Q.1 In the Hodgkin‑Huxley model, which ion channel primarily contributes to the rapid upstroke of the action potential?
Potassium (K⁺) channel
Sodium (Na⁺) channel
Calcium (Ca²⁺) channel
Leak channel
Explanation - The fast activation of voltage‑gated Na⁺ channels causes a large inward Na⁺ current, generating the rapid depolarization phase of the action potential.
Correct answer is: Sodium (Na⁺) channel
Q.2 Which of the following describes the FitzHugh‑Nagumo model?
A four‑dimensional model with detailed ion channel kinetics
A simplified two‑dimensional reduction of the Hodgkin‑Huxley model
A stochastic model of synaptic transmission
A model for long‑term potentiation
Explanation - The FitzHugh‑Nagumo model captures excitability with a fast variable (membrane voltage) and a slow recovery variable, simplifying the Hodgkin‑Huxley equations.
Correct answer is: A simplified two‑dimensional reduction of the Hodgkin‑Huxley model
Q.3 In a leaky integrate‑and‑fire neuron, the membrane time constant τ_m is given by:
τ_m = R_m C_m
τ_m = R_m / C_m
τ_m = C_m / R_m
τ_m = R_m + C_m
Explanation - The membrane time constant is the product of the membrane resistance R_m and capacitance C_m, representing how quickly the membrane potential decays toward rest.
Correct answer is: τ_m = R_m C_m
Q.4 Which control strategy is most commonly used to stabilize a neural prosthetic device against perturbations?
Proportional‑Integral‑Derivative (PID) control
Sliding‑mode control
Adaptive control
Model Predictive Control (MPC)
Explanation - PID control offers a simple yet effective method to correct errors in real‑time neural prosthetic applications, balancing speed and robustness.
Correct answer is: Proportional‑Integral‑Derivative (PID) control
Q.5 In state‑space representation of a neural mass model, the matrix A primarily describes:
Input‑output mapping
System dynamics (state evolution)
Measurement noise
Control effort
Explanation - The A matrix defines how the internal states of the system evolve over time without external inputs.
Correct answer is: System dynamics (state evolution)
Q.6 Which of the following is a typical assumption when linearizing a nonlinear neural model around an equilibrium point?
All variables are constant
Higher‑order terms are negligible
Noise dominates the dynamics
The system is time‑varying
Explanation - Linearization retains only first‑order (linear) terms, assuming higher‑order nonlinearities have minimal effect near the equilibrium.
Correct answer is: Higher‑order terms are negligible
Q.7 The Kalman filter is optimal for estimating neural states under which condition?
Non‑Gaussian noise
Nonlinear dynamics
Gaussian noise and linear dynamics
Deterministic systems
Explanation - The classic Kalman filter provides the minimum‑variance estimate when the process and measurement noises are Gaussian and the system is linear.
Correct answer is: Gaussian noise and linear dynamics
Q.8 A Hopfield network used for associative memory exhibits which type of dynamics?
Limit cycle oscillations
Chaotic trajectories
Convergent to energy minima
Divergent growth
Explanation - Hopfield networks are defined by an energy function that decreases monotonically, causing the system to settle into stable attractor states.
Correct answer is: Convergent to energy minima
Q.9 In optimal control of neural stimulation, the cost function typically penalizes:
Only the tracking error
Only the control effort
Both tracking error and control effort
Only the state deviation
Explanation - A standard quadratic cost J = ∫(xᵀQx + uᵀRu) dt balances performance (tracking) against energy usage (control effort).
Correct answer is: Both tracking error and control effort
Q.10 Bifurcation analysis in neural models helps to identify:
The exact spike timing
Parameter values where qualitative changes in dynamics occur
The amplitude of action potentials
The refractory period length
Explanation - Bifurcations mark transitions such as from quiescence to repetitive firing as system parameters are varied.
Correct answer is: Parameter values where qualitative changes in dynamics occur
Q.11 Which of the following is NOT a typical component of a closed‑loop neural interface?
Signal acquisition module
Stimulus generator
Open‑loop controller
Feedback processor
Explanation - Closed‑loop systems rely on feedback; an open‑loop controller does not incorporate feedback information.
Correct answer is: Open‑loop controller
Q.12 The Wilson‑Cowan model describes the interaction between:
Excitatory and inhibitory neuronal populations
Ion channels and pumps
Dendritic spines and axons
Glial cells and neurons
Explanation - Wilson‑Cowan equations model the average firing rates of coupled excitatory and inhibitory populations.
