Participants

The participants tested in the current study were one patient with dorsal visual stream damage (IG, age 44) and six healthy age-matched controls (three females), mean age 39 ± 9. All participants were tested using their dominant, right hand (handedness score greater than +0.50) [48] and had experience with a computer mouse and laptop touchpad.

Patient details

At the time of testing, IG was a 44 year old woman who suffered from an ischemic stroke related to acute vasospastic angiopathy in the posterior cerebral arteries established with an angiogram. Magnetic resonance imaging revealed a hyperintense signal on T2 sequencing that was fairly symmetrically, located in the posterior parietal and upper and lateral occipital cortico-subcortical regions (Figure 1C). Reconstruction of the lesion [49] indicated that it involved mainly Brodmann’s areas 19, 18, 7, a limited part of area 39. This therefore included the parieto-occipital sulcus and the caudal part of the intraparietal sulcus. IG was given a set of standard clinical tests involving visual field topography (Goldman perimetry), sensory stimulation tests (visual and tactile extinction), neurological evaluation of reflexes and muscle tone and joint movements. Visual fields showed a partial right inferior homonymous quadrantanopia (visual scotoma). The patient did not exhibit any purely motor or somatosensory deficit, any sign of sensory extinction or any sign of neglect during conventional testing (on standard line bisection, star cancellation and drawing tasks) but she demonstrated bilateral optic ataxia [24,50]. OA patients typically display in impaired online peripheral guidance of a limb [27,31] with improved accuracy when given more processing time as in delayed reaching [51], along with preserved low-level visual and motor function [34-36]. Reaching and grasping inaccuracy predominated for her right hand in her right peripheral hemifield. However, visually elicited hand movements were generally accurate when performed in foveal vision. Note that IG initially showed simultanagnosia, which prevented her to see the target and her hand at the same time, which could lead to a contradictory result [52]. In the present experiment, however, IG was tested when the bilateral optic ataxia persisted without associated clinical simultanagnosia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Task procedure, anatomical MRI scan slices.

(A) Schematic drawing of the standard center-out reaching movement (C, coronal) and the two basic manipulations: spatial plane (H, horizontal; S, para-sagittal) and cursor feedback rotation (C180°/90°). The visual stimuli were always presented on the vertical monitor. The gray circles denote the cued position before the movement. Light eye and hand symbols denote starting positions. Practice trials were performed before each condition (presented in randomized order) until it was reported that the task was sufficiently familiar for testing to begin. All conditions were performed both head-fixed and head-free in addition to a control eye only (head-fixed) and a control gaze only (head-free) condition. (B) Schematic of horizontal (black) and diagonal (gray) target locations. Note that targets in lower hemifield were not tested to avoid IG's scotoma. (C) Axial T2-weighted MRI scans of IG's cortex revealed fairly symmetrical lesions located in the posterior parietal and upper and lateral occipital cortico-subcortical regions (mainly BA 19, 18 and 7, and a limited part of area 39 and the IPS).

https://doi.org/10.1371/journal.pone.0086138.g001

Experimental procedure

Participants sat in a darkened room in front of a computer monitor placed 41 cm away, aligned with their mid-sagittal plane. They made sliding finger movements over a touch sensitive screen (Keytec Magic Screen: Model KTMT-1315: Sampling rate: 100 Hz) in order to displace a cursor from a central target to one of two horizontal (right or left), or to one of two diagonal (45° rotated from a vertical line – approximately top-right or approximately top-left) targets. Importantly, contrary to the horizontal targets which are oriented directly perpendicular to the dimly lit computer monitor border, the diagonal targets were not oriented directly towards any helpful allocentric cues such as the corner of the computer monitor. Following a two second delay, one of the peripheral visual targets (16mm in diameter) was presented 110 mm (15° visual angle) from the central target (25mm in diameter), always on the vertical monitor. Participants were trained to move the cursor as accurately and quickly as possible across the touch screen into the target, and were encouraged to maintain a consistent initial arm orientation for the different task conditions of the experiment. In addition, all participants were instructed to perform a saccade towards the peripherally-cued visual target and to maintain fixation of the target until the end of the trial. Eye movements were monitored at 250 Hz (right eye, Cambridge Systems and EyeLink II). The viewing space was calibrated using a nine-point calibration and drift correction was applied between each condition.

