The practice of yoga synchronizes human physiology through controlled postures, breathing, meditation, a set of regular physical exercises, and relaxations [
Compared to physical exercise yoga may be more effective or even better in improving health related conditions. Despite corpus of research on the subjects, the lack of evidence based on scientific approaches has limited the application of yoga as an accepted method for improvement of health [
Therefore the objective of this study is to investigate the effectiveness of yoga practice and to evaluate physiological parameters related to cognitive aspects on novice subjects. The study primarily focused the effect of yoga on cognitive behavior in terms of physiological parameters. In the current study, the yoga practice involved combined practice of easy asanas (postures), meditation, and pranayama (breathing exercise). It is known that yoga involving relaxation techniques improve the functioning of cardiovascular autonomic nervous system. Yoga is correlated with decreased sympathetic adrenergic receptor sensitivity, which might affect cardiovascular response during stress [
Heart rate variability (HRV) is a measure of deviations in the interbeat R-R intervals. It is a noninvasive method used to assess the functioning of the autonomic nervous system (ANS), which is responsible for the regulations of many physiological processes of the human being [
Generally, for HRV analysis, parameters can be computed by two methods [ Time domain measures are directly computed from the time series of the RR intervals. In the literature there are many time domain measures available for HRV analysis. In this paper the following indices are used for its analysis: mHR: mean RR intervals; mHRV: mean heart rate variability and it indicates the total amount of deviations of both instantaneous HR and RR intervals. It reflects sympathetic and parasympathetic activity of the ANS on the sinus node of the heart; SDNN: standard deviation of all NN intervals and an indicative of global HRV. It indicates all the long term elements and circadian rhythms responsible for variability in the recording interval; RMSSD: the Square roots of the mean of the sum of the squares of differences between adjacent NN intervals and it reflects the short cyclical variability in the autonomic tone that is largely vagally mediated; CVRR: coefficient of variations of RR intervals and it is used to reflect the parasympathetic nervous system activity; the important time domain parameters are shown in Table Frequency domain parameters are computed by applying fast Fourier transform (FFT) to the time series of the raw RR intervals. FFT is the most powerful and efficient algorithm used to break the HRV signal into a series of sine and cosine components. This Fourier transformed signal is further translated to power spectrum by squaring magnitude of each [ Very low frequency (VLF: 0.0033–0.04 Hz) power: the function of this frequency range is not well defined but sometimes it can be used as the index of sympathetic activity of ANS. Low frequency (LF: 0.04–0.15 Hz) power: this band is complex in nature and an index of both sympathetic and parasympathetic activity and influences HRV patterns. High frequency (HF: 0.15–0.4 Hz) power: it is the index of parasympathetic activity and is used to indicate slow changes in the HR. Very high frequency (VHF: >0.4 Hz): this frequency is generally considered as noise and has no clinical significance. LF/HF ratio: It reflects the overall balance of the ANS. The lower ratio is recommended by the task force. In normal, in resting condition, this ratio lies in the range of 1 and 2. Total power (TP): variance of all NN intervals in the frequency range less than 0.4 Hz.
The equations used to compute time domain measures.
Index | Equations | Unit |
---|---|---|
mHRV |
|
ms |
|
||
mHR |
|
bpm |
|
||
SDNN |
|
ms |
|
||
RMSSD |
|
ms |
|
||
CVRR |
|
— |
The VLF, LF, HF, and TP are expressed in ms2 units, when computed in absolute values. The important frequency domain parameters used for the computation are shown in Table
The equations used to compute frequency domain measures.
Index | Equations | Unit |
---|---|---|
|
|
% |
|
||
|
|
% |
|
||
SVI |
|
— |
|
||
LFRel.power |
|
— |
|
||
HFRel.power |
|
— |
|
||
dLFHF |
|
% |
|
||
where TP = VLF + LF + HF |
The spectral parameters of HRV are usually normalized to minimize the effect of redundancy inherent in them in most of the research work. The important frequency domain parameters are shown in Figure
The brain activity which changes continuously with time is called “electroencephalogram” (EEG), which can be used to investigate the cognitive abilities and memory executions of individuals in terms of its band of frequencies.