Correct answer is: Excitatory and inhibitory neuronal populations
Q.13 In a linear quadratic regulator (LQR) design for neural control, the matrix R must be:
Positive semi‑definite
Negative definite
Positive definite
Zero
Explanation - R weights the control effort; to ensure a unique optimal solution, R must be positive definite.
Correct answer is: Positive definite
Q.14 Which phenomenon is modeled by the Izhikevich neuron model?
Precise ion channel kinetics
Various spiking and bursting patterns with low computational cost
Synaptic plasticity mechanisms
Glial calcium waves
Explanation - Izhikevich’s model combines simplicity with the ability to reproduce many observed neuronal firing patterns.
Correct answer is: Various spiking and bursting patterns with low computational cost
Q.15 When applying pole placement to a linearized neural system, the desired poles are chosen to:
Maximize the system’s natural frequency
Ensure stability and meet performance specifications
Minimize the number of states
Increase the system order
Explanation - Pole placement assigns closed‑loop eigenvalues to achieve desired speed of response, damping, and stability.
Correct answer is: Ensure stability and meet performance specifications
Q.16 The term ‘neuronal synchrony’ in network models usually refers to:
All neurons firing at the exact same instant
Neurons having identical membrane potentials at all times
A statistical tendency of neurons to fire within a short time window
Neurons sharing the same ion channel composition
Explanation - Synchrony often describes phase‑locked or correlated firing rather than perfect simultaneous spikes.
Correct answer is: A statistical tendency of neurons to fire within a short time window
Q.17 In a stochastic differential equation describing neural noise, the term dW(t) represents:
Deterministic drift
White Gaussian noise (Wiener process)
Synaptic conductance
Membrane capacitance
Explanation - dW(t) denotes an increment of a Wiener process, modeling continuous‑time Gaussian noise.
Correct answer is: White Gaussian noise (Wiener process)
Q.18 Which of the following is a primary advantage of using a model‑based controller for deep brain stimulation (DBS)?
Eliminates the need for any sensing hardware
Allows adaptation to patient‑specific dynamics
Guarantees zero power consumption
Requires no parameter tuning
Explanation - Model‑based controllers can be tuned to individual neural responses, improving efficacy and reducing side effects.
Correct answer is: Allows adaptation to patient‑specific dynamics
Q.19 The Lyapunov function is used in neural control to:
Calculate the firing rate
Prove stability of an equilibrium point
Determine the refractory period
Estimate synaptic weights
Explanation - If a Lyapunov function can be found that always decreases, the system is stable around the equilibrium.
Correct answer is: Prove stability of an equilibrium point
Q.20 In the context of neural decoding, the term ‘inverse model’ refers to:
Predicting neural activity from motor commands
Predicting motor commands from recorded neural activity
Simulating the biophysics of a neuron
Generating synthetic spike trains
Explanation - An inverse model maps observed neural signals back to the intended movement or control command.
Correct answer is: Predicting motor commands from recorded neural activity
Q.21 Which of the following describes the purpose of a ‘dead‑zone’ nonlinearity in a neural controller?
Amplify small errors
Prevent actuator wear by ignoring small error signals
Introduce chaos into the system
Linearize the plant dynamics
Explanation - A dead‑zone suppresses control action when the error is within a small bound, reducing unnecessary actuation.
Correct answer is: Prevent actuator wear by ignoring small error signals
Q.22 When modeling a population of neurons with the mean‑field approach, the key variable often used is:
Individual spike times
Average firing rate
Exact membrane potential of each neuron
Synaptic vesicle count
Explanation - Mean‑field models replace the detailed description of each neuron with a collective variable such as the mean firing rate.
Correct answer is: Average firing rate
Q.23 The Jacobian matrix evaluated at an equilibrium point of a neural system provides information about:
The system’s input‑output gain
Local linear stability and eigenvalues
The exact spike shape
Synaptic plasticity rules
Explanation - Eigenvalues of the Jacobian indicate whether perturbations grow or decay near the equilibrium.
Correct answer is: Local linear stability and eigenvalues
Q.24 Which of the following is a characteristic of a Type I neuron in the Hodgkin classification?
Abrupt onset of firing at a non‑zero frequency
Continuous frequency‑current (f‑I) curve starting from zero
No afterhyperpolarization
Only inhibitory output
Explanation - Type I neurons can fire at arbitrarily low frequencies as the input current exceeds a threshold.
