Climbing out of the lab: Studying motor planning and coarticulation in the climbing gym

Imagine yourself summiting the Eiger North Face. The classic route demands a judicious combination of rock, ice and crack climbing, as you ascend 1400m along a breathtaking wall. Climbing such a route is physically taxing. Your sensorimotor system needs to integrate multiple sensory streams to assess local conditions and guide your muscles to ascend without falling. Hierarchical motor planning is essential in this process: at a low-level, to select the right actions based on visual, tactile and proprioceptive cues, and at a high-level, to select the right strategy based on your skills, the availability of established routes, weather conditions and other relevant contexts.

Thus, climbing is ideal for studying motor control, planning and their interaction. Two aspects make it particularly interesting for advancing our understanding. Firstly, climbing requires coordination across all limbs in precarious postures, which remains poorly understood. Secondly, climbing routes naturally expand and compress action possibilities. For instance, certain bottlenecks can only be overcome with particular sequences of whole-body movements, forcing climbers to plan ahead to execute the correct movements.

While we know a lot about motor control (1), much of this knowledge has been gained in constrained laboratory environments and tasks. While this reductionist approach is highly valuable, it falls short of understanding motor control in the complex, dynamic nature of real-world settings (2). Part of this limitation, often described as a lack of "ecological validity," refers to the inability of measurements and behaviors observed in research settings to accurately represent those occurring in the real world (3). However, the study of motor control in unconstrained environments introduces additional challenges, particularly in terms of data collection and analysis.

This is where Maseli et al. innovate (4) by carefully designing tractable climbing tasks in a hypothesis-driven way. They combined motion capture, dimensionality reduction and statistical techniques to gain insights. Their hypotheses are related to “coarticulation” - the phenomenon in which the execution of subtasks is influenced by the overall task, as well as the subtask sequence. While extensively studied in linguistics (5) and in motor control for various end-effectors (1), coarticulation remains underexplored in whole-body coordination tasks like climbing.

According to Godoy et al. (6), coarticulation is both a biomechanical necessity (driven by biomechanical constraints), and a motor control necessity (arising from the need for anticipatory motor planning to optimize efficiency). It results in a contextual modulation of the motor output. The temporal aspect of this modulation can be viewed through carryover effects; the influence of past events on future events; or anticipatory effects; the influence of future events on past events. Notably, anticipatory effects provide valuable insights into the formulation of high-level motor plans and scale with task expertise (Figure 1).

Figure 1. Coarticulation and Motor planning.

Coarticulation results in the fusion of movement parts due to motor control necessities, such as anticipation, and biomechanical constraints. Anticipation corresponds to the influence of a motor plan on the present body state. A motor plan separates the choice of the end effector motion (action selection), decisions regarding trajectory shapes (abstract kinematic representations) and postural adjustments (movement specification).

The authors first investigate the existence of coarticulation when participants are given explicit instruction about the sequence of movements to be executed. In contrast with previous studies, Maselli et.al consider whole-body kinematics and, while studying action selection, also have insights on abstract kinematic representations and movement specification. They recruited 21 participants with varying climbing experience (non-climbers, beginner-climbers and expert-climbers) and tracked 25 passive markers. In each trial, participants were first asked to mimic the climbing hand-moves, ensuring the formation of a motor plan before climbing. Participants were then asked to perform a 2-step climbing task and the influence of the second step – representing the anticipatory effect related to the final hold – was analyzed.

As expected, Maselli et al. found that most participants, regardless of their expertise, displayed some degree of coarticulation and anticipation during the climbing task. Furthermore, the authors hypothesized that the degree of coarticulation should increase with the level of expertise. Learning is particularly relevant for sport climbing – while sport climbing requires complex multimodal sensory integration, drawing upon visual, motor, haptic, and proprioceptive information, climbing performance is in fact more influenced by trainable variables rather than anthropometric characteristics (7). Additionally, Maselli et al. found that expertise modulates coarticulation. Indeed, expert-climbers displayed a high level of coarticulation and showed a more refined ability to anticipate multiple steps ahead.

By analyzing whole-body kinematics, the authors studied the contribution of each body part to the coarticulation score. While commonly discarded in standard constrained setups, the different motor strategies and styles adopted by each individual were clearly visible from the large inter-individual variability in how coarticulation emerged across different body parts and stages of the task.

Another interesting finding is the strong connection between coarticulation and goal position rather than the identity of the hand reaching that goal, suggesting that climbers structure their movements around spatial targets rather than limb-specific actions. This finding underlies the importance of the goal’s spatial aspect, such as the direction and distance, in motor planning.

In summary, Maselli et al. investigated coarticulation in sport climbing. They found that limb movements are differently controlled depending on the context. Furthermore, they found that climbing skill level influences the degree of coarticulation, with more experienced climber exhibiting greater coarticulation in the form of anticipation This study opens up many exciting possibilities for future research. While the simplicity of the climbing route allowed individuals with a wide range of climbing expertise to participate, it also limited the range of required motor skills and the degree to which coarticulation could be observed. Future studies could consider a broad variety of climbing routes. Indeed, by carefully designing the location and shapes of the holds, as well as the inclination of the wall, one can nicely adjust the complexity of the task. This further highlights the power of sports climbing as a testbed for studying motor control.

Nevertheless, the ability to study coarticulation in climbing scenarios elevates the study of motor planning toward new theoretical perspectives and should be explored at the algorithmic and computational levels. Biomechanics simulators and reinforcement learning algorithms are becoming increasingly performant, enabling the study of disregarded complex behaviors such as the evolution of high-jump (8) or skilled motor behavior (9). Furthermore, as highlighted by Cisek & Greene (2) a deeper understanding of complex motor behavior like sports climbing will require additional measurement and modeling tools. Here, our recent work on the unsupervised decomposition of hierarchical behavior provides a promising starting point (10).

Overall, the careful design and the continued improvement of tailored methodologies for movement analysis aims to enhance the ecological validity in motor control research. These advances pave the way for a more profound understanding of natural behaviors.