Heart rate (HR) is a cardiac activity metric that corresponds to the number of heartbeats per unit time (per minute). HR is commonly derived from the electrocardiogram (ECG) signal, which represents the electrical activity of the heart. HR can also be retrieved from photoplethysmography (PPG) devices that use light and photodetectors to observe volumetric changes of the blood through human skin. In the task of mental workload assessment, the topic of interest would be the variation in the interval between successive heartbeats. Heartrate variability (HRV) is a term that generally refers to changes in the inter-beat interval (IBI). The heart is known to be supplied with nerves from the sympathetic and the parasympathetic branches of the autonomic nervous system (ANS). The two systems show opposing effects on the HR, where activation of sympathetic system increases the HR and activation of parasympathetic system decreases the HR ^[1][40]. Several initial studies that were been conducted as early as in the 1980s have concluded that changes in the MWL can be observed through changes in the cardiovascular responses ^[2][30].

The traditional method for measuring mental workload using HRV metrics uses the time domain and frequency domain components ^[3][4][41,42]. The time domain category comprises statistical and geometrical measures that determine the time variability between heartbeats. Time domain analysis through statistical methods is further classified into two categories: (a) Direct measurements of Normal-to-Normal (NN) intervals, and (b) Differences between NN intervals. The NN intervals can also be derived as a geometrical pattern that can be studied using different approaches: (1) basic measurement of the pattern converted into a HRV measure, (2) interpolation of the geometrical pattern, and (3) categorisation of the geometrical pattern into classes of HRV. The analysis of HRV in the frequency domain usually involves the study of the power spectral density (PSD), which provides information about how power distributes as a function of frequency in the HRV signal. The spectral analysis of HRV is based on the Fourier theory, where a signal can be represented as the sum of sinusoidal signals consisting of amplitude, phase, and frequency components. The literature has confirmed that the spectral components of HRV can be categorised into very low frequency (VLF), low frequency (LF), high frequency (HF), and ultra-low frequency (ULF) components in short-term and long-term readings ^[5][43]. Short-term readings generally consider 2–5 min of HRV data, whereas long-term readings consider HRV data up to 24 h. The time and frequency domain components of HRV are summarised in Table 1.

^10][11][48,49]. The components and features of the RR are generally analysed using time, volume, and spectral parameters. The most popularly used features are the respiratory rate, inspiratory time, expiratory time, timing ratio, tidal volume, minute ventilation, and spectral power. A summary of RR features has been presented in Table 2.

The biggest oscillator in the human body that is involved in the maintenance of homeostasis is the phenomenon of respiration. It is well known that respiratory activation additionally influences psychological and behavioural processes along with metabolic changes ^[6][44]. RR also exhibits a close relationship with HR, as the coupling between them can provide information to be used as an index to study the vagal control of the heart ^[7][45]. Several studies have indicated that the respiratory rate (RR) is affected by emotional and cognitive demands reflecting limbic and paralimbic influences ^[8][9][46,47]. The psychophysiology effects of respiration are generally studied based on measures related to time, volume, and gas exchange ^[

Mental workload (MWL) and driver distraction are often related to each other, such that the same secondary task used to measure distraction can be used to measure MWL ^[12][50]. As the trend towards in-vehicle infotainment systems has modernised vehicles, distractions are often induced visually or cognitively ^[13][51]. Eye-related measures such as blink frequency, blink duration, and pupil dilation are recorded using head-mounted gear that cannot be used outside experimental conditions, as it causes discomfort. Further, if one opts for optical instruments such as cameras to gather eye-related data, multiple installations may be required, considering head movement and occluded situations. EDA (Electrodermal Activity) is a collective term used to define the bio-electrical changes that occur in the skin, which is one of the most useful indices of the sympathetic activity impacting sweat gland activity.

The functional concepts of EDA closely relate to psychophysiological activity, which makes it an important objective measure for understanding arousal, attention, and emotional responses. EDA is classified into two categories: skin conductance level (SCL) and skin conductance responses (SCR). SCR data are obtained through the placement of electrodes in positions such as below the distal phalanx of the index and middle finger. Similarly to with eye-related measures, the signal acquisition of the EDA components might become problematic, as data are generally affected by noise artifacts induced by driving movement.