A robot-based behavioural task to quantify impairments in rapid motor decisions and actions after stroke
© The Author(s). 2016
Received: 3 March 2016
Accepted: 4 October 2016
Published: 10 October 2016
Stroke can affect our ability to perform daily activities, although it can be difficult to identify the underlying functional impairment(s). Recent theories highlight the importance of sensory feedback in selecting future motor actions. This selection process can involve multiple processes to achieve a behavioural goal, including selective attention, feature/object recognition, and movement inhibition. These functions are often impaired after stroke, but existing clinical measures tend to explore these processes in isolation and without time constraints. We sought to characterize patterns of post-stroke impairments in a dynamic situation where individuals must identify and select spatial targets rapidly in a motor task engaging both arms. Impairments in generating rapid motor decisions and actions could guide functional rehabilitation targets, and identify potential of individuals to perform daily activities such as driving.
Subjects were assessed in a robotic exoskeleton. Subjects used virtual paddles attached to their hands to hit away 200 virtual target objects falling towards them while avoiding 100 virtual distractors. The inclusion of distractor objects required subjects to rapidly assess objects located across the workspace and make motor decisions about which objects to hit.
As many as 78 % of the 157 subjects with subacute stroke had impairments in individual global, spatial, temporal, or hand-specific task parameters relative to the 95 % performance bounds for 309 non-disabled control subjects. Subjects with stroke and neglect (Behavioural Inattention Test score <130; n = 28) were more often impaired in task parameters than other subjects with stroke. Approximately half of subjects with stroke hit proportionally more distractor objects than 95 % of controls, suggesting they had difficulty in attending to and selecting appropriate objects. This impairment was observed for affected and unaffected limbs including some whose motor performance was comparable to controls. The proportion of distractors hit also significantly correlated with the Montreal Cognitive Assessment scores for subjects with stroke (r s < = − 0.48, P < 10−9).
A simple robot-based task identified that many subjects with stroke have impairments in the rapid selection and generation of motor responses to task specific spatial goals in the workspace.
KeywordsStroke Assessment Cognitive impairments Attention Inhibition Neglect
Moving and interacting in the world requires rapid processing of the visual environment to identify potential motor goals, select a movement and finally move in a timely manner. For example, when packing groceries, we must decide where to put items based on their shape, size, fragility and other features. The selection, planning and execution of motor actions must be done rapidly to keep pace with the flow of groceries from the cashier.
There is growing evidence that sensory feedback is rapidly integrated into motor decisions [1–3]. Sensory feedback is integrated with higher-level behavioural goals to make rapid decisions on how to move and interact in the environment. Selective attention refines spatial representations of the environment into potential movement targets [1, 4]. The choice between these internal representations is then based on ‘decisional factors’ . One such factor is the recognition of combinations of visual features and their behavioural relevance [1, 5, 6]. Thus, the sensorimotor system rapidly integrates information on the environment to guide motor decisions.
Another important aspect of voluntary motor control is the ability to inhibit a motor action . When instructed, it is very automatic to simply reach towards spatial targets as they appear in the workspace . In contrast, it can be hard to avoid reaching towards a target when instructed to move in the opposite direction. In this anti-reach condition, subjects can make erroneous initial motor responses to the spatial goal and are delayed in moving in the opposite direction. This task requires the voluntary override of an automatic response to reach towards the target and involves many brain areas including frontal and parietal cortex [7, 9–12]. This ability can be impaired in persons with stroke , mild cognitive impairment , Alzheimer’s Disease , and a history of concussion , Thus, successful voluntary motor control involves processing sensory feedback not only to select motor actions but also to avoid making others.
Post-stroke disability stems from a variety of motor, sensory, and/or cognitive deficits . The ability to pack groceries described above highlights that impairments in these functional tasks may reflect not only motor impairments but also cognitive impairments. When driving, one must quickly decide on actions to apply pressure to the brake or accelerator pedals, or turn the wheel based on information from street signs, traffic signals, other traffic, and pedestrians. However, neuropsychological tests or cognitive screening tools generally separate motor and cognitive assessments – the latter often requiring verbal or written responses – and typically do not impose time limits to perform the tasks [17, 18]. Few neuropsychological assessments focus on rapid motor decisions beyond simple reaction time tests [18, 19], or timed cognitive tasks such as trail making , even though complex and time sensitive demands are often required for everyday activities.
Furthermore, many standard assessments of post-stroke functioning have problems of subjectivity, coarse ordinal scales, criteria-based scoring, and lack of responsiveness (including floor and ceiling effects) . Thus, we developed a novel approach of using a robotic assessment to provide objective, continuous measures of performance that are compared to a normative model of healthy control performance.
