top of page

Fundamental Processes of Sensorimotor Learning: Reasoning, Refinement, and Retrieval.

Learning complex motor skills is a uniquely challenging endeavor. This process requires the simultaneous coordination of numerous muscles and joints, each subject to different biomechanical constraints and metabolic demands. While some constraints on coordination are wired into our genetic code and emerge spontaneously given the right environment, the coordination demands for complex skills require considerable cognitive control over these many degrees of freedom.

 

Additionally, learning complex motor skills often involves developing new physical intuitions about how subtle differences in movement can lead to drastically different outcomes (e.g., an amputee learning to use a new myoelectric prosthesis). Yet, despite these formidable challenges, humans are capable of learning a near-limitless repertoire of movements, enabling babies to walk and talk, athletes to achieve incredible levels of skill, and patients to recover from neurological disorders.

 

How humans achieve these impressive feats of complex motor skill acquisition remains poorly understood. This gap in knowledge partly stems from a long-standing neglect of the role of cognition in motor learning because such processes are generally hard to formalize and often exhibit high variability. Conversely, cognitive science frequently overlooks the role of motor control in decision-making; for example, many decisions are constrained by the sensorimotor outcomes associated with making a choice. 

 

To make progress toward a comprehensive theory of motor learning—one that can explain the intricate cognitive–motor interactions that facilitate successful motor skill acquisition — we propose a ‘3R’ framework that integrates three fundamental concepts shared between the motor learning and cognitive science communities: reasoning, refinement, and retrieval. This new framework establishes a way to understand how new sensorimotor control policies—mappings between a learner’s state and context to an action—are acquired through reasoning, optimized through refinement, and automatized through retrieval. As part of this initiative, we are actively investigating several key questions.

 

1. How do we reason about common sensorimotor relationships during motor learning, such as rotations (e.g., adjusting for a wind-blown tennis ball) or reflections (e.g., when shaving in front of a mirror)? What makes some sensorimotor relationships easier or harder to learn than others?

 

2. How does the timing, quantity, and configuration of movement goals influence the processes of reasoning, refinement, and retrieval in motor planning?

 

3. Do left and right hemispheres have shared or distinct roles in reasoning, refinement, and retrieval?

Continuous Transformation Theory of the Cerebellum.

The notion that the cerebellum plays a causal role in cognition is now well established, with a hypothesis offered decades ago helping to inspire a large body of research in systems and cognitive neuroscience. This work has convincingly linked cerebellar function to performance in diverse areas such as working memory, attention, decision-making, child development, and social cognition. It has also led to the establishment of a new clinical diagnostic category, Cerebellar Cognitive Affective Syndrome, and spurred research into the underlying circuitry responsible for the cerebellum’s role in non-motor behavior.

 

To date, much of this research has been descriptive, focusing on questions like whether the cerebellum is involved in cognitive functions such as working memory or attention. There has been limited progress in synthesizing the growing body of literature and developing unifying accounts of how the cerebellum contributes to both motor and non-motor functions. Many researchers have been guided by the idea of the "universal cerebellar transform" (UCT), which suggests that the cerebellum performs a specialized computation that applies across various domains due to its uniform structure and widespread connections with the cerebral cortex. While this idea remains debated, the UCT framework can be a useful tool for developing and testing theories about the cognitive role of the cerebellum.

 

The concept of prediction, or forward models, is central to many theories of cerebellar function. However, to be computationally meaningful, it's important to specify the constraints on cerebellar-based predictions. To address this, we propose a theory centered on the cerebellum being crucial for predictions involving the continuous transformation of sensorimotor and cognitive states, particularly in space and time—domains that are inherently continuous. This theory brings together findings from multiple areas of research. In sensorimotor control, the cerebellum predicts the dynamics of future bodily states based on current spatial and temporal information. In sensorimotor learning, the cerebellum’s integrity is key for precise timing, such as in eyelid conditioning or complex movement coordination. When extended to cognitive domains, this continuous transformation theory may provide a unified explanation for cerebellar involvement in action planning, visual cognition, social cognition, and mental arithmetic. As part of this initiative, we are actively investigating several key questions.

 

1. How does the cerebellum influence value-based decision-making, the cognitive process by which individuals assess and choose between options based on potential rewards, risks, and anticipated outcomes?

2. What role does the cerebellum play in theory of mind—the ability to understand and infer others' thoughts and intentions?

 

3. How is the cerebellum involved in both implicit and explicit sensorimotor learning? 

Benchmarking Human Performance using Large Datasets and Emerging Technologies.

The study of human sensorimotor science dates back to the early days of experimental psychology. In 1897, George Stratton conducted a classic self-experiment, documenting the behavioral and psychological changes he experienced after wearing mirror-inverting glasses for eight consecutive days. Today, similar questions are explored through environments and virtual reality systems that allow researchers to perturb movement feedback in controlled ways.

 

Typically, studies of human sensorimotor learning are conducted with specialized equipment in laboratory settings. This approach has been highly effective in uncovering critical spatial and temporal constraints on adaptation, as well as revealing the contributions of different neural systems to this form of learning. However, these studies often rely on small, homogenous samples, increasing the risk that findings may not generalize beyond the lab. Most research has been conducted with WEIRD samples (Western, educated, industrialized, rich, and democratic), assuming that motor learning shows little variability across populations. Indeed, if sensorimotor neuroscience aims to uncover universal principles and understand how individual experiences drive behavioral variability, large-scale investigations into how task and demographic factors influence sensorimotor learning are overdue.

 

To address these challenges, we have developed web-based approaches capable of scaling motor learning data collection by orders of magnitude (at least 100x) compared to traditional methods. Our crowdsourcing approach allows us to recruit a large, diverse cohort of participants. Using these datasets, we are building cross-validated predictive models to identify the key features of successful motor learning. These models not only capture patterns that hold across the population but also reveal under-appreciated factors that shape individual differences. Understanding these differences will help instructors and clinicians tailor interventions more effectively, while guiding future hypothesis-driven lab studies to uncover the computational mechanisms behind this variability. As part of this effort, we are exploring several key research questions:

 

1. How do individual differences in age, handedness, sex, physical fitness, and cognitive abilities impact sensorimotor control (e.g., reaction time, movement accuracy, hand angle variability) and sensorimotor learning (e.g., reasoning, refinement, retrieval in motor skill acquisition)?

 

2. What is the optimal practice schedule for learning sign language, and how does different performance feedback impact learning efficiency? 

 

3. What shared and unique sensorimotor synergies underlie various manual tasks, and how can insights into these synergies guide the design of motor learning curricula to enhance dexterity?

bottom of page