Correct answer is: Continuous frequency‑current (f‑I) curve starting from zero
Q.25 In the context of neuromorphic hardware, ‘event‑driven’ computation means:
Processing at a fixed clock rate
Updating only when spikes occur
Using analog voltage levels
Running continuous differential equation solvers
Explanation - Event‑driven systems process information only on the occurrence of discrete events (spikes), saving energy.
Correct answer is: Updating only when spikes occur
Q.26 Which term best describes the phenomenon where a small change in a model parameter leads to a large change in system behavior?
Robustness
Bifurcation
Linearization
Hysteresis
Explanation - Bifurcations mark qualitative changes in dynamics caused by parameter variations.
Correct answer is: Bifurcation
Q.27 In adaptive control of a neural prosthesis, the parameter update law is often based on:
Gradient descent on a Lyapunov function
Fourier transform of the signal
Random search
Fixed gain scheduling
Explanation - A Lyapunov‑based adaptive law ensures stability while adjusting parameters to minimize error.
Correct answer is: Gradient descent on a Lyapunov function
Q.28 The term ‘synaptic weight’ in neural network models corresponds to:
Physical size of a synapse
Conductance strength of the connection
Number of neurons in the network
Membrane capacitance
Explanation - Synaptic weight determines how strongly a presynaptic neuron influences the postsynaptic neuron, often modeled as a conductance term.
Correct answer is: Conductance strength of the connection
Q.29 Which of the following is a common method for estimating hidden neural states from noisy measurements?
Fourier analysis
Extended Kalman Filter (EKF)
Fast Fourier Transform (FFT)
Principal Component Analysis (PCA)
Explanation - EKF linearizes the nonlinear system around the current estimate, allowing recursive state estimation with Gaussian noise.
Correct answer is: Extended Kalman Filter (EKF)
Q.30 In a recurrent neural network used for motor control, stability can be ensured by enforcing:
Positive eigenvalues of the weight matrix
Symmetric weight matrix with negative definite Jacobian
All weights equal to zero
Random weight initialization
Explanation - A symmetric weight matrix with negative eigenvalues ensures the Lyapunov function decreases, yielding stable dynamics.
Correct answer is: Symmetric weight matrix with negative definite Jacobian
Q.31 The term ‘phase response curve’ (PRC) quantifies:
Amplitude of action potentials
Change in oscillation period caused by a perturbation at a given phase
Resting membrane potential
Synaptic delay
Explanation - PRC describes how a brief input shifts the timing of the next spike depending on when it arrives within the cycle.
Correct answer is: Change in oscillation period caused by a perturbation at a given phase
Q.32 When designing a controller for a neural implant, which of the following safety constraints is most critical?
Minimizing computational latency
Limiting charge per phase to avoid tissue damage
Maximizing stimulation frequency
Using the highest possible voltage
Explanation - Electrochemical safety limits (charge density) are essential to prevent electrochemical reactions that can harm tissue.
Correct answer is: Limiting charge per phase to avoid tissue damage
Q.33 In the context of neural modeling, ‘slow‑fast decomposition’ refers to:
Separating excitatory and inhibitory populations
Dividing dynamics into variables that evolve on distinct time scales
Classifying neurons by size
Distinguishing synaptic from intrinsic currents
Explanation - Slow‑fast analysis exploits the fact that some state variables change much more slowly than others, simplifying analysis.
Correct answer is: Dividing dynamics into variables that evolve on distinct time scales
Q.34 Which of the following best describes a ‘limit cycle’ in a neural oscillator model?
A fixed point that never moves
A closed trajectory that attracts nearby states
A trajectory that diverges to infinity
A chaotic attractor
Explanation - Limit cycles represent stable periodic firing patterns that neighboring trajectories converge to.
Correct answer is: A closed trajectory that attracts nearby states
Q.35 In reinforcement learning applied to neural control, the ‘reward signal’ typically encodes:
The exact spike times
Performance error or task success
Synaptic weight magnitudes
Membrane capacitance
Explanation - The reward guides the learning algorithm to improve control policies based on how well the task is performed.
Correct answer is: Performance error or task success
Q.36 Which mathematical tool is most suitable for analyzing the frequency response of a linearized neural system?
Bode plot
Phase portrait
Spike raster plot
Histogram
Explanation - Bode plots display magnitude and phase versus frequency, useful for assessing gain and stability margins.
Correct answer is: Bode plot
Q.37 The ‘Hodgkin‑Huxley gating variable m’ corresponds to:
Inactivation of Na⁺ channels
Activation of Na⁺ channels
Activation of K⁺ channels
Leak conductance
Explanation - Variable m describes the probability of Na⁺ channel activation gates being open.