We recently used an object hit task to quantify simultaneous upper limb bimanual sensorimotor performance . Although this task quantified rapid motor skills, decisional processes required to perform the task were limited to identifying the trajectory of an object and selecting a limb to hit the object. As all objects were targets in this task, it did not require cognitive processes related to attending to object qualities (rather than just spatial location) to select a motor action nor require inhibiting inappropriate motor responses. These processes can be impaired following stroke [9, 23, 24].
The goal of the present study was to develop a task that examined rapid motor skills with both arms that also required greater cognitive processing. We developed a variant of the object hit task  by requiring subjects to hit 2 possible targets while avoiding all other objects in the workspace. We hypothesize that individual subjects with stroke will be impaired in enacting or inhibiting a motor response to a potential target based on sensory feedback of the object’s features and their relevance to the ongoing task. The performance of subjects with stroke was compared to a large cohort of non-disabled control subjects.
Clinically, knowledge of impairments in these more complex visuomotor skills can guide novel rehabilitation strategies to regain the ability to rapidly process sensory information for motor actions. As well, it may help to identify if individuals should return to more complex daily activities such as driving.
Participants included patients recruited from Providence Care (St. Mary’s of the Lake Hospital, Kingston, ON), the Dr. Vernon Fanning Centre and Foothills Hospital (Calgary, AB). Prospective subjects were excluded if they had other significant neurologic diagnoses (e.g., Parkinson’s disease), acute medical illness, and/or ongoing upper extremity musculoskeletal injuries. Subjects were also excluded if they appeared fatigued, reported pain associated with attempting robotic assessments or reported pain during clinical testing on strength or range of motion that would be relevant to the robotic task. Non-disabled control subjects were recruited from the Kingston, ON and Calgary, AB communities. This study was approved by the Queen’s University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board (#ANAT-024-05) and the University of Calgary’s Conjoint Health Research Ethics Board (#22123) and subjects provided informed consent.
Details of the robotic set-up have been reported previously [25, 26]. Briefly, the behavioural task was performed using a bimanual exoskeleton robot which measures limb motion (KINARM, BKIN Technologies Ltd, Kingston, ON, Canada). Participants sat in a modified wheelchair base, and their arms were fitted in supports permitting movement in the horizontal plane. Arm supports were adjusted such that the robot’s linkages aligned with the subject’s elbows and shoulders. Subjects received visual feedback from a virtual reality system which displayed fingertip position and virtual objects in the same plane as arm motion via a two-way mirror. Direct vision of the hands and arms was occluded.
One of 6 task variants were used with varying shapes designated as targets and distractors. Target pairs had similar width but were always different heights and different classes of shapes (Fig. 1a). Distractors consisted of the remaining unused shapes (shapes used as targets in other task variants), as well as two wider shapes.
Every effort was made to ensure subjects understood the task instructions. Operators usually obtained verbal confirmation that they understood which targets to hit when showing the target objects before starting the task. Reminders to hit the specific target shapes and avoid all others would be given early in the task, especially if there seemed to be confusion with similar distractor shapes (for example tall target rectangle vs. wide rectangle distractor). As well, targets hit by a paddle were knocked away and haptic feedback of the contact was provided by the robot , whereas distractors simply passed through the paddle to provide immediate feedback that it was a distractor.
Data were analyzed using MATLAB (Mathworks Inc., Massachusetts). Hand speed was filtered using a sixth-order double-pass Butterworth filter (cutoff frequency 10 Hz).
We used 14 metrics to quantify task performance in order to characterize a diverse range of sensory, motor and cognitive functions examined in this task. Most of the parameters paralleled the metrics in our previously published object hit task , a simpler version of this task in which there were no distractors. As well, a few parameters were added or modified to capture the addition of distractors in the present task.
Targets hit: The number of target objects hit away from the body.
Distractors hit: The number of distractor objects hit.
Objects hit: The number of objects hit (target hits + distractor hits).
Distractor proportion: Distractors hit divided by objects hit.
Object processing rate (objects/second): The rate of correctly processed objects (number of targets hit + distractors missed per second) at 80 % of task completion. The rate of correctly processed objects was determined at every time step (every 0.005 s) from the time the first object was hit or left the screen to the time the 240th object (80 % of task complete) was hit or left the screen. To filter this signal, we convolved the rate with a Gaussian window (MATLAB function normpdf). From this rate signal, the optimal growth curve (y = <max height > *(1-exp(−(<curvature>)* < data>))) [27, 28] for the data was calculated and the value of this curve when 80 % of the task was completed was used to approximate the maximal object processing rate for each individual subject. The rate was taken at 80 % of task complete so that performance was at or near maximum, but not at 100 % as the ratio of distractor object dropping statistically increased at the end of task. This is because there is always a 66 % chance of dropping a target object, but objects are sampled without replacement, leading to the statistical scenario of running out of target objects and only being able to drop distractor objects from a given bin at the end of the task.
