Propagating spatio-temporal neural patterns are widely evident across sensory, motor and association cortical areas. However, it remains unclear whether any characteristics of neural propagation carry information about specific behavioral details. Here, we provide the first evidence for a link between the direction of cortical propagation and specific behavioral features of an upcoming movement on a trial-by-trial basis. We recorded local field potentials (LFPs) from multi-electrode arrays implanted in the primary motor cortex of two rhesus macaque monkeys while they performed a 2-D reach task. Propagating patterns were extracted from the information-rich high-gamma band (200–400Hz) envelopes in the LFP amplitude. We found that the exact direction of propagating patterns varied systematically according to initial movement direction, enabling kinematic predictions. Furthermore, characteristics of these propagation patterns provided additional predictive capability beyond the LFP amplitude themselves, which suggests the value of including mesoscopic spatio-temporal characteristics in refining brain-machine interfaces.

Electrophysiology

Neural recordings were collected from two male rhesus macaques (Macaca mulatta), Bx and Ls. They were each implanted with a set of dual 64-electrode Utah multi-electrode arrays (Blackrock Microsystems, Salt Lake City, UT) in the left primary motor cortex. The electrode lengths were 1.5mm for Bx and 1mm for Ls, with a uniform inter-electrode distance of 400µm in the 8-by-8 grid. The arrays corresponded to the arm/hand areas of the limb trained for the task, as confirmed by electrically stimulating the cortex during surgery with surface electrodes and observing corresponding twitches prior to array implantation. All procedures regarding surgery, animal training and data collection were approved by the University of Chicago Institutional Animal Care and Use Committee (IACUC) and obey the Guide for the Care and Use of Laboratory Animals.

Behavioral task

The animals were trained to perform a reach task in different directions with their right upper limb constrained by a 2D exoskeletal robot (BKIN Technologies). All the reach directions were on a horizontal plane roughly at the height of their elbows when their upper limbs were naturally hanging. They performed the task by moving the joystick at the tip of the robot, while receiving simultaneous visual feedback of the cursor position on the screen in front of them. The task involves a hold period (600ms for Ls and 1000ms for Bx) during which the animal held the joystick steadily keeping the cursor within a central target. The peripheral target then appeared indicating the desired reach location. The animal moved the joystick to direct the cursor to the peripheral target. Each target had a radius of 1cm for Ls and 0.75cm for Bx; the distance from the central target to the peripheral target was 6cm For Ls and 5.5cm for Bx.  The peripheral target could appear in one of 8 possible locations on a circle (in this dataset, 1 means forward, 2 means rightward-forward, 3 means rightward, etc.). A trial was considered successful once the cursor reached the peripheral target and remained in the target for a period of time (100ms for Ls, 400–600ms for Bx), upon which the animal received a juice reward. In total, there were 3 sessions for each monkey, indicated by the date in the file name.

Data collection and preprocessing

Neural signals were collected with Blackrock Microsystems (Salt Lake City, UT), and bandpass was filtered from 0.3Hz to 7.5 kHz with Butterworth filter and then digitized at 30kHz. For local field potentials, the 30kHz signal was further low-pass filtered below 500Hz with a Butterworth filter and downsampled to 2kHz. From the 2kHz local field potentials, the high-gamma band signals were extracted by bandpass filtering within 200–400Hz with a Butterworth filter. From those band-pass filtered signals, we computed the Hilbert transformation to obtain the amplitude envelopes. Individual trials were aligned on movement onset defined as the time point where 15% of the peak speed was reached for that trial.

Please see this GitHub repo for python code

https://github.com/hatsopoulos-lab/macaque-spatio_temporal_pattern

Please cite this paper if you are using this dataset. 

Liang, W., Balasubramanian, K., Papadourakis, V., Hatsopoulos, N. G. (2023) Propagating spatio-temporal activity patterns across macaque motor cortex carry kinematic information. Proceedings of the National Academy of Sciences