Correct answer is: Activation of Na⁺ channels
Q.38 When implementing a neural controller on digital hardware, the main cause of discretization error is:
Analog noise
Finite sampling period (Δt)
Temperature drift
Synaptic plasticity
Explanation - Choosing a non‑infinitesimal Δt introduces errors in approximating continuous dynamics.
Correct answer is: Finite sampling period (Δt)
Q.39 In the context of brain‑computer interfaces (BCI), the term ‘decoding latency’ refers to:
Time between stimulus and neuronal response
Delay between neural signal acquisition and generation of the command output
Propagation speed of action potentials
Time required for synaptic vesicle release
Explanation - Decoding latency impacts the responsiveness of a BCI system and is critical for real‑time control.
Correct answer is: Delay between neural signal acquisition and generation of the command output
Q.40 Which of the following is a hallmark of chaotic dynamics in a neural model?
Periodic firing with constant period
Sensitive dependence on initial conditions
Single fixed point
Linear growth of membrane potential
Explanation - Chaos is characterized by exponential divergence of nearby trajectories, making long‑term prediction impossible.
Correct answer is: Sensitive dependence on initial conditions
Q.41 The ‘gain‑phase margin’ of a closed‑loop neural system is assessed to guarantee:
Maximum firing rate
Robust stability against model uncertainties
Minimum synaptic weight
Optimal spike shape
Explanation - Gain and phase margins quantify how much gain or phase shift can change before instability occurs.
Correct answer is: Robust stability against model uncertainties
Q.42 In a neural mass model, the variable representing the average membrane potential of a population is often denoted by:
V̅
I_syn
g_leak
τ_m
Explanation - V̅ (V bar) commonly denotes the mean membrane potential across the modeled neural population.
Correct answer is: V̅
Q.43 A common metric for evaluating the performance of a neural controller in a simulated task is:
Mean squared error (MSE) between desired and actual trajectories
Number of spikes per second
Membrane resistance value
Synaptic delay
Explanation - MSE quantifies tracking accuracy, a primary performance indicator for control systems.
Correct answer is: Mean squared error (MSE) between desired and actual trajectories
Q.44 Which of the following best describes a ‘feedforward’ neural network?
Connections that form cycles
Connections that only go from input to output without loops
Neurons that fire only once
Networks that adapt synaptic weights online
Explanation - Feedforward networks have a directed acyclic graph structure, unlike recurrent networks.
Correct answer is: Connections that only go from input to output without loops
Q.45 In the context of system identification for neural models, the term ‘persistent excitation’ ensures:
The system remains at rest
Sufficient richness of input signals to uniquely estimate parameters
Zero measurement noise
Immediate convergence of the estimator
Explanation - Persistent excitation guarantees that the data contains enough information for accurate parameter identification.
Correct answer is: Sufficient richness of input signals to uniquely estimate parameters
Q.46 The ‘refractory period’ in neuronal dynamics is primarily caused by:
Inactivation of Na⁺ channels and activation of K⁺ channels
Leak currents
Synaptic vesicle depletion
Glial buffering
Explanation - During the absolute and relative refractory periods, Na⁺ channels are inactivated and K⁺ channels remain open, preventing immediate re‑excitation.
Correct answer is: Inactivation of Na⁺ channels and activation of K⁺ channels
Q.47 A neural controller that updates its parameters based on the error gradient is an example of:
Model predictive control
Gradient‑based adaptive control
Bang‑bang control
Open‑loop control
Explanation - The controller uses the gradient of a cost function with respect to parameters to reduce error over time.
Correct answer is: Gradient‑based adaptive control
Q.48 When using a Wiener filter for neural signal processing, the filter is designed to:
Maximize the signal‑to‑noise ratio in the mean‑square sense
Detect spikes by thresholding
Perform Fourier transformation
Introduce non‑linearity into the signal
Explanation - The Wiener filter provides the optimal linear estimator that minimizes mean‑square error between the desired and estimated signals.
Correct answer is: Maximize the signal‑to‑noise ratio in the mean‑square sense
Q.49 In a conductance‑based model, the total membrane current I_m is given by:
I_m = C_m dV/dt + Σ g_i (V - E_i)
I_m = R_m dV/dt
I_m = V / R_m
I_m = g_leak V
Explanation - The membrane current consists of capacitive current and the sum of ionic currents (conductance × driving force).