All other parameters were defined in the same way as the object hit task .
Miss bias: Spatial position quantifying the extent to which the number of target misses deviates from being equally distributed on either side. Computed as sum of target misses in each bin (m), multiplied by the bin position (x), and then divided by the total number of target misses (sum(mx)/sum(m)). Given that the centre of the bins is x = 0, the greater the mean location of misses deviates from 0, the more misses were on the left or right side of the workspace (dependent on whether the miss bias is negative or positive, respectively).
Hand transition: Spatial transition point in subject’s hand preference for hitting targets. This is the mean of the right hand’s and the left hand’s weighted means of their respective target hit distributions. The weighted mean of each hand only includes target hits by that hand in bins where both hands have been used to hit targets (overlapping bins) and one additional bin beside the overlapping bins (where that hand has been used to hit targets). If bins do not overlap, only the leftmost bin with target hits from the right hand and the rightmost bin with target hits from the left hand are used in the weighted means.
Median error (% of targets): Point in time when subjects missed half of the target objects that they missed over the entire task.
Movement Area: The areas of space used by each hand during the task. Computed as the area of the convex hull- a complex polygon which captures the boundaries of the movement trajectories of each hand [22, 29]. Calculated for each hand separately.
Hand speed: The average hand speed calculated from each time step (5 ms) over the course of the task. Calculated for each hand separately.
Hand bias hits: The difference between the number of target hits with the right hand and the number of target hits with the left hand divided by the total number of target hits.
Hand selection overlap: The number of times successive target hits from a given bin were with different hands divided by the total number of target hits.
Hand movement bias area: Difference in movement area of the right and left hands divided by the sum of the movement area of the right and left hands.
Hand bias speed: The difference in mean hand speed of the right and left hands divided by the sum of the mean hand speed of the right and left hands.
Performance of control subjects was analyzed for any effects of age, sex, or handedness. Control values were age-regressed and Box Cox transforms were used to normalize control distributions when necessary [30, 31]. Control parameter values were then assessed for any effect of sex or handedness, and values subdivided into respective categories if effects were significant, and age regressed and Box Cox transformed again if necessary. All parameter values were converted to z-scores of the model to allow for comparison across all subjects (because age, sex, and handedness are now accounted for in the model). Individual subjects with stroke were defined as having impaired performance on a task parameter, when their z-score was >1.65 or < −1.65 for one tailed tests, or > |1.96| for two tailed tests.
A subset of subjects was assessed a second time by a different operator within 7 days of their initial assessment. An intraclass correlation was used to determine interrater reliability (significant if P < 0.05, acceptable if ICC > 0.8).
Subjects with stroke were evaluated by a trained physician, physiotherapist, or occupational therapist using a number of standardized clinical assessments. Both arms were assessed using the Chedoke-McMaster Stroke Assessment (CMSA)  to determine arm function. The CMSA is based on Brunnstrom’s stages of motor recovery post-stroke . Subjects were broadly categorized as “Left-Affected” (LA) or “Right-Affected” (RA) depending on the clinically most affected side of the body. Elbow flexor spasticity was measured by the Modified Ashworth Scale, which categorizes the amount of resistance produced by the arm in response to passively moving it through its range of motion . Functional abilities were measured with the Functional Independence Measure (FIM), which has both a motor and cognitive component . The conventional subtests of the Behavioural Inattention Test (BIT) were used to screen for deficits in spatial attention . Subjects with stroke who scored <130 on the BIT were defined as having visuospatial neglect and were analyzed separately (neglect subjects). Subjects were also screened for cognitive deficits using the Montreal Cognitive Assessment (MoCA) . The handedness of controls and subjects with stroke was determined by the Modified Edinburgh Handedness test .