Figure 1A displays a schematic of all of the experimental conditions. The participants performed a single standard condition and four decoupled conditions. There were two ways in which the visual target could be decoupled from the required hand movement: a change in the spatial plane of the hand movement or a change in the motion of the viewed cursor relative to the motion of the hand (rotated visual feedback). In the standard condition and the rotated visual feedback conditions, the touch screen was placed over the computer monitor (C, in a coronal plane). The touch screen was also placed in two other spatial planes: horizontal (H, lying flat in front of the participant aligned with their midline) and para-sagittal (S, affixed to a custom mount in line with the subject’s right shoulder). The para-sagittal plane was chosen to ensure that control of a spatial plane dissociation was not aided by previously learned rules involving a computer mouse and/or laptop touchpad (such as in H). Thus, the horizontal and para-sagittal planes represented two decoupled conditions, since the visual targets were always presented on the computer monitor. Within the coronal plane (where the touch screen was placed on the monitor), cursor feedback rotation was altered either 180° (e.g. move hand leftward to displace cursor rightward) or 90° clockwise (e.g. move hand upward to displace cursor rightward). Thus the cursor feedback rotations in the plane of the computer monitor represented the other two decoupled conditions, C90° and C180°. The C90° condition was used to present a situation where a simple rule could be implemented for the horizontal targets (e.g. right = down), but a more implicit mental rotation of the cursor was required for the diagonal targets. In order to test for effects of head restraint, each condition was performed with the head being restrained using a chin rest (head-fixed), and with the head not restrained (head-free).

In order to ensure equal understanding of the transformational rules applied in each condition, all participants were trained prior to each randomly assigned block until each subject reported that they were adequately prepared. IG felt cognitively prepared following training for a total of six trials for C, 12 for H, 27 for S, 30 for C180°, and 40 for C90°. The control subjects reported to adequately prepared following training for a mean total of 12.2 trials for C, 22.6 for H, 30.6 for S, 21.2 for C180°, and 37.7 for C90°. Following training, each participant performed 40 trials (20 head-fixed trials, 20 head-free trials) in each of the five experimental conditions (Figure 1A). In addition, in order to ensure proper oculomotor control in the patient, participants performed a sixth condition in which all participants performed 40 saccades (20 eyes-only trials, 20 gaze-only trials) towards the peripherally-cued targets without any hand movements.

For the head-free conditions, the relative changes in absolute translation and roll of the head were extrapolated from the change in position of an infrared camera positioned on the middle of the forehead (Cambridge Systems and EyeLink II) relative to four infrared calibration points. These relative changes in translation and roll were verified with video from a head-mounted video camera. In order to emulate a natural environment, all subjects were instructed to look at the visual target (i.e. foveal acquisition), but were not restricted to a certain eye path. In the darkened room, the border of the computer monitor and the hand were still visible with peripheral vision.

Data Analyses

We calculated an index reflecting the overall performance decrements observed during eye-hand decoupling for patient IG relative to the control group. The index was computed as the mean number of standard deviation units IG differed from the control group (i.e. effect size; ES) for the 17 hand and eye timing, trajectory and error variables tested (not including summary variables; see below for details). ES was calculated for each trial as a change in each decoupled condition (H, S, C180°, and C90°) relative to the standard condition for each participant, and for each target type (horizontal versus diagonal). Importantly, a positive effect size represents an increased impairment for IG relative to the controls. For example, a slower peak velocity of the hand movement for IG relative to the controls would be calculated as a positive ES.