Correct answer is: I_m = C_m dV/dt + Σ g_i (V - E_i)
Q.50 The term ‘bursting’ in neuronal activity refers to:
A single isolated spike
A rapid series of spikes followed by a quiescent period
Continuous high‑frequency firing without pause
A decrease in membrane potential
Explanation - Bursting consists of clusters of spikes separated by silent intervals, often important for coding.
Correct answer is: A rapid series of spikes followed by a quiescent period
Q.51 In the context of neural prosthetics, the term ‘closed‑loop latency budget’ refers to:
Total allowable time from sensing to stimulation to maintain effective control
Maximum number of neurons that can be recorded
Energy consumption limit
Number of electrodes that can be implanted
Explanation - Latency budgets ensure that the loop is fast enough to interact with neural dynamics without degrading performance.
Correct answer is: Total allowable time from sensing to stimulation to maintain effective control
Q.52 A ‘phase‑locked loop’ (PLL) used in neural signal processing primarily serves to:
Synchronize a reference oscillator to the phase of an incoming neural oscillation
Amplify low‑amplitude spikes
Filter out high‑frequency noise
Generate random spike trains
Explanation - PLL tracks the phase of a periodic neural signal, enabling phase‑dependent stimulation or analysis.
Correct answer is: Synchronize a reference oscillator to the phase of an incoming neural oscillation
Q.53 Which of the following is a typical assumption in the use of mean‑field models for cortical columns?
All neurons fire synchronously
Neurons within the column are statistically identical
Synaptic delays are negligible
Membrane capacitance varies widely
Explanation - Mean‑field approaches average over a homogeneous population, assuming similar dynamics for each neuron.
Correct answer is: Neurons within the column are statistically identical
Q.54 In control theory, the term ‘observability’ of a neural system indicates:
All states can be reconstructed from output measurements
The system can be stabilized with feedback
The system has no noise
All eigenvalues are negative
Explanation - Observability ensures that the internal state vector can be inferred from the measured outputs.
Correct answer is: All states can be reconstructed from output measurements
Q.55 A ‘spike‑triggered average’ (STA) is used to:
Calculate the average firing rate over time
Estimate the average stimulus preceding a spike
Determine the refractory period duration
Measure the amplitude of an action potential
Explanation - STA aligns stimulus segments around each spike and averages them to reveal the feature that most reliably elicits spikes.
Correct answer is: Estimate the average stimulus preceding a spike
Q.56 When a neural model exhibits a Hopf bifurcation, the system:
Gains a stable fixed point
Transitions from a fixed point to sustained oscillations
Becomes chaotic
Stops firing altogether
Explanation - A Hopf bifurcation creates a pair of complex conjugate eigenvalues crossing the imaginary axis, giving rise to limit cycle oscillations.
Correct answer is: Transitions from a fixed point to sustained oscillations
Q.57 In a linear time‑invariant (LTI) neural system, the impulse response h(t) is:
The Fourier transform of the output
The system’s output when the input is a Dirac delta function
The derivative of the membrane potential
A constant value for all t
Explanation - The impulse response fully characterizes an LTI system’s behavior.
Correct answer is: The system’s output when the input is a Dirac delta function
Q.58 Which of the following neural coding schemes relies on the precise timing of spikes rather than their rate?
Rate coding
Temporal (or spike‑time) coding
Population coding
Binary coding
Explanation - Temporal coding encodes information in the exact timing of individual spikes.
Correct answer is: Temporal (or spike‑time) coding
Q.59 The term ‘synaptic plasticity’ in neural models most often refers to:
Changes in membrane capacitance
Adjustment of synaptic weights based on activity
Growth of axons
Fluctuations in temperature
Explanation - Synaptic plasticity such as STDP modifies connection strengths depending on spike timing.
Correct answer is: Adjustment of synaptic weights based on activity
Q.60 In a stochastic integrate‑and‑fire model, the term ‘diffusion coefficient’ D influences:
Deterministic drift only
The variance of the membrane potential fluctuations
The refractory period length
The size of the action potential
Explanation - D scales the intensity of the Wiener process term, affecting the spread of stochastic trajectories.
Correct answer is: The variance of the membrane potential fluctuations
Q.61 When applying model reduction techniques to a high‑dimensional neural network, one common method is:
Increasing the number of neurons
Proper Orthogonal Decomposition (POD)
Adding more synaptic connections
Using higher sampling rates
Explanation - POD (or PCA) extracts dominant modes, reducing dimensionality while preserving essential dynamics.
