SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Report presents theoretical study of gains required in neurons to implement winner-take-all electronic neural network of given size and related question of maximum size of winner-take-all network in which neurons have specified sigmoid transfer or response function with specified gain.
An integrated, optoelectronic, variable thresholding neuron implemented monolithically in GaAs integrated circuit and exhibiting high differential optical gain and low power consumption is presented. Two alternative embodiments each comprise an LED monolithically integrated with a detector and two transistors. One of the transistors is responsive to a bias voltage applied to its gate for varying the threshold of the neuron. One embodiment is implemented as an LED monolithically integrated with a double heterojunction bipolar phototransistor (detector) and two metal semiconductor field effect transistors (MESFET's) on a single GaAs substrate and another embodiment is implemented as an LED monolithically integrated with three MESFET's (one of which is an optical FET detector) on a single GaAs substrate. The first noted embodiment exhibits a differential optical gain of 6 and an optical switching energy of 10 pJ. The second embodiment has a differential optical gain of 80 and an optical switching energy of 38 pJ. Power consumption is 2.4 and 1.8 mW, respectively. Input 'light' power needed to turn on the LED is 2 micro-W and 54 nW, respectively. In both embodiments the detector is in series with a biasing MESFET and saturates the other MESFET upon detecting light above a threshold level. The saturated MESFET turns on the LED. Voltage applied to the biasing MESFET gate controls the threshold.
A novel mathematical framework for the rapid learning of nonlinear mappings and topological transformations is presented. It is based on allowing the neuron's parameters to adapt as a function of learning. This fully recurrent adaptive neuron model (ANM) has been successfully applied to complex nonlinear function approximation problems such as the highly degenerate inverse kinematics problem in robotics.
This invention is an adaptive neuron for use in neural network processors. The adaptive neuron participates in the supervised learning phase of operation on a co-equal basis with the synapse matrix elements by adaptively changing its gain in a similar manner to the change of weights in the synapse IO elements. In this manner, training time is decreased by as much as three orders of magnitude.
During the past year, pre-, in- and postflight studies were conducted in association with the Axon project for Bion 10 (Cosmos 2229). Recordings were made during pre- and postflight studies, from 118 horizontal semicircular canal afferents and 27 vestibular nucleus neurons in 7 rhesus monkeys; 137 pulse rotation protocols alone were executed (548 acceleration and deceleration responses were curve fit). Usable data was obtained from 127 horizontal afferents concerning their spontaneous discharge. Curve fits and analysis was made of sinusoidal and sum of sinusoidal responses from 42 and 35 horizontal afferents, respectively. Also recordings were made from neurons inflight from the two flight animals. The mean spontaneous rate varied from 128 spikes/sec. during preflight to 92 spikes/sec during postflight (day 5) - a change of 28%. In direct contrast to the results of Cosmos 2044, the best fitted neural adaptation operator (k) and the gain of the pulse response were decreased during post flight when compared to preflight. Surprisingly, the best fitted gain and k values for the sum of sines were slightly elevated during post flight tests. The gain and phase of single sine responses were compared for pre- and post flight tests and compared to a larger population of afferents. In contrast to Cosmos 2044 results where on the first day of post flight testing the gains of the best fitted sine response were skewed toward the higher values of the Miles and Braitman distribution, the gain of the best fitted sine responses during the first day of post flight testing (day 2) during Cosmos 2229 were exactly on the mode of the Miles and Braitman distribution. Thus, at least for the periodic stimuli, (pulses and sine waves) we found no change in gain and neural adaptation during post flight testing following Cosmos 2229. This conclusion is different from the one derived following the Cosmos 2044 flight. Cosmos flight 2229 differed from Cosmos flight 2044 in several significant ways: For example, during preflight, (1) The animals preflight training was different (less well trained on the gaze task) and (2) the animals were exposed to more experimental manipulations (surgical and rotational). Inflight, (1) the animals were required to make a pointing gesture (motor response) in association with eye movements to obtain reward, (2) the inflight diet was different (more balanced), (3) the feeder for one of the animals clogged following 9 days of flight resulting in evident dehydration and probably less head motion exposure in that monkey and (4) there was limited video taping of the monkeys in space. During postflight, (1) we were unable to test the flight animals until 26 hours postflight as compared to 14.5 hours during Cosmos 2044, (2) the animals received significantly more exposure to motion stimuli during postflight testing than during Cosmos 2044. These differences in the vestibular environment will require analysis of several parameters other than just neural and eye movement responses. For example, computer programs will have to be written and used to recover and quantify the number of head movements made by each animal during flight. This activity is critical to the production of neural adaptation and increased gain.
