Optimal control framework successfully explains changes in neural modulations during experiments with Brain Machine Interfaces Miriam Zacksenhouse, Koren Beiser: Technion, Haifa, Israel Joseph E. O'Doherty, Mikhail A. Lebedev, Miguel A.L. Nicolelis: Duke University, Durham, NC, 27710, USA Recent experiments with brain-machine-interfaces (BMIs) [1] indicate that the extent of neural modulations increases abruptly upon starting to operate the interface, especially after stopping hand movements [2]. In contrast, neural modulations that are correlated with the trajectories of the controlled cursor and target remain relatively unchanged. Here we demonstrate that similar changes are obtained from an optimal controller under simulated BMI experiments, and investigate the implied nature of the enhanced modulations.

BMI experiments: The initial BMI experiments are detailed in [1], and include three control modes: Pole control in which the monkey controlled the cursor on the screen using a hand held pole, and Brain control with and without hand movements (WH and WHO, respectively), when the cursor was controlled by the predicted velocity from the recorded neural activity while the monkey initially continued but eventually stopped moving the hand. Cortical neurons were recorded simultaneously from the primary motor area (M1), the pre-motor area (PMd), and somatosensory area (S1). The BMI experiments analyzed and modeled here are similar to those detailed in [1] except that they involved center-out reaching movements instead of reaching to randomly placed targets and the neural activity was used to predict the position rather than the velocity. Data analysis: Overall neural modulations are defined as the variance of the underlying rate, and expressed as a percentage of the variance of the spike counts. Under the assumption that the spike trains are realizations of doubly stochastic Poisson processes (DSPPs), the variance of the underlying rate Λ can be estimated from the statistics of the binned spike counts N , using: var(Λ ) = var( N ) − E ( N ) [2]. (Extension to dead-time modified DSPPs involves minor bias terms [3]). The percent task-related modulations are estimated by the coefficient of determination between the neural activity and a 2 seconds window of the trajectories and kinematics of the cursor and target. Modeling: An optimal controller was constructed for reaching movements using a combined goal that penalizes for both the control effort and deviation from the target. The parameters of the goal function were selected to match the velocity profile observed during the experiments, as shown in Figure 1. State estimation integrated both visual and proprioceptive feedback, using appropriate delays and noise variances, with predictions from a linear internal model. Neural activity was simulated for M1 and PMd neurons and used to determine the BMI filter during simulated pole control, and control the position of the cursor during simulated brain control. Brain control was simulated by increasing the noise variance of the cursor's movements, and brain control without hand movements was simulated by using also an infinitely high noise variance for the proprioceptive feedback. Results and conclusions: The activity of the simulated neurons demonstrates the same phenomena observed in the recorded data: the percent neural modulations is higher in brain control without hand movements than in pole control, while the percent task-related modulations remains relatively unchanged, as depicted in Figure 2. Further simulations indicate that the source of the enhanced modulations in simulated brain control is related to increasing variability of the predicted velocity and the control signals.

Experimental speed profile

Speed (Cm/sec)

40 20 0 0 40

0.5

1

1.5

Simulated speed profile

20 0 0

0.5

1

1.5

Time (sec)

Figure 1: Experimental and simulated speed profile

Simulation Results

Experimental Results Pole control Brain control WH Brain control WOH

15

PTM & POM %

PTMModulations & POM % (%) Percent

20

20

Percent15Overall Modulations

10

5

0

Pole control Brain control WH Brain control WOH

25

10 5

M1

PMd

0

M1

PMd

Percent Task (kinematics plus target) Modulations Figure 2: Percent overall modulations and percent task-related modulations in actual and simulated BMI experiments involving center-out task and position predictions. Simulated results capture well the observed changes in neural modulations: (i) Percent overall modulation is higher in brain control without hand movements than in pole control, and (ii) There is no matching increase in task-related modulations. Acknowledgements This research was supported by the Ashkenazy Handicap Research Fund for MZ and grants from DARPA, the James S. McDonnel Foundation, NIH and NSF to MALN. References 1. Carmena JM, Lebedev MA, Crist RE, O’Doherty JE, Santucci DM, et al., Learning to control a brain–machine interface for reaching and grasping by primates. PLoS Biol 1:193–208, 2003. 2. Zacksenhouse M., Lebedev MA., Carmena JM., O'Doherty JE., Henriquez CS., Nicolelis MAL. Cortical modulations increase during early sessions with Brain-Machine Interface, PLoSONE 2(7):e619, 2007. 3. Zacksenhouse M., Lebedev MA., Nicolelis MAL., Signal-to noise ratio of binned spike counts and the timescales of neural encoding, submitted 2009.

