Jose M Carmena

University of California, Berkeley, Berkeley, California, United States

Are you Jose M Carmena?

Claim your profile

Publications (121)402.89 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Evidence suggests that the CNS uses motor primitives to simplify movement control, but whether it actually stores primitives instead of computing solutions on the fly to satisfy task demands is a controversial and still-unanswered possibility. Also in contention is whether these primitives take the form of time-invariant muscle coactivations (“spatial” synergies) or time-varying muscle commands (“spatiotemporal” synergies). Here, we examined forelimb muscle patterns and motor cortical spiking data in rhesus macaques (Macaca mulatta) handling objects of variable shape and size. From these data, we extracted both spatiotemporal and spatial synergies using non-negative decomposition. Each spatiotemporal synergy represents a sequence of muscular or neural activations that appeared to recur frequently during the animals’ behavior. Key features of the spatiotemporal synergies (including their dimensionality, timing, and amplitude modulation) were independently observed in the muscular and neural data. In addition, both at the muscular and neural levels, these spatiotemporal synergies could be readily reconstructed as sequential activations of spatial synergies (a subset of those extracted independently from the task data), suggestive of a hierarchical relationship between the two levels of synergies. The possibility that motor cortexmayexecute even complex skill using spatiotemporal synergies has novel implications for the design of neuroprosthetic devices, which could gain computational efficiency by adopting the discrete and low-dimensional control that these primitives imply.
    No preview · Article · Sep 2015 · The Journal of Neuroscience : The Official Journal of the Society for Neuroscience
  • Preeya Khanna · Jose M Carmena
    [Show abstract] [Hide abstract]
    ABSTRACT: Local field potential (LFP) activity in motor cortical and basal ganglia regions exhibits prominent beta (15-40Hz) oscillations during reaching and grasping, muscular contraction, and attention tasks. While in vitro and computational work has revealed specific mechanisms that may give rise to the frequency and duration of this oscillation, there is still controversy about what behavioral processes ultimately drive it. Here, simultaneous behavioral and large-scale neural recording experiments from non-human primate and human subjects are reviewed in the context of specific hypotheses about how beta band activity is generated. Finally, a new experimental paradigm utilizing operant conditioning combined with motor tasks is proposed as a way to further investigate this oscillation. Copyright © 2014 Elsevier Ltd. All rights reserved.
    No preview · Article · Jun 2015 · Current Opinion in Neurobiology
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Movements made by healthy individuals can be characterized as superpositions of smooth bell-shaped velocity curves. Decomposing complex movements into these simpler "submovement" building blocks is useful for studying the neural control of movement as well as measuring motor impairment due to neurological injury. One prevalent strategy to submovement decomposition is to formulate it as an optimization problem. This optimization problem is non-convex and finding an exact solution is computationally burdensome. We build on previous literature which generated approximate solutions to the submovement optimization problem. First, we demonstrate broad conditions on the submovement building block functions that enable the optimization variables to be partitioned into disjoint subsets, allowing for a faster alternating minimization solution. Specifically, the amplitude parameters of a submovement can typically be fit independently of its shape parameters. Second, we develop a method to concentrate the search in regions of high error to make more efficient use of optimization routine iterations. Both innovations result in substantial reductions in computation time across multiple non-human primate subjects and diverse task conditions. These innovations may accelerate analysis of submovements for basic neuroscience and enable realtime applications of submovement decomposition.
    Full-text · Article · May 2015 · IEEE Transactions on Biomedical Engineering
  • [Show abstract] [Hide abstract]
    ABSTRACT: A 65 nm CMOS 4.78 mm 2 integrated neuromodulation SoC consumes 348 µA from an unregulated 1.2 V to 1.8 V supply while operating 64 acquisition channels with epoch compression at an average firing rate of 50 Hz and engaging two stimulators with a pulse width of 250 µs/phase, differential current of 150 µA, and a pulse frequency of 100 Hz. Compared to the state of the art, this represents the lowest area and power for the highest integration complexity achieved to date.