A template model of human visual self-motion perception (Perrone, JOSA, 1992; Perrone & Stone, Vis. Res., in press), which uses neurophysiologically realistic "heading detectors", is consistent with numerous human psychophysical results (Warren & Hannon, Nature, 1988; Stone & Perrone, Neuro. Abstr., 1991) including the failure of humans to estimate their heading (direction of forward translation) accurately under certain visual conditions (Royden et al., Nature, 1992). We tested the model detectors with stimuli used by others in- single-unit studies. The detectors showed emergent properties similar to those of MST neurons: 1) Sensitivity to non-preferred flow. Each detector is tuned to a specific combination of flow components and its response is systematically reduced by the addition of nonpreferred flow (Orban et al., PNAS, 1992), and 2) Position invariance. The detectors maintain their apparent preference for particular flow components over large regions of their receptive fields (e.g. Duffy & Wurtz, J. Neurophys., 1991; Graziano et al., J. Neurosci., 1994). It has been argued that this latter property is incompatible with MST playing a role in heading perception. The model however demonstrates how neurons with the above response properties could still support accurate heading estimation within extrastriate cortical maps.
A template model of human visual self-motion perception, which uses neurophysiologically realistic "heading detectors", is consistent with numerous human psychophysical results including the failure of humans to estimate their heading (direction of forward translation) accurately under certain visual conditions. We tested the model detectors with stimuli used by others in single-unit studies. The detectors showed emergent properties similar to those of MST neurons: (1) Sensitivity to non-preferred flow; Each detector is tuned to a specific combination of flow components and its response is systematically reduced by the addition of nonpreferred flow, and (2) Position invariance; The detectors maintain their apparent preference for particular flow components over large regions of their receptive fields. It has been argued that this latter property is incompatible with MST playing a role in heading perception. The model however demonstrates how neurons with the above response properties could still support accurate heading estimation within extrastriate cortical maps.
The grant focuses on a gene we identified called, serum and glucocorticoid- induced kinase (SGK), during a previously funded NASA project. The abundance of SGK messenger RNA (mRNA) and protein is increased in CNS tissues from animals reared in microgravity in comparison with 1G reared animals. In the funded proposal we had three aims: 1) characterize the distribution of SGK mRNA in the developing and adult rat CNS, 2) determine if expression of enzymatically active or inactive forms of SGK in cells influenced cell morphology (neurite growth), and 2) determine if SGK is a CREB kinase - that is, a protein kinase that adds phosphate groups to the transcription factor CREB. Over the past year we have made strong progress in the two most difficult parts of the project, namely specific aims 2 and 3. In specific aim #2 we planned to express a dominant negative or a constitutively active form of SGK in PC12 cells and assay the effects on neurite growth. Several methods are available for examining the effects of a transgene on PC12 neurite growth. Relevant variables include the performance of the assay +/- serum, +/- NGF, substratum for growth, timing between transfection and assay. Over the past 8 months we have customized the assay to enable us to most readily determine the effects of transgene expression on neurite growth. We have also compared the relative utility of transfecting DNA as opposed to protein itself. We are now well positioned to study the effects of SGK on neurite growth. We have also made progress in parallel studies in primary neurons. We have made constructs which will lead to transgene expression in cultures of spinal cord neurons. Co-transfection of a reporter and the SGK constructs can now be performed.
Receptors for the Fc portion of immunoglobulin G (IgG; FcgammaRs) facilitate IgG uptake by effector cells as well as cellular responses initiated by IgG binding. In earlier studies, we demonstrated that amyotrophic lateral sclerosis (ALS) patient IgG can be taken up by motor neuron terminals and transported retrogradely to the cell body and can alter the function of neuromuscular synapses, such as increasing intracellular calcium and spontaneous transmitter release from motor axon terminals after passive transfer. In the present study, we examined whether FcgammaR-mediated processes can contribute to these effects of ALS patient immunoglobulins. F(ab')(2) fragments (which lack the Fc portion) of ALS patient IgG were not taken up by motor axon terminals and were not retrogradely transported. Furthermore, in a genetically modified mouse lacking the gamma subunit of the FcR, the uptake of whole ALS IgG and its ability to enhance intracellular calcium and acetylcholine release were markedly attenuated. These data suggest that FcgammaRs appear to participate in IgG uptake into motor neurons as well as IgG-mediated increases in intracellular calcium and acetylcholine release from motor axon terminals. Copyright 2002 Wiley-Liss, Inc.
