Healthy sex- and age-matched controls. Participants should have no history of any neurological disorder or cancer.

The QuickDASH questionnaire is a validated, shortened version of the Disabilities of the Arm, Shoulder, and Hand (DASH) tool, and it is used in breast cancer patients to assess upper extremity function and disability

The EORTC QLQ-CIPN20 is a validated questionnaire developed by the European Organisation for Research and Treatment of Cancer to assess chemotherapy-induced peripheral neuropathy (CIPN) and its impact on quality of life in cancer patients. It includes 20 items that evaluate sensory, motor, and autonomic symptoms-such as tingling, numbness, pain, and functional impairments in the hands and feet-experienced over the past week

The Brief Pain Inventory - Short Form (BPI-sf) is a 9-item, self-administered questionnaire designed to quickly assess:

1. Pain Severity: Patients rate their pain at its worst, least, average, and current level over the past 24 hours using a 0-10 numeric scale.
2. Pain Interference: Patients rate how much pain has interfered with seven aspects of daily life-general activity, mood, walking ability, normal work, relationships, sleep, and enjoyment of life-also on a 0-10 scale.
3. Pain Relief: It includes questions about pain treatments, medications, and the percentage of relief experienced.

The Multidimensional Fatigue Inventory (MFI-20) is a 20-item self-report questionnaire designed to assess five dimensions of fatigue in cancer patients and other populations:

1. General Fatigue
2. Physical Fatigue
3. Reduced Activity
4. Reduced Motivation
5. Mental Fatigue

This is a brief cognitive assessment tool designed to detect possible mild cognitive impairments. Participants are asked questions which evaluate several cognitive domains, including: Attention and concentration, executive functions, memory, language, visuospatial skills, abstract thinking, calculation and orientation. Participants can receive up to 30 points for correctly answered questions, with a score of 26 or above considered normal.

In this task, for each trial, 16 squares are displayed on a circle in front of the participant. Then, between 3 to 6 of them become filled for a second before these stimuli disappear. One second later, a question mark appears in one of the 16 squares and the participant has to indicate whether this square was filled by a stimulus or not. Primary outcome of the task is working memory capacity.

The participant will grab the handles of the robot which controls the cursor on the screen. To start a trial the participant moves the cursor to the starting position in the middle of the screen. One of four targets will appear which the participant is instructed to reach towards as quickly and as accurately as possible and stop within the target. After each reaching movement, the starting position becomes visible again and trials are repeated until each target has been repeated 5 times, so 20 trials in total. A composite score is computed by an algorithm devised by the Kinarm company and is used as outcome for this task.

During this task the robot moves the dominant hand to four different positions and the participant is asked to mirror the movement with the non-dominant hand. Larger differences in the position of the dominant and non-dominant hand are an indication of less accurate position sense. The primary outcome is the two-dimensional variability of the error between the target position and the matched position (error variability).

In this task, participants will be making arm reaching movements while their movement will be constrained by the robot to a certain angular deviation relative to a straight line to the target. Participants are then instructed to verbally indicate whether they were deviated to the left or to the right of a reference position (straight ahead). By gradually decreasing the angular deviation, it is possible to estimate how accurately participants could discriminate angular deviations. In order to obtain an efficient estimation of the perceptual boundary for each individual, a parameter estimation by sequential testing (PEST) procedure is applied. This algorithm starts with a large deviation and depending on the individual's response, it decreases the size of the deviation progressively until the deviation falls below a minimum threshold. The primary outcome is the slope of the psychometric curve obtained from the answers of the participants.

This task assesses proprioceptive function by measuring force responses to mechanical perturbations during goal-directed reaching. Participants perform rapid reaching movements toward a visual target without visual feedback of the hand. On a subset of trials, the robotic device applies lateral perturbations of varying magnitudes that deviate the hand from the target. Participants are instructed to correct the movement and reach the target. The force exerted against the perturbation is recorded as an indicator of proprioceptive detection. To account for passive arm dynamics, participants also perform trials with a large target, which minimizes active corrective responses. Comparing force responses between small- and large-target conditions allows isolation of active responses to perturbations, providing a measure of proprioceptive sensitivity.

This task looks at the unconscious integration of proprioception with vision. In this task, participants have to reach to target and stop on it. The hand of the participant is hidden from view but represented by a cursor that moves like the hand. On some trials, the cursor is deviated from its trajectory by a given angle (between -30 and 30°). On those trials, there is a discrepancy between the position of the hand and the position of the cursor. Once the participants have stopped on the target, they are requested to move back to the starting position in the absence of any visual information about their actual hand position. The direction in which the participants start to move is a readout of the integration of vision and proprioception of the hand position signals.