Optimal control framework successfully explains ...

during experiments with Brain Machine Interfaces ... Data analysis: Overall neural modulations are defined as the variance of the underlying rate, and expressed ...

56KB Sizes 5 Downloads 226 Views

Recommend Documents

OPTIMAL CONTROL SYSTEM.pdf
How optimal control problems are classified ? Give the practical examples for each classification. 10. b) Find the extremal for the following functional dt. 2t. x (t) J.

OPTIMAL CONTROL SYSTEMS.pdf
OPTIMAL CONTROL SYSTEMS.pdf. OPTIMAL CONTROL SYSTEMS.pdf. Open. Extract. Open with. Sign In. Main menu. Displaying OPTIMAL CONTROL ...

OPTIMAL CONTROL SYSTEMS.pdf
... time and fixed end state. problem. Indicate the different cases of Euler-Lagrange equation. 10. 2. a) Find the extremal of a functional. J(x) [ ] x (t) x (t) x (t) x (t) dt.

Probabilistic LMP Forecasting under AC Optimal Power Flow Framework
segment i is considered to be a straight (linear) line, with slope ai and intercept bi. The LMP–load curve is extended to include two extra segments in Fig. 2.

Unified Framework for Optimal Video Streaming
cesses (MDPs) with average–cost constraints to the prob- lem. Based on ... Internet, the bandwidth available to a streaming appli- cation is .... low the truly optimal scheduling policy. .... that is allowed by the network connection, or alternativ

Iterative Learning Control for Optimal Multiple-Point Tracking
on the system dynamics. Here, the improved accuracy in trajectory tracking results has led to the development of various control schemes, such as proportional ...

Optimal risk control and investment for Markov ...
Xin Zhanga †. Ming Zhoub. aSchool of ... Di Masi, Kabanov and Runggaldier (1994), Yao, Zhang, and Zhou (2001), Buffington and. Elliott (2002), Graziano and ...

Optimal Adaptive Feedback Control of a Network Buffer.
system to obtain a robust quasi optimal adaptive control law. Such an approach is used ..... therefore reduces to the tracking of the singular value xsing given by eq. (8). For the .... [7] I. Smets, G. Bastin, and J. Van Impe. Feedback stabilisation

Optimal Adaptive Feedback Control of a Network Buffer
American control conference 2005. Portland, Oregon, USA - Juin 8-10 2005. Optimal Adaptive Feedback Control of a Network Buffer – p.1/19 ...

The Euler approximation in state constrained optimal control
Apr 13, 2000 - IN STATE CONSTRAINED OPTIMAL CONTROL. A. L. DONTCHEV AND WILLIAM W. HAGER. Abstract. We analyze the Euler approximation to ...

Further results on linear nonquadratic optimal control ...
C.-H. Hsu is with the Department of Industrial Engineering and Management, ..... chine less the cumulative demand, i.e., total inventory or backlog.

PDF Dynamic Programming and Optimal Control, Vol. I ...
I, 4th Edition, read online Dynamic Programming and Optimal Control, Vol. .... been instrumental in the recent spectacular success of computer Go programs.

Variational optimal control technique for the tracking of ... - Irisa
IRISA/INRIA Campus de Beaulieu 35042 Rennes Cedex, France npapadak ... a term related to the discrepancy between the state vari- ables evolution law and ...

Evolution of Optimal ANNs for Non-Linear Control ...
recognition, data processing, filtering, clustering, blind signal separation, compression, system identification and control, pattern recognition, medical diagnosis, financial applications, data mining, visualisation and e-mail spam filtering [5], [4