Trials were only included in the hand and eye movement analyses if they were successfully completed within a maximum of ten seconds and performed without an initial hand direction error (DE). A DE was quantified as a hand movement that deviated greater than 45° to either side of a straight line between the central and peripheral target for three consecutive time bins occurring in the first half of the ballistic movement. Although trials including these initial miscalculations were not included in further analyses, the number of DE were calculated for each participant in a separate analysis. For each DE, we also calculated the time to recovery (TTR), which was recorded from the time from the inaccurate hand movement onset (see below) until the time point in which the trajectory was reversed towards the correct target location.

The individual hand movement data were first low-pass Butterworth reverse filtered at 10 Hz (Matlab, Mathworks Inc.). Hand movement timing was analyzed whereby hand reaction time (HRT) began when the peripheral target was presented and ended at movement onset. Hand movement onsets were scored as the point at which in which the tangential velocity exceeded 10% of its peak using a custom-written computer algorithm. The hand ballistic movement time (HMT) for all conditions began from the hand movement onset and ended at the first point in which the movement slowed to 10% peak velocity. The automatically scored onset and offset was verified visually for each trial (before any overlapping corrective movements). In order to quantify the timing for corrective movements, we analyzed corrective movement time (CMT), which began at the end of HMT (10% peak velocity) of a given trial and ended when the cursor entered the perimeter of the peripheral target (trial completion). For summary purposes, we also report total movement time (TMT), which began at hand movement onset and ended when the cursor entered the perimeter of the peripheral target. Peak velocity was recorded as the maximum tangential change in resultant x and y position over time between movement onset and when the cursor entered the perimeter of the peripheral target. Path linearity was measured using hand movement paths (path length), which were recorded as the distance travelled from movement onset to when the cursor entered the perimeter of the peripheral target. In addition, the absolute angle (in degrees) of the vector from the starting point to the point of the trajectory that corresponds to the maximum velocity relative to a straight line between the central and the peripheral target was recorded for each trial (angle at peak velocity). Hand movement accuracy parameters were determined from the participant’s mean movement endpoints for each target location and analyzed separately for distance errors (on-axis CE) and for direction errors (off-axis CE). Endpoint precision (variable error, VE) was determined by the distance of the endpoints of the individual movements from their mean movements. For summary purposes, we also report absolute error (AE), which was determined as the absolute value of each participant's mean movement endpoint errors for each target location.

Eye position data were first low-pass Butterworth reverse filtered at 50 Hz (Matlab, Mathworks Inc.) and were drift corrected prior to each trial. Eye movement timing was analyzed whereby eye reaction time (ERT) began when the peripheral target was presented and ended at saccadic onset. Eye movement onsets were scored as the point at which the resultant of the x and y trajectories exceeded 10% of the peak velocity. Eye movement time began at saccade onset and ended when the pupil entered the perimeter of the peripheral target.

Eye scan paths were recorded in order to observe the un-restricted eye movement behavior when the hand was spatially decoupled from gaze direction. The eye scan paths were only analyzed for a given trial if the corresponding hand movement trial was successfully completed. Each sampled data point obtained during the experiment that was registered as a blink was interpolated off-line using data obtained from the nearest accurate measurement before and after the point. Blinks were detected from a transient reduction in the pupil size measurement, provided by the eye tracking system. In order to be able to identify saccade-related errors, eye scan path data were recorded from eye movement onset until hand movement onset (early errors: “priming”) and from hand movement onset until entrance of peripheral target (late errors: “online updating”). The saccade-related errors were placed into three categories: 1) steps 2) look-backs, and 3) hand-biased mis-saccades (HBMS). Saccade-related errors were only coded if they occurred greater than 10% (11 mm) of a full saccade (from central to peripheral target) from the target border to ensure we were not enumerating eye movements within the target. The resulting errors were categorized as steps if an eye movement trajectory continued for at least 100 ms. Hypometric steps were defined as brief saccadic pauses occurring before reaching the peripheral target, while hypermetric steps were recorded when these small saccadic pauses occurred beyond the peripheral target towards the border of the computer monitor. Look-backs were counted when subjects reversed eye direction (towards the cursor) a minimum of 20% (22 mm) of the total amplitude from the central to peripheral target, holding at least 100 ms. HBMS were recorded if the initial and/or final saccadic endpoint was biased (greater than 10% of total distance from central to peripheral target) towards the direction of the hand during the decoupled conditions.

