FALCON 2.0: A USER-FRIENDLY PLATFORM FOR FLEXIBLE DESIGN OF CLOSED-LOOP NEUROSCIENCE EXPERIMENTS

Assessing the causality of transient neural events (e.g., hippocampal replay lasting 50-300 ms) requires closed-loop brain-computer interfaces (CL-BCIs), which can analyze up to a few hundred high-rate (>10 kHz) electrophysiological signals within a few milliseconds and deliver event-specific feedback stimulation. Although the open-source multi-threaded CL-BCI Falcon 1.0 meets this very stringent requirement, it is not user-friendly and still has a limited user base.
To overcome these limitations, we developed Falcon 2.0 (https://github.com/falcon-eyrie/falcon-core). Written in C++ like its predecessor, it can now be installed without familiarity with its code structure. To modernize the development workflow and enable a faster community-driven development, we improved its documentation, facilitated its integration with OpenEphys, added Continuous Integration/Continuous Deployment tools and re-tested its real-time performance.
Falcon 2.0 features a graphical user interface that enables the configuration of processing graphs via an intuitive drag-and-drop tool, eliminating the need for text-based graphs. Processors are now fully modular, allowing for the independent addition and removal of components during installation. Real-time capabilities are still maintained with an internal latency of 153 µs (99% confidence interval: [132, 233] µs), which is comparable with that one of Falcon 1.0.
Finally, Falcon 2.0 provides a flexible BCI platform for reproducible low-latency CL experiments, now accessible to neuroscientists with limited technical background; simultaneously, it serves as an extensible framework for neuroengineers to build upon. Thus far exclusively used in animal laboratories, we now envisage easier adoption also in the context of human electrophysiology.