The EEG is highly complex and is combination of five different frequency waveforms, namely,
Five EEG frequency bands.
Parameters | Frequency range (Hz) | Magnitude ( |
Activity | Remark |
---|---|---|---|---|
|
13–30 | <30 |
Desynchronized | Mental occupation |
|
8–13 | 50–100 |
Synchronized | Relaxed, tranquility, and wakefulness |
|
4–8 | 20–40 |
Desynchronized | Dreaming state |
|
0.5–4 | 75–150 |
Desynchronized | State of dreamless sleep |
|
30–70 | — | Synchronized | Sensory integration |
The EEG waveforms may be global or localized to the specific areas on the scalp. This kind of electrical data is important to study the correlation between yoga asanas and physiological states, because any shift in the EEG frequency range reflects the physiological arousal. The various EEG ratio indices and their physiological and cognitive interpretations are shown in Table
EEG band ratios and their physiological/cognitive activity index interpretation.
EEG band ratios | Activity/correlation | Sources |
---|---|---|
|
Heart rate (HR) | [ |
|
Performance enhancement index or “wellbeing” | [ |
|
Arousal index | [ |
|
Neural activity | [ |
|
Cognitive performance and attentional resource index | [ |
|
Task load index | [ |
( |
Executive load index | [ |
|
Brain perfusion | [ |
|
CNS arousal | [ |
( |
Sum of LF to HF ratio | [ |
|
Desynchronization | [ |
|
Synchronization | |
( |
Vigilance index | [ |
Discrete wavelet transforms (DWT) are widely used for the analysis of physiological signals as compared to the classical techniques such as fast Fourier transforms (FFT). When FFT is applied on the time series signal, the signal information is available in the form of spectral parameters. That is, the whole time domain information will be lost. It is equivalent to windowed Fourier transform and can be used to measure both the time and frequency changes of a signal [
The DWT splits the input signal into approximation (trend) and detailed coefficients (fluctuation), respectively. The approximation coefficient can further be split into a new approximation and detailed coefficients. This process is continued progressively to get a new set of approximation and detailed coefficients of a signal at various levels of decomposition [
In this study the EEG signal was acquired with sampling frequency of 500 Hz. The useful information of this signal lies in the range of 0.5–70 Hz. Hence a level of 7 using db4 was applied to decompose the EEG signal into its approximate (A1–A7) and detail (D1–D7) coefficients. After the seventh level of decomposition, the band of frequencies obtained are D1 (250–500 Hz), A1 (0–250 Hz), D2 (125–250 Hz), A2 (0–125 Hz), D3 (62.5–125 Hz), A3 (0–62.5 Hz), D4 (31.25–62.5 Hz), A4 (0–31.25 Hz), D5 (15.625–31.25 Hz), A5 (0–15.625 Hz), D6 (7.8125–15.625 Hz), A6 (0–7.8125 Hz), D7 (3.906–7.813 Hz), and A7 (0–3.906 Hz), respectively. The decomposition levels from D1 to D3 were considered as noise components and hence excluded from the analysis. The finer detailed coefficients from levels D4–D7 and final approximate coefficients from level A7 are retained as they approximately represent the EEG physiological frequency subbands of
The total number of subjects who participated in the experiment was 30 young healthy graduate and postgraduate engineering students of IIT Roorkee (male = 27, female = 3). All the subjects were right handed with normal eye sight. The study population was divided into two groups: experimental group and control group. In this study the sample size is relatively small and both groups have the same size. The study population was randomly assigned to either of the groups by block randomization method to achieve the balance. The block size of two was used. Both participants and investigators were unaware of the groups to be assigned in advance. Each group consisted of 15 subjects with two females in experimental group and one in control group. The mean and standard deviation of each group were 22.42 ± 2.30 and 23.67 ± 2.09, respectively. The same subjects were chosen for both experimental and control groups to diminish misperceiving influences and make the study more effective. Subjects with previous yoga practice, history of alcohol consumption, smoking, and any other drug consumption were excluded from this study. The participants were informed
In this study the physiological parameters such as ECG and their ratio indices have been evaluated to assess the cognitive benefits of the yoga practice along with its well established health benefits. This may provide the window for further investigation to correlate the actual measures of cognitive functions and their physiological parameters.
