In the field of neuroscience and cognitive science research, electroencephalography (EEG), near-infrared (fNIRS), oculomotor and physiological multidirectional devices have become the core tools for analyzing brain function and behavioral mechanisms. As the basic parameter of data acquisition, the choice of sampling rate is directly related to the signal accuracy, research validity and data processing efficiency This paper introduces how to choose the correct sampling rate for different devices.

First of all, we must first understand the meaning of the sampling rate: sampling rate (sampling rate) is also known as the sampling speed or sampling frequency, which refers to the number of samples per second from the continuous signal extraction and composition of discrete signals, the unit of Hertz (Hz) said.

EEG: A major feature of digital EEG is its ability to digitally sample the acquired EEG signals in real time, and the sampled data is passed through an analog-to-digital converter (ADC, A/D) to form a computer-recognizable graphic, i.e., the time-dependent waveform that we see in the computer monitor.

Frequency range of EEG signals

Typical frequency bands:

Delta (0.1-4Hz): deep sleep, anesthesia.

Theta (4-8Hz): light sleep, memory encoding.

Alpha (8-13 Hz): resting closed eye state.

Beta (13-30Hz): wakefulness, cognitive activity.

Gamma (30-100Hz): highly focused, neural synchronization.

Higher frequency components (e.g., 100-200 Hz): localized field potentials (LFP), commonly found in intracranial EEG (iEEG).

Nyquist sampling theorem (math.)specific application

Theoretical requirements:

The sampling rate should be at least twice the highest frequency of the signal (i.e. Nyquist frequency) to avoid spectral aliasing.

Practical applications:

To ensure the effectiveness of the anti-alias filter, the sampling rate is usually set to 2.5-3 times the highest target frequency. Example:

When studying Gamma waves (up to 100Hz), the sampling rate should be ≥250Hz (100×2.5).

If the high-frequency components above 100 Hz are to be preserved (e.g., high-frequency oscillations of iEEG), the sampling rate should be ≥ 500 Hz.

NIR:

fNIRS reflects neural activity by measuring changes in cerebral cortical blood oxygen concentration (e.g., oxyhemoglobin HbO₂, deoxyhemoglobin HbR), and its signaling nature is a hemodynamic response (HDR) with the following characteristics:

Signal frequency range:

The frequency of a typical HDR is mainly centered on 0.01-0.1 Hz (corresponding to a period of 10 seconds-100 seconds), and is associated with changes in cerebral blood flow and oxygenation caused by neural activity.

Physiological noise frequency:

Heartbeat-related fluctuations: approximately 1 Hz (60 beats/minute);

Respiration-related fluctuations: approximately 0.2-0.5 Hz (12-30 beats/minute);

Mayer wave (low-frequency fluctuation of blood vessels): about 0.05 Hz.

Core principles of sample rate selection

Nyquist sampling theorem (math.)

The sampling rate should be at least twice the highest frequency of the signal (Nyquist frequency). Example:

If you focus on HDR itself (maximum frequency 0.1 Hz), the theoretical minimum sampling rate is 0.2 Hz;

If heartbeat noise (1 Hz) is to be preserved, the minimum sampling rate is 2 Hz.

In practice, to avoid aliasing and to preserve the filtering margin, the sampling rate is usually set to 3-5 times the highest target frequency.

Physiological Multi-Conductor:Physiological polygraph can synchronously record a variety of physiological signals (e.g., ECG, EEG, EMG, respiration, etc.), and the selection of its sampling rate needs to be based on a combination of the frequency characteristics of different signals, the purpose of the study, and the performance of the equipment.

Frequency characteristics and sampling rate benchmarks of common physiological signals

Core principles of sample rate selection

1. Nyquist sampling theorem and safety factor

The sampling rate should be ≥ 2 × the highest frequency of the signal (Nyquist frequency), in practice, to avoid aliasing, usually use the3-5 times safety factor.

Example:EMG Maximum frequency 500 Hz, theoretical minimum sampling rate 1000 Hz, practical recommendation 1500-2000 Hz.

2. Multi-conductor signal synchronization and equipment compatibility

Uniform sampling rate:Most multidirectional instruments require the same sampling rate (e.g. 1000 Hz) for all channels to ensure time stamp synchronization.

Solution:Low-frequency signals (e.g., respiration) can be downsampled after the fact, but high-frequency signals (e.g., EMG) must meet the sampling requirements.

Equipment Limitations:Sampling rates of 1000 Hz, 2000 Hz, 5000 Hz are common in commercial multidetector instruments (e.g., Biopac), and 10 kHz or more in high-end research equipment.

3. Purpose of the study and need for signaling details

Clinical Diagnosis:Signal characteristics (e.g., changes in the ST segment of the ECG) need to be recorded with high fidelity, and the sampling rate is usually ≥1000 Hz.

Basic scientific research:

Low-frequency signals (e.g., autonomic activity):50-200 Hz is sufficient;

High-frequency events (e.g., EMG action potentials):≥1000 Hz is required;

Fast transient signals (e.g., eye movements, EMG bursts):≥ 500 Hz is required.

Eyetracker: The eyetracker sampling rate indicates the number of times per second that the eyetracker collects data such as eye position, gaze points, sweeps, and so on. For example, a sampling rate of 100 Hz means that 100 data points are recorded per second.

Key Factors Affecting Sample Rate Selection

1. Types of research and scientific questions

Basic cognitive research (e.g., visual attention, neural mechanisms of eye movements):

Demand:Need to capture millisecond-level eye movement details (e.g., sweep latency, microeye hop frequency).

Recommendation:Choose a high sampling rate (500Hz-2000Hz), e.g., 1000Hz to accurately analyze the relationship between gaze duration and eye-beat distance when studying eye-beat patterns in reading.

Research on application scenarios (e.g., human-computer interaction, advertising effectiveness):

Demand:Focus on macro indicators such as gaze point distribution, area of interest (AOI) dwell time, etc.

Recommendation:Low to medium sampling rates (100Hz-300Hz) are sufficient, e.g., to analyze the hotspots of users' attention when browsing the web, 100Hz can meet the basic needs.

Clinical and neurological disease research (e.g., Parkinson's, dyslexia):

Demand:Both temporal accuracy (e.g., anomalous eye movement patterns) and long-time data stability need to be addressed.

Recommendation:Medium to high sampling rates (300Hz-1000Hz), e.g., 500Hz to capture subtle tremor tracks when tracking nystagmus.

2. Temporal characteristics of the experimental tasks

Rapid eye movement tasks (e.g., visual search, dynamic stimulus recognition):

Features:Fast eye movements (sweep speeds up to 500°/sec) require high-frequency sampling to avoid data loss.

Example:When studying user responses to flashing advertisements, 1000Hz accurately records the temporal correlation between the onset of the sweep and the presentation of the stimulus.

Slow gaze tasks (e.g., static image viewing, sustained reading):

Features:Eye movements are predominantly gaze-based, with low sampling rate requirements.

Example:When analyzing the viewing sequence of a painting, 200 Hz is sufficient to record the sequence of gaze points.

Above, that is how to choose the sampling rate of different devices all the introduction should be noted that, not the higher the sampling rate means that the equipment is more accurate, the better the signal quality, the sampling rate is too high will lead to a large amount of data and difficult to analyze, the sampling rate is too low will lead to distortion of the signal and the loss of key information. We need to choose the appropriate sampling rate according to the basis of the principle of the equipment, experimental needs, and research direction..