Although humans hold great advantages over other species as subjects for biomedical research, they also bring major disadvantages. One is that among the many rhythmic physiological signals that can be recorded, there is no sure way to know which individual change precedes another, or which change represents cause and which represents effect. In an attempt to deal with the inherent complexity of research conducted in intact human subjects, we developed and used a structural equation model to analyse responses of healthy young men to pharmacological changes of arterial pressure and graded inspiratory resistance, before and after vagomimetic atropine. Our model yielded a good fit of the experimental data, with a system weighted R2 of 0.77, and suggested that our treatments exerted both direct and indirect influences on the variables we measured. Thus, infusions of nitroprusside and phenylephrine exerted all of their direct effects by lowering and raising arterial pressure; the changes of R-R intervals, respiratory sinus arrhythmia and arterial pressure fluctuations that these drugs provoked, were indirect consequences of arterial pressure changes. The only direct effect of increased inspiratory resistance was augmentation of arterial pressure fluctuations. These results may provide a new way to disentangle and understand responses of intact human subjects to experimental forcings. The principal new insight we derived from our modelling is that respiratory gating of vagal-cardiac motor neurone firing is nearly maximal at usual levels of arterial pressure and inspiratory motor neurone activity.
Locomotor activity in many species undergoes pronounced alterations in early postnatal life, and environmental cues may be responsible for modifying this process. To determine how these events are reflected in the nervous system, we studied rats reared under two different conditions-the presence or absence of gravity-in which the performance of motor operations differed. We found a significant effect of rearing environment on the size and complexity of dendritic architecture of spinal motor neurons, particularly those that are likely to participate in postural control. These results provide evidence that neurons subserving motor function undergo activity-dependent maturation in early postnatal life in a manner analogous to sensory systems.
We tested the hypothesis that exposure to microgravity reduces the neuronal release of catecholamines and blood pressure responses to norepinephrine and angiotensin. Eight men underwent 30 days of 6 degrees head-down tilt (HDT) bedrest to simulate exposure to microgravity. Plasma norepinephrine and mean arterial blood pressure (MAP) were measured before and after a cold pressor test (CPT) and graded norepinephrine infusion (8, 16 and 32 ng/kg/min) on day 6 of a baseline control period (C6) and on days 14 and 27 of HDT. MAP and plasma angiotensin II (Ang-II) were measured during graded Ang-II infusion (1, 2 and 4 ng/kg/min) on C8 and days 16 and 29 of HDT. Baseline total circulating norepinephrine was reduced from 1017ng during the baseline control period to 610 ng at day 14 and 673ng at day 27 of HDT, confirming a hypoadrenergic state. An elevation of norepinephrine (+178 ng) to the CPT during the baseline control period was eliminated by HDT days 14 and 27. During norepinephrine infusion, similar elevations in plasma norepinephrine (7.7 pg/ml/ng/kg/min) caused similar elevations in MAP (0.12 mmHg/ng/kg/min) across all test days. Ang-II infusion produced higher levels of plasma Ang-II during HDT (47.3 pg/ml) than during baseline control (35.5 pg/ml), while producing similar corresponding elevations in blood pressure. While vascular responsiveness to norepinephrine appears unaffected, impaired neuronal release of norepinephrine and reduced vascular responsiveness to Ang-II might contribute to the lessened capacity to vasoconstrict after spaceflight. The time course of alterations indicates effects that occur within two weeks of exposure.
We have previously reported that, in the rat dorsolateral septal nucleus (DLSN), metabotropic glutamate receptor (met-GluR) agonists evoked a slow depolarization accompanied by an increase in membrane conductance and burst firing. We have speculated that the burst firing elicited by met-GluR agonists may be due to activation or enhancement of a non-specific cation current, which exists in some DLSN neurons. Now we report that a slow afterdepolarization (sADP) mediated by a non-specific cation current was potentiated by both 1S,3R-ACPD and quisqualate. In addition, met-GluR agonists unmask a sADP in DLSN neurons which did not show a sADP under control conditions. Our data suggest that a non-specific cation current can be potentiated by activation of the met-GluR.
We introduce a method for extracting and analyzing neuronal activity time signals from video of the cortex of a live animal. The signals correspond to the firing activity of individual cortical neurons. Activity signals are based on the changing fluorescence of calcium indicators in the cells over time. We propose a cell segmentation method that relies on a user-specified center point, from which the signal extraction method proceeds. A stabilization approach is used to reduce tissue motion in the video. The extracted signal is then processed to flatten the baseline and detect action potentials. We show results from applying the method to a cortical video of a live mouse.