Statistical analyses

In order to determine if successful learning occurred following training in each task (albeit not necessarily complete visuomotor adaptation), initial paired t-tests were performed for each participant between the first five trials and the last five trials performed for each condition and each target type. After confirming a performance plateau for all participants, the data from the individual patient and the control group were analyzed separately. To screen for the effects of head movement (head-fixed versus head-free) on each condition, we initially conducted three-way repeated measures ANOVAs with condition, target type (horizontal versus diagonal targets), and head movement as within-subject factors on the control group. For patient IG, we initially conducted fixed-effect intra-subject ANOVAs also with condition, target type and head movement as within-subject factors. No condition × head-movement interactions were observed following either statistical test. Therefore, all further analyses were pooled across head-movement conditions for each task condition. In addition, we also initially screened for an effect of timing (i.e. "priming" versus "online updating") depending on condition and target type. Overall, since IG’s baseline data (i.e. direct visuomotor control) did not differ from the control group for any dependent variable (p′>0.05; see explanation for modified t-test performed below), all further analyses focused on the eye-hand “decoupling” (i.e. decoupled – coupled) aspect of non-standard visuomotor control. Specifically, IG’s data was presented as the mean change in her “decoupled” relative to her “coupled” performance for each condition and each target type. For the control group, eye-hand decoupling was determined as the relative change in performance between decoupled and coupled reaching for each dependent variable and each target type. In order to control for baseline differences across control subjects, we statistically removed (i.e. covaried for) the effects of the coupled task from that of the decoupled tasks. All repeated measures ANOVA results were reported with Greenhouse-Geisser-corrected p-values, and all post hoc comparisons were corrected for multiple comparisons (Bonferroni).

Inter-group analyses between IG and the control group were performed using modified t-tests [11,53] for each condition and for each visual target type. Importantly, for accurate comparison of a case to a control group, the modified t-tests utilized in the current study adjusted the critical t-value depending on the variability (standard deviation) and group size of our control group (for details, see 53). Therefore, alpha levels for all inter-group analyses were adjusted to 5% at p′<0.05 [53]. Importantly, the corrected alpha level for each modified t-test was then corrected for multiple comparisons (Holm-Bonferroni). In addition, an index of the number of standard deviation units that IG's score differed from a randomly chosen control subject (“effect size”) was calculated for each modified t-test to demonstrate the magnitude of the difference between groups [53]. One exception was during the comparison of the change in the number of eye “look-backs” between IG and the control group. In this case, the control group did not perform such errors (mean 0 ± 0), and therefore, no statistical comparison could be performed. In order to assess the level of eye-hand coupling in both IG and our control subjects, separate correlation analyses were performed between the eye and the hand reaction times (ERT and HRT) for the both the direct and the decoupled conditions.

Effect size

We calculated an index reflecting the overall performance decrements of eye-hand decoupling for patient IG relative to the control group (i.e. change in effect size; ES, see Methods for details). We observed both condition (ANOVA, F_3,2255 = 19.6, p<0.0001) and target (ANOVA, F_1,2255= 23.0, p<0.0001) main effects for ES. Similar to previous results in unilateral OA [11], IG's performance was the most compromised relative to the control group for the visuomotor rotations (C180° and C90°). Importantly, we expand these previous findings by observing that IG’s performance was significantly more compromised towards the diagonal targets relative the horizontal targets (p<0.0001, see Figure 2). Specifically, post hoc analyses revealed a greater ES for C90° relative to H and S (p<0.0001), and C180° relative to H (p<0.0001) and S (p<0.001). For details on the dependent variables comprising the ES, see below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Effect size for decoupled conditions between IG and the control subjects.