Subject demographics and clinical information
Demographic information of subjects included in the experiment
Stroke (n = 157)
Controls (n = 309)
Age in yearsa
64 (25–90; 13.8)
53 (18–93; 19.4)
time since strokea
11 (1–49; 11.2) days
affected arm (L/R)
FIM- motor subscorecd
FIM- total scorecd
lesion location (number of subjects)
C + SC
Cb + Br
LA (n = 94)
RA (n = 63)
BIT < 130
visual field deficit
CMSA- arm subscoref
[5 13 14 4 19 14 24] g
[5 8 9 5 13 7 15] h
[0 0 0 0 8 22 63] g
[0 0 0 0 1 14 47] h
[9 3 10 8 30 16 16] i
[4 7 6 7 12 11 15] h
[0 0 0 0 1 32 59] i
[0 0 0 0 2 18 42] h
[76 10 1 6 0 0]g
[61 1 0 0 0 0]g
[91 2 0 0 0 0]h
[48 8 4 0 2 0]h
Figure 1b-d displays the distribution of target and distractor hits and misses for a control subject and two subjects with stroke. The control subject (Fig. 1b) was very effective in hitting targets (136/200) and avoiding distractors (94/100). Control subjects gradually missed more targets, especially lateral ones, as task difficulty increased. The RA subject with stroke (Fig. 1c) hit fewer targets (117/200) and more distractors (47/100) than the control. The LA neglect subject (Fig. 1d) hit even fewer targets (83/200) and a similar amount of distractors (40/100). This subject also hit very few objects with their left hand and very few on the left side of the workspace.
Impairments identified using the robot-based task
Task performance, interrater reliability, and clinical correlations. Task parameter sensitivity is defined by the corresponding z-score cutoff range
Subjects with stroke (% impaired)
Parameters (normative models)
BIT > =130
BIT < 130
Intraclass correlation (P)
Object processing rate
Spatial and temporal performance
Hand specific performance
Hand bias (hits)
Hand selection overlap
Total hand bias area
Hand bias speed
Total hand area
95 % (>5)
Distractor proportion identified more individual subjects with stroke as impaired (39 % LA, 51 % RA, 79 % neglect; see Fig. 2b and Table 2) compared to distractors hit (15 % of subjects with stroke impaired). Object processing rate was also a parameter that identified a large proportion of subjects with stroke (Fig. 2c): controls mostly had a processing rate between 1.5 to 3 objects per second, whereas the object processing rate of most subjects with stroke was below 2.
Subjects with stroke who hit fewer objects also tended to hit a higher proportion of distractors (Spearman correlation; controls: r s = 0.16, P = 0.006; subjects with stroke: r s = −.33, P = 3×10−5). Twenty nine percent of subjects with stroke displayed impairments in both object hits and distractor proportion (Fig. 2d, upper left quadrant). In contrast, 29 % of subjects with stroke had impairments in only object hits (Fig. 2d, lower left quadrant) and 16 % had impairments in only distractor proportion (Fig. 2d, upper right quadrant). All neglect subjects were impaired in at least one of these two parameters, and 64 % were impaired in both parameters.
Motor and distractor-related impairments in performance of the affected arm commonly co-occurred. Of the subjects whose affected arm was impaired in distractor proportion, the same arm was usually also impaired in objects hit (97 % LA, 81 % RA) and/or hand speed (85 % LA, 63 % RA) (Fig. 3c). Distractor proportion was also negatively correlated with the number of object hits by the affected arm of subjects with stroke (r s = −0.58, P < 10−10). Although this would be expected, a negative correlation was not observed in the non-dominant arm of controls (r s = 0.15, P = 0.008).
In contrast, this coupling of motor and distractor-related impairments was less common for the unaffected arm. For the subjects with stroke who had impaired distractor proportion with their unaffected arm, the majority did not have impaired object hits (47 % LA, 82 % RA) or hand speed (58 % LA, 75 % RA) with the same arm (Fig. 3d). The correlation between distractor proportion and objects hit was weaker for unaffected arm performance of subjects with stroke (r s = −0.15, P = 0.07). A negative correlation was not observed for the dominant arm of controls (r s = 0.18, P = 0.001).
Neglect subjects who had impaired distractor proportion with their affected arm were usually impaired in hitting objects and/or hand speed with the same arm (100 and 92 % impaired in both parameters, respectively). Impairments in distractor proportion for their unaffected arm were less likely to co-occur with impairments in objects hit and/or hand speed with that arm (36 and 36 % impaired in both parameters, respectively).
We aggregated the number of task parameters that each subject was impaired in (Table 2). Most subjects with stroke (82 %) were impaired in more task parameters than 95 % of controls (>5 parameters). This included all but one subject with neglect.
Correlations with standard clinical assessments
The relationship between task performance of subjects with stroke and Functional Independence Measure (FIM), Montreal Cognitive Assessment (MoCA), and Behavioural Inattention Test (BIT) scores is shown by the corresponding Spearman correlations
Parameters (normative models)
Spearman correlations r s (P)
Object processing rate
Spatial and temporal performance
Hand specific performance
Hand bias (hits)
Hand selection overlap
Total hand bias area
Hand bias speed
Total hand area
Spasticity, as measured by Modified Ashworth, showed modest correlations with a few task parameters: the number of objects hit with the right hand (r s = −0.32, P = 4.2×10−5), movement area with the right hand (r s = −0.34, P = 1.6×10−5), and hand speed with the right hand (r s = −0.33, P = 2.6×10−5).