Correct answer is: Proper Orthogonal Decomposition (POD)
Q.62 The term ‘gain scheduling’ in adaptive neural control refers to:
Changing controller gains based on operating conditions
Keeping gains constant for all conditions
Scheduling spikes in time
Randomly varying gains
Explanation - Gain scheduling selects appropriate controller parameters for different regions of the operating space.
Correct answer is: Changing controller gains based on operating conditions
Q.63 Which of the following is a key advantage of using the Izhikevich model over the Hodgkin‑Huxley model for large‑scale simulations?
Higher biological realism
Faster computational speed while preserving diverse firing patterns
Exact representation of ion channel kinetics
Inclusion of detailed dendritic morphology
Explanation - Izhikevich’s model balances simplicity and ability to replicate many neuronal firing types, making it suitable for large networks.
Correct answer is: Faster computational speed while preserving diverse firing patterns
Q.64 In a neural prosthetic system, the ‘artifact rejection’ stage is necessary to:
Increase the amplitude of recorded spikes
Remove stimulation‑induced electrical artifacts from the recorded signals
Delay the control command
Adjust synaptic weights
Explanation - Artifacts can obscure neural activity; removal is essential for accurate sensing and closed‑loop operation.
Correct answer is: Remove stimulation‑induced electrical artifacts from the recorded signals
Q.65 When linearizing the FitzHugh‑Nagumo model around its rest state, the resulting linear system will have:
Two complex conjugate eigenvalues with positive real part
One zero eigenvalue and one negative eigenvalue
Two real eigenvalues, both negative
A single positive eigenvalue
Explanation - At the resting (stable) equilibrium, the linearized system exhibits negative real eigenvalues indicating stability.
Correct answer is: Two real eigenvalues, both negative
Q.66 Which of the following control architectures is most suitable for handling multi‑objective optimization in neural prostheses (e.g., minimizing error while reducing power consumption)?
Single‑objective PID control
Multi‑objective Model Predictive Control (MPC)
Bang‑bang control
Open‑loop stimulation
Explanation - MPC can incorporate several cost terms and constraints, allowing simultaneous optimization of performance and energy usage.
Correct answer is: Multi‑objective Model Predictive Control (MPC)
Q.67 In the context of neural oscillations, the term ‘phase‑amplitude coupling’ (PAC) describes:
Interaction where the phase of a low‑frequency rhythm modulates the amplitude of a high‑frequency rhythm
Synchronization of all neurons at the same frequency
Linear superposition of two sine waves
A type of synaptic plasticity
Explanation - PAC is a cross‑frequency coupling phenomenon observed in many brain areas and relevant for decoding neural states.
Correct answer is: Interaction where the phase of a low‑frequency rhythm modulates the amplitude of a high‑frequency rhythm
Q.68 Which of the following is a typical method for stabilizing an unstable neural limit cycle during simulation?
Increasing the time step size
Adding a small damping term to the equations
Removing all nonlinearities
Setting all conductances to zero
Explanation - A damping term can shift eigenvalues to have negative real parts, stabilizing the cycle without drastically altering dynamics.
Correct answer is: Adding a small damping term to the equations
Q.69 In a spiking neural network, the term ‘refractory reset’ typically refers to:
Setting the membrane potential to a fixed value after a spike
Increasing the synaptic weight
Changing the neuron’s type
Turning off the neuron permanently
Explanation - After a spike, many models reset V to a reset potential (often V_reset) and enforce a refractory period.
Correct answer is: Setting the membrane potential to a fixed value after a spike
Q.70 The ‘Cramer‑Rao bound’ in the context of neural parameter estimation provides:
The maximum possible error of an estimator
A lower bound on the variance of any unbiased estimator
The exact value of the parameter
A method for generating spikes
Explanation - The Cramer‑Rao bound quantifies the best achievable precision for unbiased estimators given the data.
Correct answer is: A lower bound on the variance of any unbiased estimator
Q.71 When performing a frequency analysis of a neural signal, which transform is most appropriate for non‑stationary data?
Fourier Transform
Wavelet Transform
Laplace Transform
Z‑Transform
Explanation - Wavelet analysis provides time‑frequency localization, suitable for signals whose spectral content changes over time.
Correct answer is: Wavelet Transform
Q.72 In the context of neural prosthetic control, the term ‘haptic feedback’ refers to:
Visual display of neural activity
Providing tactile sensations to the user based on system state
Increasing stimulation amplitude
Recording EEG signals
Explanation - Haptic feedback closes the loop by delivering physical sensations, enhancing the sense of control.
