display a uniform distribution of directional eye- and hand-movement signals. This is coherent to what observed by PET studies in humans performing reaches to ispi- and contralateral visual targets (Kertzman et al., 1997). Interestingly, unilateral superior parietal lesions result in optic ataxia, which is characterized by a reaching disorder to visual targets (Perenin and Vighetto, 1988) that, when damage is on the right or left hemisphere, affects reaches of either hand in the contralateral visual field (‘visual field effect’), while in left-damaged patients, in addition, reaching with the right hand is inaccurate in the ispilesional field as well (‘hand effect’). Thus, in optic ataxia the reaching disorder does not show the strict directional polarity described for directional hypokinesia.
Finally, it is worth stressing that network models of neglect rely heavily on anisotropy of spatial representations (Pouget and Sejnowski, 1997; Pouget and Driver, 2000). The assumption is that a gradient of spatial representation is embedded in the activity of neurons in inferior parietal cortex, such that the lesion in one hemisphere results in lack of information concerning the contralesional space. This impairs the performance of the neural net model across different tasks, which might include those referring to motor plans toward contralesional space. The results of our study can offer neurophysiological underpinnings and new material to theoretical models of neglect.
This study was supported by funds from MIUR of Italy.
1Dipartimento di Fisiologia umana e Farmacologia, Università di Roma ‘La Sapienza’, Piazzale A. Moro 5, 00185 Rome, Italy and 2Dottorato di Ricerca in Neurofisiologia, Università di Roma ‘La Sapienza’, Piazzale A. Moro 5, 00185 Rome, Italy