The interrater reliability of the task parameters is shown in Table 2 for subjects (13 controls and 10 subjects with stroke) assessed in the task twice. Intraclass correlation coefficients were often high: ICC > =0.8, for 76 % of parameters. Lower reliability values were generally associated with parameters that identified fewer subjects with stroke as impaired and thus had a relatively small range of values across the control and stroke populations. In contrast, higher reliability values tended to be associated with parameters that identified more subjects with stroke as impaired and thus tended to have a larger range of values across the inter-rater sample.
The current study quantified impairments in stroke survivors to rapidly hit certain objects (targets) while avoiding all other objects (distractors). Up to 78 % of subjects with stroke had impairments in individual global, spatial, temporal, or hand-specific task parameters. The task instructions were simple, minimizing the impact of comorbid language impairment . The task was completed in ~3 min yet provided a wide range of information related to sensorimotor and cognitive function. Most parameters had high inter-rater reliability providing an objective approach to measure impairments and track recovery.
The object hit and avoid task is a variant of an object hit task in which subjects had to rapidly locate and hit all objects moving in the workspace . The present task extended this approach by requiring the subject to select amongst many options when moving and interacting in the environment. Total objects hit quantified each subject’s ability to make rapid motor actions, regardless of whether they hit the correct objects or not. Subjects with stroke almost always hit fewer objects with their more affected side, and this arm’s performance was more correlated with FIM scores than the unaffected side. Thus, the reduction and asymmetry of the ability to make rapid motor actions is quantitatively measured by the object hit and avoid task, and may have importance in the ability to complete activities of daily living.
We used a large number of parameters to quantify a broad range of sensory, motor and cognitive functions necessary to perform this task. For healthy subjects, some of these measures were highly correlated, but nevertheless captured different functions. For instance, the correlation between target hits and object hits was very strong for controls (r = 0.81). The reason why both parameters were measured rather than choosing only one was because it was important to differentiate between the ability to make fast and accurate movements, and the ability to make correct motor decisions om whether an object was a correct reach target or not. Thus, these metrics represent different domains of performance. Furthermore, subjects with stroke do not necessarily follow this typical pattern of performance. As shown in Fig. 2d, some subjects with stroke hit a high proportion of distractors and others do not, showing the value of each parameter to identify different impairments that do not necessarily co-occur in some individuals with stroke.
The inclusion of both target and distractor objects in the current task added an additional cognitive load to the previous object hit task. This is important as many different cognitive processes are necessary to perform daily activities, and their impairment after stroke is a significant cause of disability . The present object hit and avoid task focused on a few key processes.
First, demands on the attentional system are high in a visual search task, as it requires differentiating target and distractor stimuli . Rapid parallel processing of the entire visual workspace can be employed to find a target amongst many distractors with minimal effort if the target has a unique feature separate from distractors that makes it ‘pop out’. In contrast, focused attention is required to serially analyze each stimulus if the target can only be differentiated from the distractors by a conjunction of features. The greater attentional demands required for a conjunction versus a feature visual search task results in greater reaction time for both controls and subjects with stroke who do not have visuospatial neglect . Subjects with visuospatial neglect also show significantly increased times to detect targets in a conjunction search task (regardless of which side of the workspace was tested), when compared to the performance of controls and subjects with stroke. The object hit and avoid task is representative of a conjunctive visual search as targets could only be differentiated from distractors by attending to the geometry (circular, three- or four-sided) and relative dimensions (tall, wide or equal) of each object (see Methods-Behavioural Task). Correspondingly, BIT scores correlated with many individual task parameters, as well as the total number of parameters impaired. Although correlations were weak to moderate, all were in the expected direction: greater task impairment associated with greater clinical impairment.
In the current study, participants are required to either enact a reach toward the target, or actively avoid hitting a distractor. Despite visual feedback, haptic feedback, and initial reminders on the need to hit only two types of objects and avoid the rest, over half of the subjects with stroke hit a greater proportion of distractors than 95 % of controls. Subjects with stroke were twice as likely to be impaired in this parameter if they also had neglect.
The ability to inhibit a motor action is an important cognitive function of voluntary motor behaviour . Motor decisional processes mediate the initiation of an automatic motor response to a new stimulus with the voluntary response required by the task . This ability to inhibit stimulus-driven and enact task-driven motor responses can be measured by eye movements in the anti-saccade task  and arm movements in an anti-pointing task . In both tasks, subjects must inhibit a movement to the appearance of a visual stimulus and move to the equal and opposite location. Subjects with stroke having damage to frontal lobes have been shown to make erroneous saccades towards a stimulus in an anti-saccade task . Subjects with stroke and visual neglect show greater endpoint errors and longer reaction times in an anti-pointing condition (on both sides of space) than controls or subjects with stroke who do not have visuospatial nelgect . Distractor proportion in the current study correlated with BIT scores just as anti-pointing impairments correlated with the severity of neglect.
