We measured electroencephalography (EEG) data streams from participants wearing a wireless EEG headset in two modes: eyes open and eyes closed. Then we analyzed the data by computing the correlation coefficients for a pair of electrodes in each measurement mode. We also plotted and visually inspected the associated scatter plots. We observed that for the electrodes selected, the signals were more strongly correlated in the eyes closed mode and relatively weakly correlated in the eyes open mode. In most measurements, the signals were dissimilar. These observations could be harnessed to inform expedient placement of EEG electrodes and efficient selection of data stream channels for further analysis.


Download data is not yet available.


The accuracy and reliability of data in EEG recordings highly depends on proper placement of EEG (electroencephalogram) electrodes. These electrodes are strategically positioned on the scalp to capture electrical activities generated by the brain. On the reach of research or clinical objectives, the selection of electrode positions and data stream channels should therefore follow international standards or conventions to meet specific requirements.

Bigdely-Shamlo et al. [1] proposed a preprocessing pipeline (PREP) which consists of standardized preprocessing approach for EEG data. In this research, electrode placement was described, discussing the use of the 10–20 system and subsequently, provide guidelines for electrode selection and positioning. Oostenveld et al. [2] on the other hand, focused on the insensitivity of MEG and EEG data to skull defects, laying more emphasis on electrode placement. The research reiterated the influence of skull defects on the accuracy of EEG recordings and recommends electrode placements in the presence of skull defects. The placement of the electrode is also considered important in event-related potentials [3], [4]. Different cognitive processes are associated with specific brain regions, and electrodes can be placed over these regions to capture the corresponding electrical activity. Hence, certain features need to be considered when selecting electrode placements for ERPs. To optimize electrode placements to maximize the signal-to-noise ratio by minimizing artifacts and interference from non-brain sources, Speier et al. [5] consider electrode configurations for high-density EEG studies on auditory event-related potentials. Insights into effective strategies for maximizing signal quality considering the impact of electrode density, placement, and montage on the signal-to-noise ratio can be realized [6].

The number of electrodes and configuration has direct impact on high spatial resolution in EEG [7], [8]. The trade-offs between electrode density, scalp coverage, and signal-to-noise ratio should be balanced to achieve high spatial resolutions. However, low-resolution brain electromagnetic tomography (LORETA) to approximate electrical sources in the brain based on scalp can be used in EEG recordings [9]. Even with the association between different scalp regions and their practical implications in EEG data analysis, there still remains a method to standardize the reference point to a hypothetical point—see Dong [10]. It is important to note that the number of electrodes used contributes to a result, as higher sensor density allows for better spatial resolution and more precise localization of ERP components. Investigations into the impact of electrode density and placement on the performance of a brain-computer interface (BCI) using steady-state visually evoked potentials (SSVEPs) have shown that appropriate electrode placement can optimize signal quality and reduce artifacts in SSVEP-based BCIs [11]. However, the impact of electrode placement, impedance, and contact quality on EEG signal quality and the occurrence of artifacts can be identified and removed using blind source separation procedures [12], [13].

Mapping techniques generally involve representing spatial information or relationships between objects or features in a defined area. The concept of EEG micro-states with their corresponding mapping using space-oriented adaptive segmentation can also be utilized on the structural distribution of alpha activity across different scalp regions and its relationship with underlying brain micro-states [14]. Placing electrodes in appropriate positions and carefully selecting data stream channels can help researcher identify and remove artifacts, ensuring cleaner and more reliable EEG data.

The American Electroencephalographic Society [15] provides guidelines for standard electrode position nomenclature. The article provides an inclusive overview of electrode placements based on the 10–20 system with diagrams and descriptions of the electrode positions.

It becomes crucial to consult these guidelines and adhere to best practices to ensure methodological rigor and to enhance the comparability and reliability in conducting research involving ERPs and cognition. To strengthen this, Luck and Yee [16] gave distinctive overview of different ERP components and their use in studying cognition which covers various recording standards and publication criteria.

In order to promote the same consistency and rigor in recording and analyzing ERPs in cognitive research, Keil et al. [17] presented guidelines and recommendations for EEG studies, which are highly relevant to ERP research.

However, Handy and Picton [18] presented a study that can serve as a reference in reporting ERP research findings in a journal article.