    No preview · Article · Apr 2015 · IEEE Journal of Solid-State Circuits
  • [Show abstract] [Hide abstract]
    ABSTRACT: Brain-machine interface (BMI) technology has tremendous potential to revolutionize healthcare by greatly im-proving the quality of life of millions of people suffering from a wide variety of neurological conditions. Radio-frequency identi-fication (RFID)-inspired backscattering is a promising approach for wireless powering of miniature neural sensors required in BMI interfaces. We analyze the functionality of millimeter-size loop antennas in the wireless powering of miniature cortical implants through measurements ina human head equivalent liquid phantom and in the head of a postmortem pig. For the first time, we present the design and measurement of a miniature 1x1x1 mm3 backscattering device based on a cubic loop connected with an RFID integrated circuit (IC). Our measurement results show that this very small loop receives sufficient electro-magnetic powertoactivatethe IC whenthedeviceisimplanted in a pig's head. This demonstrates the feasibility of extremely small implant antennas in challenging wireless biomedical systems
    No preview · Article · Feb 2015 · IEEE Transactions on Antennas and Propagation
  • [Show abstract] [Hide abstract]
    ABSTRACT: Emerging applications in brain-machine interface systems require high-resolution, chronic multisite cortical recordings, which cannot be obtained with existing technologies due to high power consumption, high invasiveness, or inability to transmit data wirelessly. In this paper, we describe a microsystem based on electrocorticography (ECoG) that overcomes these difficulties, enabling chronic recording and wireless transmission of neural signals from the surface of the cerebral cortex. The device is comprised of a highly flexible, high-density, polymer-based 64-channel electrode array and a flexible antenna, bonded to 2.4 mm x 2.4 mm CMOS integrated circuit (IC) that performs 64-channel acquisition, wireless power and data transmission. The IC digitizes the signal from each electrode at 1 kS/s with 1.2 μV input referred noise, and transmits the serialized data using a 1 Mb/s backscattering modulator. A dual-mode power-receiving rectifier reduces data-dependent supply ripple, enabling the integration of small decoupling capacitors on chip and eliminating the need for external components. Design techniques in the wireless and baseband circuits result in over 16x reduction in die area with a simultaneous 3x improvement in power efficiency over the state of the art. The IC consumes 225 μW and can be powered by an external reader transmitting 12 mW at 300 MHz, which is over 3x lower than IEEE and FCC regulations.
    No preview · Article · Jan 2015 · IEEE Journal of Solid-State Circuits
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Brain-machine interfaces (BMI) hold great po-tential to improve the quality of life of many patients with disabilities. The neural decoder, which expresses the mapping between the neural signals and the subject's motion, plays an important role in BMI systems. Conventional neural decoders are generally in the form of a kinematic Kalman filter which does not possess an explicit mechanism to deal with the unavoid-able mismatch between the biological system and the model of the system used by the decoder. This paper presents a novel design of a neural decoder that uses a one-step model predictive controller to generate a control signal that compensates for the inherent model mismatch. The effectiveness of the proposed decoding algorithm compares favorably to the state-of-the-art Kalman filter in numerical simulations with different degrees of model mismatch.
    Full-text · Conference Paper · Dec 2014
  • Krishna V Shenoy · Jose M Carmena
    [Show abstract] [Hide abstract]
    ABSTRACT: Brain-machine interfaces (BMIs) aim to help people with paralysis by decoding movement-related neural signals into control signals for guiding computer cursors, prosthetic arms, and other assistive devices. Despite compelling laboratory experiments and ongoing FDA pilot clinical trials, system performance, robustness, and generalization remain challenges. We provide a perspective on how two complementary lines of investigation, that have focused on decoder design and neural adaptation largely separately, could be brought together to advance BMIs. This BMI paradigm should also yield new scientific insights into the function and dysfunction of the nervous system. VIDEO ABSTRACT: Copyright © 2014 Elsevier Inc. All rights reserved.