The assessment of rapid visuomotor skills post-stroke has potentially useful applications when rehabilitation goals are to regain high function. The object hit and avoid task may be very predictive of the ability to drive, return to work, or maintain complete independence as these skills require the ability to make many rapid motor decisions daily. We show that impairments in these skills are not always captured by currently used pen and paper cognitive screening tools such as the MoCA. Also, since this task relies on many domains of function to be successful, it may be a good indicator of overall stroke recovery.
The measurement of cognitive function after stroke, as measured by the MoCA, correlated moderately with distractor proportion, but only modestly with the number of distractor hits. As well distractor proportion also identified more subjects as impaired as compared to distractor hits. These differences reflect the fact that some control subjects hit a substantive number of distractors, but they also hit many targets. This is why we measured distractor proportion which quantified the ratio between distractors hit and total objects hit.
This task is also part of a larger research program to design a battery of robotic assessment tasks to create a quantitative diagnostic assessment of sensory, motor, and cognitive impairments post-stroke . The use of a robotic assessment provides objective, continuous measures of performance that are responsive to small changes and compared to a normative model of healthy control performance. This overcomes issues of subjectivity, coarse ordinal scales, criteria-based scoring, and lack of responsiveness (including floor and ceiling effects) seen in many standard assessments of post-stroke functioning. We have also developed assessments of visually-guided reaching , bimanual control , limb position sense , kinesthesia , and limb afferent feedback for action . The goal is that information from this assessment battery may be used collectively to provide more precise and responsive tools to guide individualized rehabilitation care.
Successful performance in the current task requires many sensorimotor and cognitive skills, thus failure can reflect many potential impairments in sensory, motor and cognitive functions. In order to identify unique impairments in individual participants, it is important to consider the type of parameters that show poor performance. For example, subjects who have impairments in the number objects hit, but not distractor proportion, may have underlying sensorimotor impairments, but no cognitive impairments. These subjects may be better candidates for sensorimotor rather than more cognitive-related rehabilitation. Future work is required to identify whether these patterns of impairment can predict the best type of rehabilitation for each individual.
The object hit and avoid task provides a simple and fast approach to quantify the use of attention and selection to perform rapid motor actions with the arms. Most subjects with stroke were found to be impaired when performing this task, especially those with neglect. Many parameters had high inter-rater reliability and correlated with various clinical measures of impairments and ability to perform daily activities.
Behavioural inattention test
- C + SC:
Cortical + subcortical
- Cb + Br:
Cerebellar + brainstem
Chedoke-McMaster stroke assessment
Functional independence measure
Montreal cognitive assessment
The authors would like to thanks S Appaqaq, H Bretzke, MJ Demers, M Metzler, K Moore, J Peterson, M Piitz, and J Yajure for their help with data collection, patient recruitment, and technical support.
This work was supported by the Ontario Research Fund – Research Excellence (ORF-RE 04–47), Canadian Institute of Health Research operating grants (MOP 106662), and the Heart and Stroke Foundation of Canada (G-13-0003029). SH Scott was supported by a GSK-CIHR chair in Neuroscience (XGG124631 & 279791). None of these funding sources had any role in the design of the study, the collection, analysis, and interpretation of data, nor in writing the manuscript.
Availability of data and materials
Please contact the author for data requests.
TCB assisted in data collection, and was the primary person responsible for conducting the data analysis, and writing the manuscript. CRL assisted with data analysis and drafting of the manuscript. SPD assisted with data collection and drafting of the manuscript. SDB assisted with data collection and drafting of the manuscript. KEN assisted with data analysis and drafting of the manuscript. SHS participated in the design of the study, data analysis, and drafting of the manuscript. All authors read and approved the final manuscript.
SHS is co-founder and chief scientific officer of BKIN Technologies that commercializes the KINARM robot.
Consent for publication
Ethics approval and consent to participate
This study was approved by the Queen’s University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board (#ANAT-024-05) and the University of Calgary’s Conjoint Health Research Ethics Board (#22123) and subjects provided informed consent.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Cisek P. Cortical mechanisms of action selection: the affordance competition hypothesis. Philos Trans R Soc Lond B Biol Sci. 2007;362:1585–99.PubMedPubMed CentralGoogle Scholar
- Cisek P, Pastor-Bernier A. On the challenges and mechanisms of embodied decisions. Philos Trans R Soc Lond B Biol Sci. 2014;369.