EEG source localization can introduce errors Ollikainen et al. [19]. This research analyzed the effect of local skull conductivity inhomogeneity on source estimation accuracy. It showed that neglecting inhomogeneity of the skull conductivity can lead to localization errors of about 1 cm in the equivalent current dipole estimation. Odabaee et al. [20] corroborated the accuracy of the head model used in EEG source localization and its significant impact on solutions of the EEG inverse problem using empirical methods to indirectly estimate skull conductivity. This study revealed that a focal cortical activity is only registered by electrodes within few centimeters from the source. The study then admits that the conventional 10 to 20 channel neonatal EEG acquisition systems gives a significantly spatially under-sampled scalp EEG and may, consequently, give distorted pictures of focal brain activities.

Ramon et al. [21] considered the influence of head models on EEG simulations and inverse source localizations with emphasis on the structure of the cerebrospinal fluid (CSF) and gray and white matter. The study showed that these anatomical surfaces could severely influence conductivity in a head model which in turn influences the scalp potentials and the inverse source localizations. The results indicate that the complexity of head models strongly influences the scalp potentials and the inverse source localizations.

The previous research projects reviewed have demonstrated the position of proper electrode placement in reducing artifacts and present approaches to handle and eliminate them. They equally provide valuable insights into the relationship between scalp regions, underlying brain areas, and functional processes in EEG recordings. But only a few considered the analysis of Electroencephalography (EEG) Electrode-pair Correlations.

The aim of this study is to analyze electrode-pair correlations in Electroencephalography (EEG) data to provide insights into brain activity and connectivity. This study uses the Emotiv EPOC wireless EEG device, the capabilities of which have been demonstrated and characterized by researchers [4], [22], [23], [24].

Materials and Methods

Participant Recruitment and Experiments

Two healthy adult participants without any history of neurological disease volunteered to have their EEG data captured in two modes, namely, eyes open and eyes closed. A recording session lasted approximately 1 minute. We used an Emotiv EPOC wireless EEG headset to capture the EEG data streams.

We utilized two electrodes, T8 and P8 as depicted in Fig. 1. Recordings produced by the Emotiv EPOC account for the reference measurements and pass through an inbuilt high pass filter.

Fig. 1. Electrode placement for the experiments.

Ethical Approval

All participants read, understood, and signed informed consent forms and the studies were approved by the Research Ethics Committee at Topfaith University.

Data Availability

The raw EEG data supporting the conclusions of this study are available upon reasonable request from the corresponding author.


We calculate the pairwise correlation coefficients for the data streams captured at the two monitored electrodes.

Furthermore, we create the scatter plots for the data streams and analyze them to determine how the data streams are correlated.

The Pearson product-moment correlation coefficient, r, can be obtained as shown in (1). (1)r=∑i=1n(Xi−X¯)(Yi−Y¯)∑i=1n(Xi−X¯)2∑i=1n(Yi−Y¯)2

In (1), X¯ is the sample mean for the data {X}={X1,X2,…,Xn}, that is, the EEG data stream captured by one of the electrodes in the pair, T8 or P8 in Fig. 1 while Y¯ represents the sample mean for the data {Y}={Y1,Y2,…,Yn}, that is, the EEG data stream captured by the other electrode of the electrode pair, T8 or P8 in Fig. 1. The number of samples in the data set is encapsulated in the variable named n in (1).


In Figs. 2 to 5, we display the scatter plots we obtained for the electrode pairs for Participant 1 and Participant 2. Fig. 2 shows results for Participant 1 with Eyes Open while Fig. 3 illustrates results for Participant 1 with Eyes Closed. Similarly, Fig. 4 depicts results for Participant 2 with Eyes Open while Fig. 5 is the scatter plot for Participant 2 with Eyes Closed.

Fig. 2. Scatter plot for Participant 1: Eyes open.

Fig. 3. Scatter plot for Participant 1: Eyes closed.

Fig. 4. Scatter plot for Participant 2: Eyes open.

Fig. 5. Scatter plot for Participant 2: Eyes closed.

Table I provides a listing of the correlation coefficients measured for Participant 1 and Participant 2 under the Eyes Open and Eyes Closed modes.

Participant/mode Eyes open Eyes closed
Participant 1 0.1049 0.5041
Participant 2 0.3942 0.6459
Table I. Correlation Coefficients

The vertical axis is for the EEG data stream from the T8 electrode while the horizontal axis is for the EEG data stream from the P8 electrode. All measurements are in microvolts.