    No preview · Article · Nov 2014 · Neuron
  • [Show abstract] [Hide abstract]
    ABSTRACT: A major hurdle in brain-machine interfaces (BMI) is the lack of an implantable neural interface system that remains viable for a substantial fraction of the user's lifetime. Recently, sub-mm implantable, wireless electromagnetic (EM) neural interfaces have been demonstrated in an effort to extend system longevity. However, EM systems do not scale down in size well due to the severe inefficiency of coupling radio-waves at those scales within tissue. This paper explores fundamental system design trade-offs as well as size, power, and bandwidth scaling limits of neural recording systems built from low-power electronics coupled with ultrasonic power delivery and backscatter communication. Such systems will require two fundamental technology innovations: 1) 10-100μm scale, free-floating, independent sensor nodes, or neural dust, that detect and report local extracellular electrophysiological data via ultrasonic backscattering, and 2) a sub-cranial ultrasonic interrogator that establishes power and communication links with the neural dust. We provide experimental verification that the predicted scaling effects follow theory; (127μm)(3) neural dust motes immersed in water 3cm from the interrogator couple with 0.002064% power transfer efficiency and 0.04246ppm backscatter, resulting in a maximum received power of ∼0.5μW with ∼1 nW of change in backscatter power with neural activity. The high efficiency of ultrasonic transmission can enable the scaling of the sensing nodes down to 10's of μm. We conclude with a brief discussion of the application of neural dust for both central and peripheral nervous system recordings, and perspectives on future research directions.
    No preview · Article · Aug 2014 · Journal of Neuroscience Methods
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we examine the use of beamforming techniques to interrogate a multitude of neural implants in a distributed, ultrasound-based intra-cortical recording platform known as Neural Dust [1]. We propose a general framework to analyze system design tradeoffs in the ultrasonic beamformer that extracts neural signals from modulated ultrasound waves that are backscattered by free-floating neural dust (ND) motes. Simulations indicate that high-resolution linearly-constrained minimum variance beamforming sufficiently suppresses interference from unselected ND motes and can be incorporated into the ND-based cortical recording system.
    No preview · Article · Aug 2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: Brain-machine interface (BMI) performance has been improved using Kalman filters (KF) combined with closed-loop decoder adaptation (CLDA). CLDA fits the decoder parameters during closed-loop BMI operation based on the neural activity and inferred user velocity intention. These advances have resulted in the recent ReFIT-KF and SmoothBatch-KF decoders. Here we demonstrate high-performance and robust BMI control using a novel closed-loop BMI architecture termed adaptive optimal feedback-controlled (OFC) point process filter (PPF). Adaptive OFC-PPF allows subjects to issue neural commands and receive feedback with every spike event and hence at a faster rate than the KF. Moreover, it adapts the decoder parameters with every spike event in contrast to current CLDA techniques that do so on the time-scale of minutes. Finally, unlike current methods that rotate the decoded velocity vector, adaptive OFC-PPF constructs an infinite-horizon OFC model of the brain to infer velocity intention during adaptation. Preliminary data collected in a monkey suggests that adaptive OFC-PPF improves BMI control. OFC-PPF outperformed SmoothBatch-KF in a self-paced center-out movement task with 8 targets. This improvement was due to both the PPF's increased rate of control and feedback compared with the KF, and to the OFC model suggesting that the OFC better approximates the user's strategy. Also, the spike-by-spike adaptation resulted in faster performance convergence compared to current techniques. Thus adaptive OFC-PPF enabled proficient BMI control in this monkey.
    No preview · Article · Aug 2014

  • No preview · Conference Paper · Jul 2014
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Brain-Machine Interfaces (BMI) have strong po-tential to benefit a large number of disabled people but current decoding algorithms suffer from the following shortcomings. First, BMI decoding algorithms are often trained offline, but this paradigm ignores the discrepancy between the Manual Control (MC) and the Brain Control (BC) modes of operation. Second, the standard neural decoder, the Kalman filter, does not explicitly take into account the control of movements by neural activity. To address these problems, we propose a biologically motivated neural decoder structure by explicitly adding a control signal and unmeasureable neural activity. Since the parameter estimation problem is underdetermined, we propose a new parameter estimation method that minimizes the discrepancy between the MC and BC. We demonstrate the effectiveness of our methods by synthesizing MC and BC data in a Linear Quadratic (LQ) optimal control setting with a partial loss of neural control in BC, and show that the proposed decoder is more robust to a partial loss of neural control than a standard Kalman filter that does not utilize any reparameterizations.