Mean number of standard deviation units (i.e. effect size; ES) in which IG differed from control group for each dependent variable during the decoupled conditions by target type (horizontal = light gray, diagonal = dark gray). Note the increase in ES for the diagonal targets relative to the horizontal targets, and for the rotated visuomotor tasks (C180° and C90°) relative to the spatial plane dissociations ( H and S). Error bars denote 95% Confidence Intervals.

https://doi.org/10.1371/journal.pone.0086138.g002

Hand and eye movement timing

In order to assess both predictive and online updating deficits as a result of OA, we analyzed eye and hand movement preparation and execution. Figure 3 shows the overall changes in hand and eye movement timing from baseline (direct, standard visuomotor control) for all subjects across the four decoupled conditions. For details on hand movement timing differences between IG and the control group depending on condition and target location, see Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Hand and eye movement timing data for IG and the control group.

Mean change in (A) hand and (C) eye reaction times and (B) hand and (D) eye total movement times in ms for both groups for the four decoupled conditions (H = horizontal; S = para-sagittal; C180° = coronal 180°; C90° = coronal 90°) relative to standard reaching for each target type (horizontal versus diagonal). Both head-fixed and head-free conditions were pooled for each subject. Error bars denote 95% Confidence Intervals. *p′<0.05; **p′<0.01; ***p′<0.001, corrected for multiple comparisons (Holm-Bonferroni).

https://doi.org/10.1371/journal.pone.0086138.g003

Eye-hand coupling

We analyzed the impact that eye movement planning (ERT) had on hand movement planning (HRT) in both direct and decoupled situation. As such, to assess eye-hand coupling across conditions and targets for IG and the control group, we ran correlation analyses.

IG displayed the strongest eye-hand coupling (i.e. positive correlation) for C90° towards the diagonal targets (r = 0.79, p<0.0001). Eye-hand coupling was also strong during her performance of C90° towards the horizontal targets (r = 0.71, p<0.001). Moderate positive correlations were observed for direct reaching towards both horizontal (r = 0.60, p<0.01) and diagonal targets (r = 0.51, p<0.05). ERT and HRT were not correlated for either H, S, or C180° (p>0.05).

The control group displayed the strongest eye-hand coupling during direct reaching towards the diagonal targets (r = 0.57, p<0.0001). Similarly, moderate eye-hand coupling was observed during S only towards the diagonal targets (r = 0.47, p<0.0001). Moderate positive correlations were also observed for C90° towards both horizontal (r = 0.45) and diagonal (r = 0.45) targets (p<0.0001). Low positive correlations were observed for C180° towards horizontal (r = 0.24) and diagonal (r = 0.20) targets (p<0.05). No significant correlation was observed in the control group for H (p>0.05).

Hand movement endpoints and trajectories

Analyses of hand movement endpoints of the initial ballistic motor plan, as well analyses of the entire trajectory were performed to assess the integrity of the predictive motor plan and the online correction in OA. Figure 4 displays the ballistic endpoint across all conditions for IG and for a typical control subject. For examples of full hand and eye movement trajectories between diagonal and horizontal targets during decoupled eye-hand coordination, see Figure 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Individual hand endpoint ellipses for IG and a typical control subject.

Hand movement trajectories endpoints of ballistic motor plan to four peripherally-cued targets from the home target in (A) coronal (B) horizontal (C) para-sagittal (D) coronal 180° and (E) coronal 90°. Both eye and gaze conditions were pooled for all subjects. Open and filled ellipses represent 95% confidence intervals for IG and a typical control, respectively. Trajectories (green lines), start points (closed red circles), endpoints for successful trial (closed blue circles) and direction error trials (open black circles) represent IG’s data only. Circles with cross-hatching represent starting and ending target location. Note the systematic endpoint errors for IG during decoupled reaching especially when orienting towards the diagonal (off-axis) targets, and the accurate reaching for the control subject.

https://doi.org/10.1371/journal.pone.0086138.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Example trials during coronal 90° condition towards diagonal and horizontal targets for IG and a typical control subject.