Vision-Language-Action (VLA) models enable test-time behavioral steering via neuron-level interventions, but existing methods use fixed strengths and operate in open loop. This static modulation fails under evolving task dynamics, leading to overcorrection, oscillations, and reduced task success—especially for temporal attributes like speed. We propose CTRL-STEER, a control-theoretic framework that casts activation steering as closed-loop feedback with adaptive, time-varying interventions. Instead of assuming neurons encode temporal concepts, we steer along motion-aligned residual directions and regulate intervention magnitude via feedback. We instantiate this with both PID and reinforcement learning controllers that jointly optimize concept adherence and task success. Experiments on fine-tuned OpenVLA policies across four LIBERO suites show improved stability and a better steering–success trade-off over fixed-coefficient baselines, without retraining the base model.
Cognitive impairment is one of the many symptoms reported by individuals suffering from long-COVID and other post-viral infection disorders such as myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). A common factor among these conditions is a sustained immune response and increased levels of inflammatory cytokines. Tumor necrosis factor alpha (TNF-α) and interleukin-6 (IL-6) are two such cytokines that are elevated in patients diagnosed with long-COVID and ME/CFS. In this study, we characterized the changes in neural functionality, secreted cytokine profiles, and gene expression in co-cultures of human iPSC-derived neurons and primary astrocytes in response to prolonged exposure to TNF-α and IL-6. We found that exposure to TNF-α produced both a concentration-independent and concentration-dependent response in neural activity. Burst duration was significantly reduced within a few days of exposure regardless of concentration (1 pg/mL – 100 ng/mL) but returned to baseline after 7 days. Treatment with low concentrations of TNF-α (e.g., 1 and 25 pg/mL) did not lead to changes in the secreted cytokine profile or gene expression but still resulted in significant changes to electrophysiological features such as interspike interval and burst duration. Conversely, treatment with high concentrations of TNF-α (e.g., 10 and 100 ng/mL) led to reduced spiking activity, which may be correlated to changes in neural health, gene expression, and increases in inflammatory cytokine secretion (e.g., IL-1β, IL-4, and CXCL-10) that were observed at higher TNF-α concentrations. Prolonged exposure to IL-6 led to changes in bursting features, with significant reduction in the number of spikes in bursts across a wide range of treatment concentrations (i.e., 1 pg/mL–10 ng/mL). In combination, the addition of IL-6 appears to counteract the changes to neural function induced by low concentrations of TNF-α, while at high concentrations of TNF-α the addition of IL-6 had little to no effect. Conversely, the changes to electrophysiological features induced by IL-6 were lost when the cultures were co-stimulated with TNF-α regardless of the concentration, suggesting that TNF-α may play a more pronounced role in altering neural function. These results indicate that increased concentrations of key inflammatory cytokines associated with long-COVID can directly impact neural function and may be a component of the cognitive impairment associated with long-COVID and other post-viral infection disorders.
Fentanyl is one of the most common opioid analgesics administered to patients undergoing surgery or for chronic pain management. While the side effects of chronic fentanyl abuse are recognized (e.g., addiction, tolerance, impairment of cognitive functions, and inhibit nociception, arousal, and respiration), it remains poorly understood what and how changes in brain activity from chronic fentanyl use influences the respective behavioral outcome. Here, we examined the functional and molecular changes to cortical neural network activity following sub-chronic exposure to two fentanyl concentrations, a low (0.01 μM) and high (10 μM) dose. Primary rat co-cultures, containing cortical neurons, astrocytes, and oligodendrocyte precursor cells, were seeded in wells on either a 6-well multi-electrode array (MEA, for electrophysiology) or a 96-well tissue culture plate (for serial endpoint bulk RNA sequencing analysis). Once networks matured (at 28 days in vitro ), co-cultures were treated with 0.01 or 10 μM of fentanyl for 4 days and monitored daily. Only high dose exposure to fentanyl resulted in a decline in features of spiking and bursting activity as early as 30 min post-exposure and sustained for 4 days in cultures. Transcriptomic analysis of the complex cultures after 4 days of fentanyl exposure revealed that both the low and high dose induced gene expression changes involved in synaptic transmission, inflammation, and organization of the extracellular matrix. Collectively, the findings of this in vitro study suggest that while neuroadaptive changes to neural network activity at a systems level was detected only at the high dose of fentanyl, transcriptomic changes were also detected at the low dose conditions, suggesting that fentanyl rapidly elicits changes in plasticity.
Rhythm variations of neurons of reticular formation during lateral overloads