    Full-text · Conference Paper · Jun 2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: Neuroplasticity may play a critical role in developing robust, naturally controlled neuroprostheses. This learning, however, is sensitive to system changes such as the neural activity used for control. The ultimate utility of neuroplasticity in real-world neuroprostheses is thus unclear. Adaptive decoding methods hold promise for improving neuroprosthetic performance in nonstationary systems. Here, we explore the use of decoder adaptation to shape neuroplasticity in two scenarios relevant for real-world neuroprostheses: nonstationary recordings of neural activity and changes in control context. Nonhuman primates learned to control a cursor to perform a reaching task using semistationary neural activity in two contexts: with and without simultaneous arm movements. Decoder adaptation was used to improve initial performance and compensate for changes in neural recordings. We show that beneficial neuroplasticity can occur alongside decoder adaptation, yielding performance improvements, skill retention, and resistance to interference from native motor networks. These results highlight the utility of neuroplasticity for real-world neuroprostheses.
    No preview · Article · Jun 2014 · Neuron
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Closed-loop decoder adaptation (CLDA) is an emerging paradigm for both improving and maintaining online performance in brain-machine interfaces (BMIs). The time required for initial decoder training and any subsequent decoder recalibrations could be potentially reduced by performing continuous adaptation, in which decoder parameters are updated at every time step during these procedures, rather than waiting to update the decoder at periodic intervals in a more batch-based process. Here, we present recursive maximum likelihood (RML), a CLDA algorithm that performs continuous adaptation of a Kalman filter decoder's parameters. We demonstrate that RML possesses a variety of useful properties and practical algorithmic advantages. First, we show how RML leverages the accuracy of updates based on a batch of data while still adapting parameters on every time step. Second, we illustrate how the RML algorithm is parameterized by a single, intuitive half-life parameter that can be used to adjust the rate of adaptation in real time. Third, we show how even when the number of neural features is very large, RML's memory-efficient recursive update rules can be reformulated to be computationally fast so that continuous adaptation is still feasible. To test the algorithm in closed-loop experiments, we trained three macaque monkeys to perform a center-out reaching task by using either spiking activity or local field potentials to control a 2D computer cursor. RML achieved higher levels of performance more rapidly in comparison to a previous CLDA algorithm that adapts parameters on a more intermediate timescale. Overall, our results indicate that RML is an effective CLDA algorithm for achieving rapid performance acquisition using continuous adaptation.
    Preview · Article · Jun 2014 · Neural Computation
  • [Show abstract] [Hide abstract]
    ABSTRACT: Wireless body-centric sensing systems have an important role in the fields of biomedicine, personal healthcare, safety, and security. Body-centric radio-frequency identification (RFID) technology provides a wireless and maintenance-free communication link between the human body and the surroundings through wearable and implanted antennas. This enables real-time monitoring of human vital signs everywhere. Seamlessly integrated wearable and implanted miniaturized antennas thus have the potential to revolutionize the everyday life of people, and to contribute to independent living. Low-cost and low-power system solutions will make widespread use of such technology become reality. The primary target applications for this research are body-centric sensing systems and the relatively new interdisciplinary field of wireless brain-machine interface (BMI) systems. Providing a direct wireless pathway between the brain and an external device, a wireless brain-machine interface holds an enormous potential for helping people suffering from severely disabling neurological conditions to communicate and manage their everyday life more independently. In this paper, we discuss RFID-inspired wireless brain-machine interface systems. We demonstrate that mm-size loop implanted antennas are capable of efficiently coupling to an external transmitting loop antenna through an inductive link. In addition, we focus on wearable antennas based on electrically conductive textiles and threads, and present design guidelines for their use as wearable-antenna conductive elements. Overall, our results constitute an important milestone in the development of wireless brain-machine interface systems, and a new era of wireless body-centric systems.
    No preview · Article · Jun 2014 · IEEE Antennas and Propagation Magazine
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Brain-machine interfaces are not only promising for neurological applications, but also powerful for investigating neuronal ensemble dynamics during learning. We trained mice to operantly control an auditory cursor using spike-related calcium signals recorded with two-photon imaging in motor and somatosensory cortex. Mice rapidly learned to modulate activity in layer 2/3 neurons, evident both across and within sessions. Learning was accompanied by modifications of firing correlations in spatially localized networks at fine scales.