Example hand (represented by the rotated cursor) and eye movement trajectories for the entire trial towards (A) diagonal and (B) horizontal targets. Note for IG (left side) the increase in path length towards diagonal target and the increase in late (online updating) look-backs (closed arrows), and relatively accurate hand movement trajectories accompanied by an early (priming) look-back (closed arrow) towards the horizontal target.

https://doi.org/10.1371/journal.pone.0086138.g005

Eye movement errors

Although the hand data for this OA patient demonstrated impaired performance, IG did complete all trials within the given time limit (10 seconds). The reason for her successful completion of each trial becomes clear when looking at the eye movement data. Although all subjects were instructed to foveate the peripherally-cued visual target, eye movements were not restricted. Similar to previous results observed in unilateral OA patients [11], several additional oculomotor errors were observed in this bilateral OA patient (see Figure 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Eye errors performed by IG and the control group.

Change in eye (A) hypometric steps (B) hypermetric steps (C) look-backs and (D) hand-biased mis-saccades for IG (circles) and controls (bars) for the four decoupled conditions (H = horizontal; S = para-sagittal; C180° = coronal 180°; C90° = coronal 90°) for each target type (horizontal = light gray, diagonal = dark gray) relative to standard reaching. Head-fixed and head-free conditions were pooled for all subjects. Error bars denote 95% Confidence Intervals. Note a marked increase in look-backs for IG during the performance of C90° towards diagonal targets. *p′<0.05; **p′<0.01; ***p′<0.001, corrected for multiple comparisons (Holm-Bonferroni).

https://doi.org/10.1371/journal.pone.0086138.g006

Head movement

In order to observe behavior in a more “natural” environment, we repeated all conditions “head-free”. During such head-free conditions, we recorded the change in cumulative head movement (translation and roll, see Methods) for decoupled reaching relative to standard reaching. IG performed larger head movements (condition main effect; ANOVA, F_3,72 = 41.7, p<0.0001) during C90° relative to the other decoupled conditions (p<0.0001). No differences were observed between conditions or target-type for the control group.

Although allowing free head movement did not improve hand or eye performance in either group, we did observe inter-group differences in the amount of head movement between IG and the controls during decoupled reaching. IG utilized more head movement than the controls did during the performance of C90°, for both horizontal (t = 14.9, p′<0.001, effect size = 16.1) and for diagonal (t = 7.2, p′<0.01, effect size = 7.7) targets.

In summary, although free head movements did not improve hand or eye movement performance in any participant, IG performed larger head movements in one of her more challenging conditions (C90°).

Strategic control versus sensorimotor recalibration of a decoupled limb in optic ataxia

Strategic control plays an integral role in cognitive-motor integration. However, cognitive rules alone are not sufficient; the incorporation of a rule into a motor plan must be complimented by online sensorimotor recalibration of a decoupled limb in space. Similar to previous reports in OA [11,51], IG displayed intact strategic control in the current study, particularly for well-categorized canonical movements. Meanwhile, her deficits were the greatest towards the non-canonical diagonal targets. These decrements were observed in the form of increased planning and execution time for the eye and the hand, hypometric reaching, and increased hand path length. Her deficits were markedly smaller for mappings which relied predominantly on explicit rule integration, suggesting an independent pathway for processing strategic control in non-standard visually-guided reaching (see below).

