Manipulating Arm movement Preparation via a comPuter Interface for Neural decodinG

Project Information

MAPPING

Grant agreement ID: 101108533

DOI

10.3030/101108533

Project closed

EC signature date 12 May 2023

Start date 1 November 2023

End date 31 October 2025

Funded under

• Marie Skłodowska-Curie Actions (MSCA)

Total cost

No data

EU contribution

€ 211 754,88

Coordinated by

UNIVERSITE D'AIX MARSEILLE
France

Periodic Reporting for period 1 - MAPPING (Manipulating Arm movement Preparation via a comPuter Interface for Neural decodinG)

Reporting period: 2023-11-01 to 2025-10-31

Summary of the context and overall objectives of the project

The MAPPING project was motivated by a central challenge in systems neuroscience: understanding how neural activity in motor-related cortical areas supports the preparation of voluntary movements. While the recent dynamical systems theory of motor planning has provided a powerful conceptual framework—proposing that movement preparation consists of optimizing neural states in premotor cortex—experimental tests of its predictions have largely been confined to highly constrained laboratory paradigms. This limitation has hindered progress toward a more general understanding of how planning operates in complex, naturalistic behaviors. MAPPING addressed this gap by developing experimental and analytical approaches capable of probing preparatory neural dynamics across both constrained and unconstrained movement contexts in order to test novel predictions of the dynamical systems theory.

The overall objective of the project was to gain mechanistic insight into how preparatory neural states are organized, updated, and exploited to generate voluntary actions. To achieve this, MAPPING combined non-human primate behavioral paradigms, chronic intracortical recordings from premotor cortex, and population-level decoding methods. The project pathway to impact was structured around generating fundamental knowledge about motor planning principles that are robust across contexts, and translating this knowledge into generalizable decoding frameworks. By identifying invariant, low-dimensional neural representations of planned movement direction, the project lays the groundwork for future advances in adaptive brain–computer interfaces and neuroprosthetic control, while also contributing to theory-driven progress in motor neuroscience.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The project was structured around three interlinked scientific work packages. First, baseline behavioral and neural data were acquired and analyzed using a delayed center-out reaching task performed by macaque monkeys. Detailed analyses of hand kinematics, eye movements, and premotor spiking activity established reproducible preparatory neural states associated with different movement directions. These preparatory states were shown to lie on a low-dimensional circular manifold, consistent with predictions of the dynamical systems theory of motor planning, and provided the foundation for subsequent decoding work.

Building on these results, the second phase focused on the development of a decoding framework to extract single-trial estimates of planned movement direction from population neural activity. An offline pipeline combining dimensionality reduction and geometric projection methods was designed to relate ongoing premotor dynamics to positions along the preparatory manifold. This approach enabled accurate decoding of intended movement direction across untrained targets and variable starting positions, demonstrating a context-independent neural representation of motor plans. Although originally conceived as a real-time BCI implementation, the offline strategy substantially improved signal-to-noise ratio and statistical power, yielding robust and generalizable results.

In the final phase, the decoding framework was applied to a more naturalistic, unconstrained task in which animals freely selected the order of multiple reach targets. By projecting premotor activity during movement execution onto the preparatory manifold, the project tested competing theoretical hypotheses about whether complex action sequences are planned holistically or decomposed into submovements prepared online. Regression analyses revealed continuous evolution of preparatory neural activity during execution, supporting parallel planning of upcoming submovements. These findings represent a major scientific achievement, resolving a long-standing debate in the field and significantly extending the applicability of the dynamical systems framework beyond simple laboratory tasks.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The main results of MAPPING demonstrate that motor planning is supported by low-dimensional, invariant neural representations in premotor cortex that generalize across task constraints and behavioral contexts. The identification of a shared preparatory manifold underlying both constrained and unconstrained movements provides strong evidence that complex action sequences are planned incrementally, with future movement components prepared in parallel with ongoing execution. These results advance fundamental understanding of motor control and provide a unifying framework linking neural dynamics to flexible behavior.

In terms of potential impact, the project offers clear opportunities for further uptake in both basic and applied research. From a scientific perspective, follow-up work will require extending these approaches to additional cortical areas, larger behavioral repertoires, and—eventually—human recordings to test cross-species generality. From a translational standpoint, further development and demonstration are needed to integrate the decoding framework into real-time brain–computer interface systems, including validation under clinical constraints such as neural variability and long-term signal instability. Access to interdisciplinary collaborations, computational resources, and targeted funding will be key to supporting this transition.

While no immediate intellectual property has been generated, the simplicity, robustness, and generality of the decoding approach make it well suited for future commercialization efforts in neuroprosthetics and assistive technologies. Successful uptake will depend on continued methodological refinement, alignment with regulatory and ethical frameworks governing neural interfaces, and engagement with industrial and clinical partners. Overall, MAPPING delivers a coherent set of results with strong potential to influence both theoretical neuroscience and the next generation of adaptive neural decoding technologies.