Two versions of this task will be performed, distally (finger) and proximally (shoulder). Each version includes two different phases, a perception and a reproduction phase. In the perception phase, participants will be asked to memorize the force applied by the robot on the non-dominant arm. In the reproduction phase, participants will be required to move a slider that controls the amount of force exerted by the robot on the hand, with their dominant hand. The participant will try and reproduce the previously perceived force by manipulating the slider.

The distal version will be performed using a small lever actuated by a motor. Participant will place their left index finger under the lever. Participants will again be asked to perceive and remember the force in the perception phase, and reproduce it using a slider in the reproduction phase.

For this task several variations to assess wrist and shoulder proprioception will performed. The experimenter will rotate the limb of the participant around a joint from a neutral position to a target position determined by a given angle (15, 30 or 45°). The participant will be asked to reproduce this position with the contralateral limb. All movements happen in the absence of any visual feedback about the limb position. Each target position is repeated ten times. For the shoulder task, the arms will be moved in the scapular plane (shoulder abduction). For the wrist, it will be in the sagittal plane (wrist flexion).

At the start of each trial, the experimenter will place the active limb on the home position (~20cm away from the body midline and ~20cm in front of the torso). The experimenter will place the finger with the pulp of the finger placed upward at one of three possible positions (the target). The three possible target locations are ~20cm from the body midline, contralateral to the home position. They will be ~15, 20 and 25cm in front of the torso. The participants will be instructed to make a swift movement with the index finger of the active limb to the target finger, wait there for approximately 1s and then go back straight to the home position. Participants will perform 10 trials per target location, which will be pseudo-randomized (each target location will be used once per cycle of three trials).

Cognitive attention is an important factor for correct performance for many of the above mentioned tasks. Early-onset fatigue is a frequent problem for patients who have just undergone breast cancer treatment, thereby possibly affecting the results of our study. Therefore, participants will be asked to indicate their fatigue levels after every robotic task using a visual analogue scale (VAS) on a tablet, with the score ranging from 0-100 with zero no fatigue and 100 complete exhaustion.

Fear for movement in patients who have just undergone surgery could impact the motor control and thereby affect the results of several proprioceptive tasks. Thus, participants will also be asked to indicate their level of fear to perform tasks after completion of each task, with the score ranging from 0-100 with zero being not reluctant and 100 extremely reluctant.

In this task the detection threshold is tested using standardized monofilaments (Optihair2-Set, Marstock Nervtest, Germany) which exert forces between a range from 0.25 to 512 mN, depending on the thickness of the filament. The test starts by the researcher providing stimulation with the thickest monofilament which exerts the highest force. If the participant is able to detect the force, the researcher goes down in monofilament size until the participant is unable to detect a monofilament. The smallest size detected by the participant is established as the infra threshold. The process is then repeated in the opposite direction, with increasing size instead of decreasing. This is done until the participant is able to detect the stimulation again, which is then noted as the supra limit. This is repeated until five infra and supra limits each have been noted.

During this task the pain threshold is assessed using a set of standardized pinpricks (MRC's PinPrick stimulators) which exert a force within a range between 8-512 mN. These pinpricks are weighted, which are progressively increasing, thereby increasing the exerted force. The participants are asked to describe the stimulation either as a blunt or as a sharp stimulation. The test is started using the pen with the lowest force exertion, and if the participant describes the stimulation as blunt the next increased progressive pen will be used. This will continue until the participant describes the subsequent stimulation as sharp. When this happens, that pen is noted as the supra threshold. The process is then repeated in the opposite direction, now with deceasing force stimuli, until the participants describe the stimulation as blunt, which will then be noted as the infra threshold. This is repeated until both infra and supra have been noted 5 times each.

The vibration detection threshold is used in the QST as a proprioceptive measurement. It is performed with a Rydel-Seiffer tuning fork which vibrates at a frequency of 128 Hz. During the assessment, the tuning fork is placed on several boney structures, including the thumb, wrist, elbow, and the shoulder. The tuning fork is struck causing it to vibrate and is then subsequently placed on the boney structure. The tuning fork has a damping scale, which is a triangle with a numbered scale along the height of the triangle. The visibility of the tip of the triangle while the fork is vibrating will depend on the amplitude of the vibration. While the tuning fork is placed on a boney structure, the participant has to indicate when they cannot detect the vibration anymore. At this moment, the researcher will read the number from the scale at which the tip of the triangle is currently visible. This represents the amplitude of the vibration the participant is able to detect.

In this task, participants are presented with five plates (2x2cm). There is a small bump in one of them. The participant is asked to indicate on which plate a bump is present and how confident they are about their choice on a scale from 1 to 3 (not certain at all, more or less certain, without any doubt). There are five different bump heights: 5, 10, 15, 20 and 25 μm. Each bump height is presented three times. This is done without vision as the plates are hidden from the participants. The primary outcome is the bump detection threshold, defined as the lowest bump height that was successfully detected in two of the three trials and the next higher bumps were successfully detected on 3/3 trials.