We observed an increase in performance difficulty for patient IG relative to the control group (i.e. ES; see Figure 2) for the visuomotor rotations relative to the spatial plane dissociations and for the diagonal relative to the horizontal targets. The overall deficit across participants for C90° in particular relative to the other conditions confirms that even an adequate comprehension of a rule is not sufficient to adapt to non-canonical off-axis situations [39-41]. Specifically, the inversion of a single axis required for the computation of C90° required increases in RT as well as eye and hand movement errors across participants. However, equal performance across targets by the control group implies a specific deficit observed for IG towards the diagonal targets. In contrast, cognitive rules were previously useful for patient IG (horizontal spatial plane dissociation) after a brief training [34]. In the current study, IG’s relatively successful performance in the horizontal plane towards horizontal targets indicates that learning of a rule is sufficient in situations with reliable allocentric cues and previous experience. Therefore, a flexible balance appears to exist when learning the rules needed for strategic control versus the gradual adaptation needed for sensorimotor recalibration during decoupled eye-hand coordination. In the current situation, although all participants had an adequate comprehension of the required rule for each task, and were performing at a plateau during the task, they were not necessarily fully adapted to the different transformational manipulations. In fact, each participant had very little difficulty switching between the randomly assigned conditions (i.e. showed no after-effects), indicating that all conditions involved a strong explicit component. However, the implicit component of each condition and target type becomes apparent when observing the deficits seen as a result of OA.

These findings suggest that an intact caudal SPL is not crucial for decoupled eye-hand movements when relying on a cognitive rule or a stereotyped motor plan (formulated via previous experience), but is integral for the realignment of decoupled vision and proprioception during novel situations or where there is no reliable rule. This latter situation likely requires the close monitoring of visual and proprioceptive information processed in this region of the brain, while the former situation could rely on intact fronto-temporal circuits for movement planning and guidance (see below).

Potential neurological substrates for decoupled visuomotor control

Although the dorsal stream of the proposed perception-action model [56] has been well accepted as a primary network for the control of “vision for action”, it has become clear that it is not entirely functionally segregated from the control of “vision for perception” [34,35,57,58]. Instead, the dorsal stream appears to contribute to the integration of cognitive visuo-perceptual skills with complex visuomotor skills [59]. As such, it has become apparent that overlapping yet distinct cortical networks exist which control the specific components involved in decoupled eye-hand coordination.

Notably, damage to a crucial node involved in the peripheral guidance of limb in space (caudal SPL) [11,29,31] appears to result in an inability to successfully integrate the two proposed streams. Such egocentric guidance of conflicting visual and proprioceptive information in peripheral space [11,33] and online updating required following target displacement [25-27,29,30,60] are primarily affected from dorsal stream damage. This suggests that the OA deficit includes an impaired integration of conscious awareness of eye-centered metrics with transient online representations of limb-centered metrics. Supporting the contribution of this brain region to non-standard visuomotor mapping, we have observed reduced firing rates in caudal SPL neurons in (intact) non-human primates performing similar eye-hand decoupled reaching tasks [61]. Taken together, we propose that the caudal SPL contributes to the required inhibition of the natural tendency to reach towards where one looks by monitoring the relationships between the behavioral goal and the location of the involved effectors in space, and communicating this information to frontal lobe structures involved in planning the biomechanical details of the specific movement.

The current data provide evidence of a functional spectrum from strategic control to sensorimotor recalibration of decoupled visuomotor control. Reaching in a well-learned canonical situation such as in the horizontal plane (H: when using a computer mouse), does not rely fully on an intact caudal SPL. Rather, the premotor cortex may receive indirect inputs from more ventral connections into the prefrontal cortex via the infero-temporal cortex or via the IPL [35], which may carry the crucial information to guide the movement. Evidence in support of these alternate connections comes from reach studies on OA patients employing a long delay between the cue and the movement [62-64]. Such connections require more processing time [65] and carry rule-based and allocentric information, which is impaired in patients such as DF with lateral occipital [66] ventral stream damage [67]. Fast, implicit guidance of a limb in peripheral space, on the other hand, relies on the combination of peripheral perception of motion with an appropriate reach vector command. It has since been suggested by [68] that the motion sensitive area MT, an area previously thought to be explicitly within the dorsal stream, and its connections with MST and IPL [69,70], may serve as an integral node in the suggested interaction between dorsal and ventral streams [35,58,71,72].