The practice of yoga schedule consisted of prayer, pranayama (breathing techniques), and simple yogic postures. Explanations on stress management, importance of meditation, and yoga in everyday life were also briefed.
The subjects practiced yoga under observation of trained yoga instructor. During practice session various types of asanas (postural exercises), pranayama (breathing techniques), and Standing asanas (postures): they consisted of surya namaskar, dandasana, urdhave asana, trikonasana, ardha asana, hasta padasana, mahavir asana, and vatayanasana. Sitting asanas (postures): they include mandook asana and oorm asana, ushtra asan, ardha matsyendrasana, vakrasana, supt asana, matsyendrasana, uttan mandukasana, vakasana, mayoor asan, padm vak asan, padma mayurasana, pashchimottanasan, eka padangusthasana, vipreet pad asan, and purna chakrasana. Asana (posture) lying on back: this includes uttan pad asana, pawanmuktasana, market asan, shreeshan, sarvangasana, halasana, setu bandhasana, and chakrasana. Asana (posture) lying on stomach: this includes naukasana, yan asan, shalabh asan, and dhanurasana. Pranayama (breathing) and kriya include Anulom-vilom, kapalbhati, Ujjayi pranayama, Bhramari pranayama, sheetali pranayama, sheetkari pranayama, surya bheda pranayama, bhastrika pranayama, bahya pranayama, udgeeth pranayama, kaki mudra, and shanmukhi mudra. Everyday practice session was concluded with prayer and meditation.
Both ECG and EEG signals were recorded simultaneously using BIOPAC MP150 System (EEG100C = 10 nos and ECG100C = 3 nos) with Acqknowledge 4.0 software.
ECG signal was recorded using five electrodes by connecting to left and right wrinkles and left and right arms and one electrode at chest. Before fixing the electrodes they were cleaned and electrode gel was applied to reduce the skin resistance to get good quality of recording signal. EEG signal was recorded by fixing the CAP100C on the scalp of the subjects. This cap was made of Lycra type fabric with 20 reusable tin electrodes attached to it, according to the international 10–20 norms. Before fixing the cap on subjects scalp, the electrodes were cleaned with saline water and electrode gel was applied. This maintains the resistance below 5 kΩ between the scalp and electrodes.
All the data were collected during 6 p.m. to 7.30 p.m. at the yoga centre of the temple premises in two stages. The data collected at the beginning of the intervention period was considered as the first stage, during which five minutes of baseline ECG and EEG signals were recorded from subjects of both experimental and control groups in sitting position with eyes closed. The baseline signal was saved on the hard disk for offline processing.
Both ECG and EEG signals were recorded with a sampling frequency of 512 Hz to have better resolution of R-R time interval series and EEG activity.
The experimental group practiced yoga for a period of five months for 1.5 hr per day in the evening from 6 p.m. to 7.30 p.m. The control group was asked not to practice any form of yogic practices or physical exercises during this period. The end of five months yoga training period was considered as second stage. During this stage again both ECG and EEG data were collected from experimental and control group for a period of 10 minutes. During recording, subjects were asked to minimize eye blinks and avoid body movements to minimize any artifacts that could be introduced. If artifacts were introduced due to uncontrolled body movements or eye blinks or due to technical reasons, the recording time was prolonged for a few more minutes. The data was again saved on the hard disk for offline processing. These data were used for the evaluation of various cognitive functions in terms of physiological parameters.
Though maximum care was taken, the recorded data was contaminated with many artifacts. Manual editing was performed for both ECG and EEG signals. The RR intervals were then extracted from the Acqknowledge 4.0 software which uses modified Pan and Tompkins algorithm. The intervals less than 300 ms and above 1200 ms were eliminated from time series data set and were saved in text format for further processing using MATLAB 7.1. Any data whose standard deviation was less than or equal to three times the standard deviation was considered outliers and removed from the data before determining the time domain parameters of heart rate variability (HRV). The artifact free data was segmented into five groups with 10 seconds segments each. The average value of each 10 seconds data was used in the analysis. The important time domain measures of HRV such as mean HRV, SDANN, RMSSD, and mean HR were computed.