- Scott SH. A functional taxonomy of bottom-up sensory feedback processing for motor actions. Trends Neurosci. 2016;39:512–26.PubMedGoogle Scholar
- Treue S. Neural correlates of attention in primate visual cortex. Trends Neurosci. 2001;24:295–300.PubMedGoogle Scholar
- Tanaka K, Saito H, Fukada Y, Moriya M. Coding visual images of objects in the inferotemporal cortex of the macaque monkey. J Neurophysiol. 1991;66:170–89.PubMedGoogle Scholar
- Eskandar EN, Richmond BJ, Optican LM. Role of inferior temporal neurons in visual memory. I. Temporal encoding of information about visual images, recalled images, and behavioral context. J Neurophysiol. 1992;68:1277–95.PubMedGoogle Scholar
- Munoz DP, Everling S. Look away: the anti-saccade task and the voluntary control of eye movement. Nat Rev Neurosci. 2004;5:218–28.PubMedGoogle Scholar
- Day BL, Lyon IN. Voluntary modification of automatic arm movements evoked by motion of a visual target. Exp Brain Res. 2000;130:159–68.PubMedGoogle Scholar
- Guitton D, Buchtel HA, Douglas RM. Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal-directed saccades. Exp Brain Res. 1985;58:455–72.PubMedGoogle Scholar
- Pierrot-Deseilligny C, Muri RM, Ploner CJ, Gaymard B, Demeret S, Rivaud-Pechoux S. Decisional role of the dorsolateral prefrontal cortex in ocular motor behaviour. Brain. 2003;126:1460–73.PubMedGoogle Scholar
- Hawkins KM, Sayegh P, Yan X, Crawford JD, Sergio LE. Neural activity in superior parietal cortex during rule-based visual-motor transformations. J Cogn Neurosci. 2013;25:436–54.PubMedGoogle Scholar
- Tippett WJ, Alexander LD, Rizkalla MN, Sergio LE, Black SE. True functional ability of chronic stroke patients. J Neuroeng Rehabil. 2013;10:1.Google Scholar
- Salek Y, Anderson ND, Sergio L. Mild cognitive impairment is associated with impaired visual-motor planning when visual stimuli and actions are incongruent. Eur Neurol. 2011;66:283–93.PubMedGoogle Scholar
- Tippett WJ, Sergio LE. Visuomotor integration is impaired in early stage Alzheimer’s disease. Brain Res. 2006;1102:92–102.PubMedGoogle Scholar
- Brown JA, Dalecki M, Hughes C, Macpherson AK, Sergio LE. Cognitive-motor integration deficits in young adult athletes following concussion. BMC Sports Sci Med Rehabil. 2015;7:1.Google Scholar
- Teasell R, Hussein N. Clinical consequences of stroke. In: Evidence based review of stroke rehabilitation (wwwebrsrcom). 2013.Google Scholar
- Salter K, Campbell N, Richardson M, Mehta S, Jutai J, Zettler L, Moses M, McClure A, Mays R, Foley N, Teasell R. Outcome measures in stroke rehabilitation. In: Evidence Based Review of Stroke Rehabilitation (wwwebrsrcom). 2013.Google Scholar
- Cumming TB, Brodtmann A, Darby D, Bernhardt J. Cutting a long story short: reaction times in acute stroke are associated with longer term cognitive outcomes. J Neurol Sci. 2012;322:102–6.PubMedGoogle Scholar
- Ballard C, Stephens S, Kenny R, Kalaria R, Tovee M, O’Brien J. Profile of neuropsychological deficits in older stroke survivors without dementia. Dement Geriatr Cogn Disord. 2003;16:52–6.PubMedGoogle Scholar
- Sanchez-Cubillo I, Perianez J, Adrover-Roig D, Rodriguez-Sanchez J, Rios-Lago M, Tirapu J, Barcelo F. Construct validity of the trail making test: role of task-switching, working memory, inhibition/interference control, and visuomotor abilities. J Int Neuropsychol Soc. 2009;15:438.PubMedGoogle Scholar
- Scott SH, Dukelow SP. Potential of robots as next-generation technology for clinical assessment of neurological disorders and upper-limb therapy. J Rehabil Res Dev. 2011;48:335–53.PubMedGoogle Scholar
- Tyryshkin K, Coderre AM, Glasgow JI, Herter TM, Bagg SD, Dukelow SP, Scott SH. A robotic object hitting task to quantify sensorimotor impairments in participants with stroke. J Neuroeng Rehabil. 2014;11:47.PubMedPubMed CentralGoogle Scholar
- Erez AB, Katz N, Ring H, Soroker N. Assessment of spatial neglect using computerised feature and conjunction visual search tasks. Neuropsychol Rehabil. 2009;19:677–95.PubMedGoogle Scholar