Correct answer is: Providing tactile sensations to the user based on system state
Q.73 Which of the following best describes the role of the ‘learning rate’ in a gradient‑descent based neural controller?
Determines the size of each parameter update step
Specifies the maximum firing rate
Controls the membrane time constant
Sets the noise level in the system
Explanation - A higher learning rate leads to faster adaptation but can cause instability; a lower rate yields slower convergence.
Correct answer is: Determines the size of each parameter update step
Q.74 In a closed‑loop deep brain stimulation system, the ‘feedback signal’ is typically derived from:
Battery voltage
Local field potentials (LFPs) recorded near the stimulation site
External temperature sensors
User’s voice
Explanation - LFPs provide information about the ongoing neural state and are used to adjust stimulation parameters in real time.
Correct answer is: Local field potentials (LFPs) recorded near the stimulation site
Q.75 A ‘zero‑order hold’ (ZOH) in digital control of neural systems is used to:
Interpolate input signals between sampling instants
Maintain the control input constant over each sampling interval
Introduce random noise into the system
Accelerate the computation
Explanation - ZOH holds the most recent digital command value until the next update, approximating a continuous‑time input.
Correct answer is: Maintain the control input constant over each sampling interval
Q.76 The ‘synchronization index’ between two neural signals is commonly computed using:
Cross‑correlation at zero lag
Phase‑locking value (PLV)
Mean firing rate difference
Spike count per second
Explanation - PLV measures consistency of phase differences across trials, quantifying synchrony.
Correct answer is: Phase‑locking value (PLV)
Q.77 When designing a neural controller, the term ‘robustness’ generally refers to:
The controller’s ability to maintain performance despite model uncertainties and disturbances
The speed at which spikes are generated
The amount of power consumed
The size of the electrodes
Explanation - Robustness is a key design goal to ensure reliable operation under varying biological conditions.
Correct answer is: The controller’s ability to maintain performance despite model uncertainties and disturbances
Q.78 In the context of neural modeling, a ‘population vector’ is used to:
Represent the summed direction of activity across a group of neurons
Measure the voltage across a single neuron
Calculate the refractory period
Determine synaptic delay
Explanation - Population vectors combine the preferred directions of many neurons, weighted by their firing rates, to decode movement direction.
Correct answer is: Represent the summed direction of activity across a group of neurons
Q.79 Which of the following is a typical approach for reducing over‑fitting when training a neural model for control tasks?
Increasing the number of hidden layers indefinitely
Applying L2 regularization (weight decay)
Using a single training sample
Eliminating all noise from the data
Explanation - L2 regularization penalizes large weights, encouraging smoother models and reducing over‑fitting.
Correct answer is: Applying L2 regularization (weight decay)
Q.80 In a conductance‑based neuron model, the reversal potential E_i for a given ion species determines:
The maximum firing rate
The voltage at which that ionic current changes direction
The membrane capacitance
The spike amplitude
Explanation - When V = E_i, the driving force (V‑E_i) becomes zero, and no net current flows for that ion.
Correct answer is: The voltage at which that ionic current changes direction
Q.81 The ‘delay differential equation’ (DDE) is often employed in neural modeling to capture:
Instantaneous synaptic transmission
Finite propagation delays in synaptic or axonal pathways
Membrane capacitance
Noise in ion channels
Explanation - DDEs incorporate explicit time delays, reflecting realistic transmission times in neural circuits.
Correct answer is: Finite propagation delays in synaptic or axonal pathways
Q.82 Which of the following is NOT a typical component of the cost function in an optimal control problem for neural stimulation?
Tracking error term
Control energy term
Synaptic plasticity term
Terminal state penalty
Explanation - Standard optimal control formulations focus on performance and energy; plasticity is usually modeled separately.
Correct answer is: Synaptic plasticity term
Q.83 When implementing a Kalman filter for neural state estimation, the ‘process noise covariance’ Q represents:
Uncertainty in the measurement devices
Uncertainty in the system dynamics (model error)
The exact value of the state vector
The sampling frequency
Explanation - Q quantifies how much the true state may deviate from the predicted state due to unmodeled dynamics or stochastic influences.
Correct answer is: Uncertainty in the system dynamics (model error)
Q.84 In the context of neuromodulation, the term ‘burst stimulation’ refers to:
A continuous single‑pulse train
A series of high‑frequency pulses grouped into bursts, separated by silent intervals
Randomly timed pulses
Low‑frequency sinusoidal stimulation
Explanation - Burst stimulation can enhance efficacy while reducing side effects compared to continuous high‑frequency stimulation.