The frequency domain parameters, namely, VLF, LF, HF, LF : HF ratio, and VHF, were extracted from the FFT algorithm. The normalized values were computed by dividing the respective frequencies by total power minus VLF. The normalization reduces the effect of LF and HF on total power. In conformity with the task force recommendation, The artifact free signal of two minutes duration was used for computing frequency domain parameters.
Results are grouped in two parts: firstly, HRV analysis, which includes time and frequency domain parameters; secondly, cognitive performance evaluation based on EEG engagement indices. Both EEG and ECG signals reflect global arousal or alertness of the brain [
The heart rate variability (HRV) is an indicator of cardiac ANS and HR is controlled by neural activity [
There was significant reduction in LF power (
Control group showed significant increase of LF power (
The LF and HF band power of HRV are expressed in normalized units. The representation of these frequency band powers in normalized units articulates the degree of control exercised and the relative balance of two limbs of the autonomic nervous system [
Student's paired
The various time and frequency domain parameters of both yoga and control group are shown in Table
Time and frequency domain parameters, before and after intervention of yoga and control group.
Parameters | Yoga group | Control group | ||||
---|---|---|---|---|---|---|
Before | After |
|
Before | After |
|
|
mHRV (ms) | 757.21 ± 65.37 | 813.29 ± 78.91 | 0.0304 | 874.67 ± 91.55 | 855.72 ± 70.48 | 0.2505 |
mHR (bpm) | 79.79 ± 7.74 | 74.57 ± 7.05 | 0.0389 | 69.50 ± 9.43 | 70.50 ± 6.22 | 0.2759 |
SDNN (ms) | 44.43 ± 21.76 | 52.14 ± 23.27 | 0.0012 | 53.22 ± 21.69 | 53.83 ± 19.51 | 0.9044 |
RMSSD (ms) | 39.93 ± 23.65 | 55.21 ± 22.78 | 0.0058 | 54.83 ± 29.46 | 49 ± 27.89 | 0.1999 |
SDNN/RMSSD | 0.77 ± 0.37 | 1.11 ± 0.57 | 0.0039 | 0.77 ± 0.37 | 1.21 ± 0.28 | 0.1336 |
VLF (n.u) | 4.02 ± 0.96 | 11.63 ± 9.83 | 0.0177 | 3.42 ± 9.91 | 14.04 ± 10.06 | 0.0007 |
LF (n.u) | 123.06 ± 21.10 | 56.85 ± 10.16 | 0.0002 | 45.14 ± 13.09 | 41.57 ± 16.60 | 0.0000 |
HF (n.u) | 27.39 ± 8.44 | 43.15 ± 10.16 | 0.0003 | 46.48 ± 4.61 | 2.20 ± 5.79 | 0.0269 |
|
71.33 ± 5.67 | 51.16 ± 9.51 | 0.0000 | 46.42 ± 10.70 | 70.81 ± 7.13 | 0.0000 |
|
27.39 ± 8.44 | 43.15 ± 10.16 | 0.0003 | 46.48 ± 4.61 | 42.20 ± 5.79 | 0.0269 |
TP (n.u) | 171.99 ± 22.41 | 111.63 ± 9.83 | 0.0000 | 95.03 ± 11.29 | 97.81 ± 12.44 | 0.0000 |
LF : HF | 4.99 ± 1.98 | 1.411 ± 0.45 | 0.0000 | 0.99 ± 0.32 | 3.55 ± 1.56 | 0.0000 |
The decrease in HR could be due to combined effect of elements of yoga. The reduction in stress after yoga could be other possible reason for improved HRV in this study. The previous researches suggest that yoga practice results in neurophysiological balance by lowering level of cholinesterase and catecholamines. Further, this result increased parasympathetic and decreased sympathetic activity. The results of this study are in concurrence with previous studies [
The regular practice of yoga for a period of five months by young healthy engineering students resulted in the increase of
The various cognitive behavior parameters have been evaluated based on various EEG indices such as
Mean powers of EEG frequency bands averaged across all the lobes of the brain before and after yoga intervention.