Difference vector computation in optic ataxia

Decoupled eye-hand coordination requires ongoing overt foveal monitoring of the visual target with covert peripheral visual feedback of the limb and cursor position, along with proprioceptive feedback of the decoupled limb position. If the generated motor plan has been miscalculated, a difference vector must be continually updated online to compensate accordingly. Such a miscalculation will, in the OA patient, result in hypometric reaching (towards the direction of gaze) when reaching towards extra-foveal targets [31,73,74], proprioceptive targets [33], and foveated visual targets decoupled from the moving limb [11].

Distinct functional regions have been proposed within the SPL, with segregated areas for reaching (parietal reach region; PRR) [75,76], located within the medial bank of the IPS [77], and saccades (parietal eye fields; PEF) [78], located on the lateral bank of the IPS [77,79]. According to this segregated view, depending on the location of the lesion site, the decoupled eye-hand coordination deficits seen in OA patients could result from either impaired eye-centered coding within the PEF or a breakdown of limb-centered coding within PRR. Others have suggested the impairment in OA to result from the inability to simultaneously represent spatial orientations of decoupled end effectors when guiding a limb in peripheral visual space [80]. Thus, during decoupled eye-hand coordination OA patients may have difficulty transforming eye-centered information about the visual goal into a limb-centered motor goal [26,28,80,81]. Alternatively, evidence for reaching deficits resulting from temporary deactivation of parietal area PEc in non-human primates [47], has been thought to result from a breakdown in the combination of the preferred direction of eye and hand position relative to a visual target goal into a common state within the dynamic, context-dependent global tuning field of individual parietal neurons [82,83]. Thus, without a functioning SPL, the frontal cortex may not be provided with updated accurate eye-hand position signals (see 82). This breakdown of the global tuning field could explain the altered neural outputs (spikes) [61] from SPL and inputs (local field potentials) [84] into PMdr when required to formulate and maintain an accurate difference vector during decoupled eye hand coordination.

In the present study, we observed that IG had the greatest deficits for those decoupled conditions in which she could not rely on reliable allocentric aids (such as the computer monitor border) and/or simple cognitive rules (such as hand up = cursor right). When forced to rely on online updates from conflicting proprioceptive and visual information, IG displayed impaired hand movement correction when attempting to adjust for inaccurate ballistic movements. In an attempt to compensate for impaired online updating in peripheral vision, IG overtly foveated the cursor position via additional eye movements (look-backs). Such overt foveal updating of limb or target position has been previously shown to be beneficial for predicting upcoming hand movements [26,85-87]. Additional evidence for IG’s reliance on her vision for her hand movements comes from the high correlation between her eye and hand movement planning timing. During C90°, particularly towards the diagonal targets, IG’s hand movements varied relative to her eye movements to allow for efficient eye-hand coupling.

We propose that the deficits seen in this OA patient arise as a result of a failed transformation between guiding sensory information and required limb movement when strategic control is not possible. Our findings support the involvement of caudal SPL in the monitoring of gaze, limb, and target location needed for difference vector computation in decoupled reaching, a computation required for successful visuomotor transformations. The question remains if IG’s extensive previous experience has enabled neuroplasticity for decoupled visuomotor control and therefore can be accurately generalized to other OA patients. In fact, compensatory activity has been observed for IG in both occipito-temporal and in occipito-parietal regions surrounding her lesions during both immediate and delayed extra-foveal reaching [88]. However, similar observations of a reliance on strategic control during decoupled visually-guided reaching in a newly-tested OA patient MFL [11] provide evidence that IG’s deficits result from her caudal SPL lesions. Further, IG’s deficits are in line with the documented role of the caudal SPL in coding the required difference vector during decoupled reaching [61,89,90].