**1. Introduction**

During our entire lives we acquire new skills, adapt to new situations, and learn to make decisions, which is known as "motor learning" [1]. Motor learning is an ongoing process and is initiated in movement preparations, during which our motor system explores the environment and the different possibilities to gradually improve motor control ("reinforcement learning") [2]. Performing a movement results from the synchronous interaction of multiple systems with the environment ("dynamic systems model") [3,4]. Within this dynamic systems model, our central nervous system (CNS) has to choose a pathway, from a vast number of movement options, to perform the action. In other words, our CNS has to overcome the "degrees of freedom (DOF) problem" and has to select one strategy to execute the desired movement [3]. This reflects the high redundancy of our motor system, with the plausible consequence that our actions become highly variable as well and that we are unable to repeat the

exact same movement in repetitive occupational tasks. This phenomenon is termed motor variability, which has been defined as variance in movement of an individual who performs under similar task conditions [5].

The study of Madeleine et al. [6] illustrates that experienced workers showed more motor variability than novices. Sandlund et al. [7] even proposed that motor variability could be considered an individual or personal factor predicting which workers would be prone to develop work-related musculoskeletal disorders (WRMSD), theoretically implying that there could be an optimal level of motor variability. Several studies showed muscular activity levels and motor learning being negatively associated, meaning that muscular activity decreases along the process of motor learning in gaining experience [6,8]. Furthermore, muscular activity was higher when exposed to new environments compared to familiar environments [9,10].

In an occupational context, researchers and practitioners aim to design workplaces by means of ergonomics tools or work organization (e.g., task rotation) that lead to a decreased workload and reduce the risk of WRMSD. In this respect, the role of motor variability in relation to WRMSD has caught the attention of many researchers. Mainly, laboratory studies have been conducted to assess motor variability in an occupational context, since experienced workers from the field have limited availability and conducting field studies generally requires additional resources. Laboratory studies often include inexperienced subjects or novices to test what the effect of ergonomically improved tools or conditions is in a specific work simulation, judged by measures of physical requirements, work performance, and motor variability (e.g., [11,12]). Measures of physical requirements are the static, median or mean, and peak levels of muscular activity, according to the exposure variation analysis [13], median or mean of kinematic variables like joint angles, movement velocity or acceleration, and median or mean of the heart rate. Measures of motor variability are linear metrics like cycle-to-cycle standard deviation or coefficient of variation [14], or nonlinear metrics like entropy-based metrics and coordination metrics [15]. The difficulty of interpreting laboratory assessments among novices is that motor learning may influence study results and respective effects, expressed in terms of, e.g., muscular effort, could be misinterpreted. For example, early phases of the motor learning process could result in higher levels of muscle activity, as shown in a previous study [16].

The aim of this study is to investigate whether physical requirements and motor variability of novices, as a function of adjusted motor control strategies, change over a three-day repeated fine motor task. For investigating the physical requirements, we calculated the static (10th percentile), median (50th percentile), and peak (90th percentile) levels of muscular activity, the mean forearm acceleration, and the mean heart rate. For investigating motor variability, we calculated the cycle-to-cycle variability (coefficient of variation, CV) of muscular activity and forearm acceleration. We have used an exploratory approach because this study was originally not designed to investigate aspects of motor learning, but to assess the test-retest reliability of different normalization procedures of surface electromyography [17]. We hypothesized that measures of physical requirements and motor variability would decrease over days. With the results of this study, we aim to highlight methodological issues that should be considered when designing studies to minimize misinterpretation of results and increase practical relevance and whether we can rely on performing just one measurement to assess the risk of musculoskeletal disorders.

#### **2. Materials and Methods**

#### *2.1. Study Population*

Subjects were excluded if they had any acute or cardiovascular diseases, impaired range of motion of the neck and upper extremities, or neurological impairments. A total of 65 subjects were recruited for study participation, of which 8 dropped-out due to methodological or organizational issues. The final study population consisted of 57 healthy subjects, of which 30 were female and 27 were male. Details with regard to the subjects can be found in Table 1. All subjects were inexperienced, meaning that none of them had specific experience in the repetitive task with the two specific screwing and fastening tools. The study (260/201BO2) was approved by the local ethics committee of the University of Tübingen (Germany) and all subjects signed the informed consent prior to participation.


**Table 1.** Description of the study population, as a whole and broken down by sex. Results are displayed as value or as mean ± standard deviation (SD).