The range of motion for the arms will be tested using a digital goniometer (EasyAngle, Meloq, Stockholm, Sweden)). The degree of maximal active movement will be recorded of: 1) humerothoracic elevation in the scapular plane; 2) arm external rotation in 0° humerothoracic abduction position and 3) arm external rotation in 90° humerothoracic abduction position. The digital goniometer will be placed on the upper arm and active range of motion (°) are assessed at both sides for the three movement tasks. For arm elevation, the participant will start with the arms straight alongside the body and the thumbs pointing forward and without shoulder girdle elevation. For the external rotation task at 0° and 90° of abduction, the elbow will be flexed to 90°. For the external rotation task at 90° of abduction, an abduction pillow will be added to support the upper arm.

For the arm, a perimeter, which is a flexible stainless-steel bar with a tapeline fixed every 4 cm and a weight of 20 g at the end of each tapeline, will be used to measure the circumference, allowing us to calculate the volume. The perimeter will be placed on each arm of the participant subsequently allowing the circumference of each arm to be determined in 4 cm intervals. Afterwards, the volume of the arm is calculated using a truncated cone, the relative arm volume (%) will be calculated by the formula: ((volume of the affected arm - volume of the non-affected arm) / volume of the non-affected arm) x 100. Correction for arm dominance was done by adjusting the arm volume of the non-dominant arm with 3.3% because literature indicates that the non-dominant arm is on average 3.3.% smaller than the dominant arm.

For the hand, bilateral water displacement method using a volumeter will be used. The volumeter is filled with water, allowed to settle at room temperature, and the patient is asked to lower their hand slowly, with the forearm pronated and the fingers resting in adduction, into the volumeter until the web of the middle and ring finger rest on the stop dowl. The patient remains still until there is more than 5 s between each drip at the overflow. The displaced water will be collected in a beaker. The amount of water displacement is weighed in kilograms, and converted to milliliters (1kg =1000mL). Afterwards the total volume (mL) of the arm and hand can be calculated by taking the sum of the two measurements.

Overall proprioceptive accuracy across upper-limb tasks, represented by spatial error in position/target matching and reach tasks. Measures are derived from robotic tasks (including arm position matching, and perceptual boundary tasks) and video-based tasks (including the joint position matching and finger-to-finger reaching tasks). Example metrics: mean absolute error

Accuracy and consistency of perceived vs produced force during force-matching tasks (finger and whole-arm robotic force matching). Includes measures of mean error and variability in force reproduction.

Performance on tasks assessing the integration of proprioceptive and visual information, quantified during the integration of vision and proprioception task by looking at the influence of the visual perturbation on the direction of the return movement (same outcome as in Van De Plas and Orban de Xivry 2025, https://www.biorxiv.org/content/10.1101/2025.05.22.655043v2.full).

Response to externally applied mechanical perturbations quantified during a robotic reaching task by isolating active force responses from passive arm dynamics. The outcome is derived from force exerted against perturbations during small-target trials relative to wide-target trials, summarized as the slope of a psychometric function reflecting the probability of detecting and counteracting perturbations across perturbation magnitudes.

Through the mechanical detection threshold task, the geometrical mean for the 5 infra thresholds measured and the 5 supra thresholds are determined, representing the somatosensory detection threshold.

Through the mechanical pain threshold task the geometrical mean for the 5 infra thresholds measured and the 5 supra thresholds are determined, representing the somatosensory pain threshold.

Detection threshold is expressed through the Rydel-Seiffer score (0-8), which represents the smallest amplitude of vibration (frequency of 64 Hz) the participant is able to detect. A higher Rydel-Seiffer score equals a lower vibration amplitude.

Through the bumps task, the geometrical mean for the 3 infra thresholds measured and the 3 supra thresholds are determined, representing the tactile threshold.

The hand volume is measured by water displacement, expressed grams of water. To describe potential confounding factors to performance, such as oedema.

The range of motion is measured using a digital goniometer to assess potential confounding factors to performance. The range of motion measured in degrees is compared between both arms for each of the 3 movements.

Participant's self-reported fatigue levels of the day, to determine possible confounding factor of fatigue across the two separate sessions. This is measured using visual analogue scales (VAS), scored from 0 to 100.

Fatigue is also monitored during the KINARM sesssion using the VAS after each robotic task. The scale is also ranged from 0 to 100.

For both VAS, 0 equals 'No fatigue' and 100 'Completely exhausted'.

Fear of movement is measured during the KINARM sesssion using visual analogue scales after each robotic task. The scale ranges from 0 to 100, 0 being 'Not fearful at all', and 100 'Extremely fearful'.