EEG band powers | |||||
---|---|---|---|---|---|
|
|
|
|
| |
Yoga group | |||||
Before yoga | 3.80 ± 0.93 | 2.32 ± 0.49 | 3.95 ± 0.70 | 21.36 ± 3.43 | 4.22 ± 0.42 |
After yoga | 3.65 ± 0.69 | 4.91 ± 1.63 | 5.67 ± 1.68 | 15.88 ± 2.57 | 5.79 ± 1.06 |
Control group | |||||
Before yoga | 2.98 ± 1.36 | 3.86 ± 0.96 | 3.92 ± 0.97 | 17.57 ± 2.54 | 4.46 ± 0.89 |
After yoga | 2.94 ± 1.33 | 3.72 ± 0.82 | 3.87 ± 0.90 | 21.12 ± 3.87 | 4.37 ± 0.78 |
The increase of frontal
The ratio
This ratio increased (47.11%) in yoga group while decreased in control group (2.43%). The increases of this ratio indicate enhanced cognitive functions such as attention. The decreases of this ratio reflect reduction in the core capabilities of cognitive functions. This ratio increased in all lobes of the brain but the maximum increase was observed in parietal (78.05%), central (53.12%), and temporal (45.04%) lobes. It increased to (38.24%) and (19.76%) in frontal and occipital lobes, respectively. The ratio (
Cognitive index parameters of yoga group before and after yoga intervention. Postintervention values are shown within the parenthesis.
EEG indices | Yoga group | ||||
---|---|---|---|---|---|
Frontal | Central | Parietal | Occipital | Temporal | |
|
4.621 (2.806) | 4.951 (2.063) | 5.059 (2.973) | 8.792 (4.609) | 4.604 (2.480) |
|
0.216 (0.356) | 0.202 (0.485) | 0.198 (0.336) | 0.114 (0.217) | 0.217 (0.403) |
|
0.612 (0.846) | 0.625 (0.957) | 0.523 (0.931) | 0.607 (0.727) | 0.564 (0.818) |
|
0.132 (0.302) | 0.126 (0.464) | 0.103 (0.313) | 0.069 (0.158) | 0.123 (0.330) |
|
0.109 (0.222) | 0.105 (0.312) | 0.086 (0.234) | 0.062 (0.130) | 0.101 (0.235) |
|
4.621 (2.806) | 4.951 (2.063) | 5.059 (2.973) | 8.792 (4.609) | 4.604 (2.480) |
( |
5.590 (3.926) | 5.964 (3.200) | 6.168 (3.961) | 10.112 (5.907) | 5.614 (3.243) |
|
Global EEG band power ratios: before and after yoga intervention.
EEG indices | Yoga group | Control group | ||
---|---|---|---|---|
Before yoga | After yoga | Before yoga | After yoga | |
|
5.405 | 2.800 | 4.485 | 5.460 |
|
0.185 | 0.357 | 0.228 | 0.183 |
|
0.588 | 0.865 | 0.986 | 0.962 |
|
0.109 | 0.309 | 0.220 | 0.176 |
|
0.092 | 0.228 | 0.180 | 0.149 |
|
5.405 | 2.800 | 4.485 | 5.460 |
( |
6.472 | 3.820 | 5.623 | 6.590 |
|
46.392 | 27.876 | 40.456 | 47.080 |
The ratio
EEG indices and their values: before and after yoga intervention.
Parameters | Yoga group | Control | ||
---|---|---|---|---|
Before | After | Before | After | |
|
0.937 | 0.980 | 0.879 | 0.885 |
( |
4.08 | 2.05 | 3.32 | 2.89 |
|
1.70 | 1.16 | 101.39 | 103.94 |
|
0.197 | 0.364 | 0.209 | 2.488 |
|
9.198 | 3.235 | 5.530 | 4.721 |
EEG index values of yoga group in different lobes of the brain: before and after yoga intervention. Postintervention values are shown in parenthesis.