- Rossit S, Malhotra P, Muir K, Reeves I, Duncan G, Harvey M. The role of right temporal lobe structures in off-line action: evidence from lesion-behavior mapping in stroke patients. Cereb Cortex. 2011;21:2751–61.PubMedGoogle Scholar
- Scott SH. Apparatus for measuring and perturbing shoulder and elbow joint positions and torques during reaching. J Neurosci Methods. 1999;89:119–27.PubMedGoogle Scholar
- Dukelow SP, Herter TM, Moore KD, Demers MJ, Glasgow JI, Bagg SD, Norman KE, Scott SH. Quantitative assessment of limb position sense following stroke. Neurorehabil Neural Repair. 2010;24:178–87.PubMedGoogle Scholar
- von Bertalanffy L. On the von Bertalanffy growth curve. Growth. 1966;30:123–4.Google Scholar
- Allen KR. A method of fitting growth curves of the von bertalanffy type to observed data. J Fish Res Board Can. 1966;23:163–79.Google Scholar
- Cormen TH, Leiserson CE, Rivest RL. Introduction to algorithms. Cambridge: MIT press; 1990.Google Scholar
- Lowrey C, Jackson C, Bagg S, Dukelow S, Scott S. A novel robotic task for assessing impairments in bimanual coordination post-stroke. Int J Phys Med Rehabil S. 2014;3:2.Google Scholar
- Box GE, Cox DR. An analysis of transformations. J R Stat Soc B Methodol. 1964;211–252.
- Gowland C, Stratford P, Ward M, Moreland J, Torresin W, Van Hullenaar S, Sanford J, Barreca S, Vanspall B, Plews N. Measuring physical impairment and disability with the chedoke-McMaster stroke assessment. Stroke. 1993;24:58–63.PubMedGoogle Scholar
- Brunnstrom S. Motor testing procedures in hemiplegia: based on sequential recovery stages. Phys Ther. 1966;46:357.PubMedGoogle Scholar
- Bohannon RW, Smith MB. Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther. 1987;67:206–7.PubMedGoogle Scholar
- Keith RA, Granger CV, Hamilton BB, Sherwin FS. The functional independence measure: a new tool for rehabilitation. Adv Clin Rehabil. 1987;1:6–18.PubMedGoogle Scholar
- Wilson B, Cockburn J, Halligan P. Development of a behavioral test of visuospatial neglect. Arch Phys Med Rehabil. 1987;68:98–102.PubMedGoogle Scholar
- Nasreddine ZS, Phillips NA, Bedirian V, Charbonneau S, Whitehead V, Collin I, Cummings JL, Chertkow H. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53:695–9.PubMedGoogle Scholar
- Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9:97–113.PubMedGoogle Scholar
- Pasi M, Salvadori E, Poggesi A, Inzitari D, Pantoni L. Factors predicting the Montreal cognitive assessment (MoCA) applicability and performances in a stroke unit. J Neurol. 2013;260:1518–26.PubMedGoogle Scholar
- Cicerone KD, Dahlberg C, Kalmar K, Langenbahn DM, Malec JF, Bergquist TF, Felicetti T, Giacino JT, Harley JP, Harrington DE. Evidence-based cognitive rehabilitation: recommendations for clinical practice. Arch Phys Med Rehabil. 2000;81:1596–615.PubMedGoogle Scholar
- Treisman AM, Gelade G. A feature-integration theory of attention. Cogn Psychol. 1980;12:97–136.PubMedGoogle Scholar
- Theeuwes J, Kramer AF, Hahn S, Irwin DE. Our eyes do not always go where we want them to go: capture of the eyes by new objects. Psychol Sci. 1998;9:379–85.Google Scholar
- Hallett PE. Primary and secondary saccades to goals defined by instructions. Vision Res. 1978;18:1279–96.PubMedGoogle Scholar
- Coderre AM, Zeid AA, Dukelow SP, Demmer MJ, Moore KD, Demers MJ, Bretzke H, Herter TM, Glasgow JI, Norman KE, et al. Assessment of upper-limb sensorimotor function of subacute stroke patients using visually guided reaching. Neurorehabil Neural Repair. 2010;24:528–41.PubMedGoogle Scholar
- Semrau JA, Herter TM, Scott SH, Dukelow SP. Robotic identification of kinesthetic deficits after stroke. Stroke. 2013;44:3414–21.PubMedGoogle Scholar
- Bourke TC, Coderre AM, Bagg SD, Dukelow SP, Norman KE, Scott SH. Impaired corrective responses to postural perturbations of the arm in individuals with subacute stroke. J Neuroeng Rehabil. 2015;12:7.PubMedPubMed CentralGoogle Scholar