    Full-text · Article · Apr 2014 · Nature Neuroscience
  • [Show abstract] [Hide abstract]
    ABSTRACT: Brain-machine interfaces (BMIs) are dynamical systems whose properties ultimately influence performance. For instance, a 2D BMI in which cursor position is controlled using a Kalman filter (KF) will, by default, create an attractor point that "pulls" the cursor to particular points in the workspace. If created unintentionally, such effects may not be beneficial for BMI performance. However, there have been few empirical studies exploring how various dynamical effects of closed-loop BMIs ultimately influence performance. In this work, we utilize experimental data from two macaque monkeys operating a closed-loop BMI to reach to 2D targets and show that certain dynamical properties correlate with performance loss. We also show that other dynamical properties represent tradeoffs between naturally competing objectives, such as speed versus accuracy. These findings highlight the importance of fine-tuning the dynamical properties of closed-loop BMIs to optimize taskspecific performance.
    No preview · Article · Mar 2014 · IEEE transactions on neural systems and rehabilitation engineering: a publication of the IEEE Engineering in Medicine and Biology Society
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Electrical microstimulation studies provide some of the most direct evidence for the neural representation of muscle synergies. These synergies, i.e., coordinated activations of groups of muscles, have been proposed as building blocks for the construction of motor behaviors by the nervous system. Intraspinal or intracortical microstimulation (ICMS) has been shown to evoke muscle patterns that can be resolved into a small set of synergies similar to those seen in natural behavior. However, questions remain about the validity of microstimulation as a probe of neural function, particularly given the relatively long trains of supratheshold stimuli used in these studies. Here, we examined whether muscle synergies evoked during ICMS in two rhesus macaques were similarly encoded by nearby motor cortical units during a purely voluntary behavior involving object reach, grasp, and carry movements. At each microstimulation site we identified the synergy most strongly evoked among those extracted from muscle patterns evoked over all microstimulation sites. For each cortical unit recorded at the same microstimulation site, we then identified the synergy most strongly encoded among those extracted from muscle patterns recorded during the voluntary behavior. We found that the synergy most strongly evoked at an ICMS site matched the synergy most strongly encoded by proximal units more often than expected by chance. These results suggest a common neural substrate for microstimulation-evoked motor responses and for the generation of muscle patterns during natural behaviors.
    Full-text · Article · Mar 2014 · Frontiers in Computational Neuroscience
  • [Show abstract] [Hide abstract]
    ABSTRACT: Substantial improvements in neural-implant longevity are needed to transition brain-machine interface (BMI) systems from research labs to clinical practice. While action potential (AP) recording through penetrating electrode arrays offers the highest spatial resolution, it comes at the price of tissue scarring, resulting in signal degradation over the course of several months [1]. Electrocorticography (ECoG) is an electrophysiological technique where electrical potentials are recorded from the surface of the cerebral cortex, reducing cortical scarring. However, today's clinical ECoG implants are large, have low spatial resolution (0.4 to 1cm) and offer only wired operation.
    No preview · Conference Paper · Feb 2014

Publication Stats

4k Citations
402.89 Total Impact Points


  • 2007-2015
    • University of California, Berkeley
      • Department of Electrical Engineering and Computer Sciences
      Berkeley, California, United States
  • 2013
    • University of Melbourne
      • Department of Electrical and Electronic Engineering
      Melbourne, Victoria, Australia
  • 2010
    • Atomic Energy and Alternative Energies Commission
      Fontenay, Île-de-France, France
  • 2003-2007
    • Duke University
      • Department of Biomedical Engineering (BME)
      Durham, North Carolina, United States
  • 2000-2002
    • The University of Edinburgh
      • • School of Informatics
      • • Institute of Perception, Action and Behaviour (IPAB)
      Edinburgh, Scotland, United Kingdom
  • 2001
    • Duke University Medical Center
      • Department of Neurobiology
      Durham, North Carolina, United States