Parameters | Brain lobes | ||||
---|---|---|---|---|---|
Frontal | Central | Parietal | Occipital | Temporal | |
|
1.032 (0.893) | 0.988 (0.879) | 0.902 (1.011) | 0.757 (0.770) | 0.990 (1.312) |
( |
3.468 (0.534) | 3.669 (0.566) | 4.050 (0.402) | 6.294 (0.363) | 3.589 (0.363) |
|
1.634 (2.806) | 1.599 (2.063) | 1.913 (2.973) | 1.648 (4.609) | 1.772 (2.480) |
|
0.210 (0.814) | 0.205 (0.524) | 0.219 (0.614) | 0.150 (0.570) | 0.219 (0.657) |
|
7.550 (3.316) | 7.916 (2.156) | 9.675 (3.194) | 14.492 (6.341) | 8.159 (3.032) |
EEG index values of control group in different lobes of the brain: before and after yoga intervention. Postintervention values are shown in parenthesis.
Parameters | Brain lobes | ||||
---|---|---|---|---|---|
Frontal | Central | Parietal | Occipital | Temporal | |
|
0.914 (0.944) | 0.940 (0.940) | 0.760 (0.760) | 0.825 (0.825) | 0.930 (0.931) |
|
0.424 (0.356) | 0.326 (0.261) | 0.271 (0.241) | 0.262 (0.218) | 0.486 (0.438) |
|
3.428 (4.186) | 3.711 (4.891) | 6.412 (7.311) | 5.513 (6.817) | 4.385 (4.970) |
|
0.716 (0.741) | 0.445 (0.445) | 0.524 (0.523) | 0.409 (0.409) | 1.433 (1.433) |
|
3.269 (4.185) | 3.958 (5.251) | 5.747 (6.647) | 6.633 (8.531) | 4.235 (4.839) |
The relative EEG band powers of
EEG band powers of yoga group in various lobes of the brain: before and after intervention.
Global EEG band powers of yoga group: before and after intervention.
The total band power (global) of
CNS activity, engagement index, and executive load index of yoga group: before and after intervention.
EEG band powers of control group in various lobes of the brain: before and after intervention.
There were no significant changes in EEG band powers, engagement indices, total power, and cognitive performances indices among control group. The various performances parameters of control group are shown in Figures
Global EEG band powers of control group: before and after intervention.
Global EEG band powers of control group in various lobes of the brain: before and after intervention.
CNS activity, engagement index, and executive load index of control group: before and after intervention.
The regular practices of yoga for a period of five months by young healthy engineering students enhance different types of cognitive skills. Apart from cognitive, the yoga practice resulted in many health benefits such as improvement in heart rate variability. The ratio SDNN/RMSSD increased while the ratio LF/HF decreased. This indicates improvement in the parasympathetic activity and decrease in sympathetic activity. Hence the current results suggest that the practice of yoga modifies the sympathovagal balance towards parasympathetic activation, improved the heart rate variability, and enhanced sense of wellbeing. Since the study population is young healthy engineering graduates, it would be interesting to investigate whether the yoga practice could result in improvement in the academic performances.
In a nutshell, it is proved beyond doubt that yoga practices resulted in effective improvements in physiological parameters, indirectly improving psychological parameters and various cognitive functions. The results of this study greatly encourage further investigation to study whether the practice of yoga could also enhance academic performance.
Limitations of this study include small number of samples and lack of dedicated control group and methodological differences. Further investigation can be done by employing psychological tests to evaluate cognitive behavior.
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors acknowledge the Principal of PDA College of Engineering, Kalaburagi, for deputing H. Nagendra to pursue Ph.D. degree and Department of Electrical Engineering IIT Roorkee for providing an opportunity under Quality Improvement Programme (QIP). They also acknowledge the students who participated in this study and give special thanks to the yoga instructor.