Correct answer is: A series of high‑frequency pulses grouped into bursts, separated by silent intervals
Q.85 Which of the following statements about the 'Fano factor' in neural spike trains is correct?
It is the ratio of mean spike count to variance
It is the ratio of variance to mean spike count
It measures the average inter‑spike interval
It equals zero for Poisson processes
Explanation - Fano factor = variance / mean; for a Poisson process it equals 1.
Correct answer is: It is the ratio of variance to mean spike count
Q.86 In a linearized neural control system, the eigenvalues of the A matrix determine:
The shape of action potentials
Stability and speed of the system’s response
The number of synapses
The amplitude of the control input
Explanation - Eigenvalues with negative real parts indicate a stable system; their magnitude relates to response speed.
Correct answer is: Stability and speed of the system’s response
Q.87 Which of the following techniques is specifically designed to handle non‑Gaussian measurement noise in neural state estimation?
Extended Kalman Filter (EKF)
Particle Filter
Linear Quadratic Regulator (LQR)
Bode Plot
Explanation - Particle filters use a set of samples to represent arbitrary probability distributions, coping with non‑Gaussian noise.
Correct answer is: Particle Filter
Q.88 The term ‘synaptic delay’ in a neural network model typically ranges from:
0.1–1 ms
1–5 ms
10–20 ms
50–100 ms
Explanation - Synaptic transmission, including neurotransmitter release and receptor activation, generally introduces a delay of about 1–5 ms.
Correct answer is: 1–5 ms
Q.89 When a neural controller includes a ‘dead‑time’ (pure time delay) in its feedback loop, the system’s stability can be improved by:
Increasing the feedback gain arbitrarily
Adding a phase‑lead compensator
Removing the delay completely
Reducing the sampling rate
Explanation - A phase‑lead compensator adds positive phase margin, helping to counteract the destabilizing effect of time delay.
Correct answer is: Adding a phase‑lead compensator
Q.90 In a neural field model, the kernel function describes:
The shape of an action potential
Spatial connectivity strength between points in the tissue
The voltage‑dependent gating kinetics
The refractory period duration
Explanation - The kernel defines how activity at one location influences neighboring locations across the cortical sheet.
Correct answer is: Spatial connectivity strength between points in the tissue
Q.91 Which of the following best describes a ‘model‑based decoder’ for a brain‑computer interface?
A system that directly maps spike counts to cursor positions without a neural model
A decoder that uses an explicit mathematical model of neural population dynamics to infer intended movement
A hardware circuit that amplifies neural signals
A random number generator
Explanation - Model‑based decoders leverage knowledge of neural dynamics to improve prediction accuracy over purely statistical approaches.
Correct answer is: A decoder that uses an explicit mathematical model of neural population dynamics to infer intended movement
Q.92 In the context of neural control, the term ‘actuator saturation’ refers to:
The inability of the actuator to exceed its maximum output limits
The maximum firing rate of a neuron
The highest possible stimulation frequency
The largest possible membrane potential
Explanation - Saturation occurs when the control signal demands exceed the physical capabilities of the actuator (e.g., stimulator current limit).
Correct answer is: The inability of the actuator to exceed its maximum output limits
Q.93 When performing a stability analysis of a nonlinear neural system using Lyapunov’s direct method, one must find a function V(x) that is:
Positive definite and its time derivative negative definite
Negative definite and its time derivative positive definite
Constant for all x
Zero for all x
Explanation - V(x) > 0 for x ≠ 0 and \u{dV}/dt < 0 guarantee asymptotic stability of the equilibrium.
Correct answer is: Positive definite and its time derivative negative definite
Q.94 In a spiking neural network, the term ‘population coding’ implies that:
Each neuron encodes an entire stimulus individually
Information is represented by the collective activity pattern of many neurons
Only the fastest neurons are used for coding
Coding is based solely on spike amplitude
Explanation - Population coding leverages the distributed representation across a group of neurons to improve robustness and resolution.
Correct answer is: Information is represented by the collective activity pattern of many neurons
Q.95 Which of the following is a typical symptom of an unstable neural controller in a prosthetic system?
Steady, smooth movement
Oscillatory or trembling motion
No movement at all
Reduced power consumption
Explanation - Instability often manifests as high‑frequency oscillations or jitter in the output, degrading performance.
Correct answer is: Oscillatory or trembling motion