As is evident from Table I, the correlation coefficients in the Eyes Open mode for Participant 1, Participant 2 are 0.1049 and 0.3942, respectively. For the Eyes Closed mode, the correlation coefficients for Participant 1 and Participant 2 are 0.5041 and 0.6459, respectively. These results indicate that the correlation coefficient is lower for the Eyes Open mode than for the Eyes Closed mode for both participants. In the Eyes Open mode, the participants are immersed in s stimulus richer environment than in the Eyes Closed mode. Consequently, it is plausible that the greater variability in the recorded signals in the Eyes Open mode reflects increased brain activity due to the ability of the participants to perceive a wider variety of stimuli, including visual stimuli, in the Eyes Open mode compared to the Eyes Closed mode. The scatter plot in the Eyes Open mode for Participant 1 (Fig. 2) indicates greater variability and dissimilarity compared to the Eyes Closed mode shown in Fig. 3. Similarly, the scatter plot in the Eyes Open mode for Participant 2 (Fig. 4) indicates greater variability and dissimilarity compared to the Eyes Closed mode shown in Fig. 5. These observations are in line with the measured correlation coefficient values.

Bonita et al. [25] have demonstrated the resilience of the correlation coefficient in the presence of noise and its suitability as a reliable similarity metric.

By performing this analysis for different combinations of electrodes under varying experimental conditions, the results could be used to help inform expedient placement of electrodes and selection of channels for a wide variety of applications.


We have analyzed correlation coefficient measures and scatter plot graphs for a pair of EEG electrodes for participants in Eyes Open and Eyes Closed settings. The results we observed are amenable to application as a means of informing expedient placement of EEG electrodes in a wide variety of settings.


  1. Bigdely-Shamlo N, Mullen T, Kothe C, Su KM, Robbins KA. The PREP pipeline: standardized preprocessing for large-scale EEG analysis. Front Neuroinform. 2015;9:16. doi: 10.3389/fninf.2015.00016.
     Google Scholar
  2. Oostenveld R, Praamstra P, Stegeman DF, van Oosterom A. Insensitivity of MEG and EEG data to skull defects. Electroen Clin Neuro. 2001;98(4):352–6. doi: 10.1016/S0013-4694(00)00266-2.
     Google Scholar
  3. Picton TW, Bentin S, Berg P, Donchin E, Hillyard SA, Johnson R, et al. Guidelines for using human event-related potentials to study cognition: recording standards and publication criteria. Psychophysiol. 2000;37(2):127–52. doi: 10.1111/1469-8986.3720127.
     Google Scholar
  4. Williams NS, McArthur GM, Badcock NA. It’s all about time: precision and accuracy of Emotiv event-marking for ERP research. PeerJ. 2021;9:e10700. doi: 10.7717/peerj.10700.
     Google Scholar
  5. Speier W, Deshpande A, Pouratian N. A method for optimizing EEG electrode number and configuration for signal acquisition in P300 speller systems. Clin Neurophysiol. 2015;6(126):1171–7.
     Google Scholar
  6. Jeon S, Chien J, Song C, Hong J. A preliminary study on precision image guidance for electrode placement in an EEG study. Brain Topogr. 2018;31:174–85. doi: 10.1007/s10548-017-0610-y.
     Google Scholar
  7. Bleichner MG, Mirkovic B, Debener S. Identifying auditory attention with ear-EEG: cEEGrid versus high-density cap-EEG comparison. Journal of Neural Engineering. 2016;13(2016):066004 (13pp). doi: 10.1088/1741-2560/13/6/06600.
     Google Scholar
  8. Ryynänen O, Hyttinen JAK,Malmivuo J. Study on the spatial resolution of EEG—Effect of electrode density and measurement noise. The 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp. 4409–12, San Francisco, CA, USA, 2004. doi: 10.1109/IEMBS.2004.1404226.
     Google Scholar
  9. Pascual-Marqui RD, Michel CM, Lehmann D. Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain. Int J Psychophysiol. 1994;Oct;18(1):49–65. doi: 10.1016/0167-8760(84)90014-x.
     Google Scholar
  10. Dong L, Zhao L, Zhang Y, Yu X, Li F, Li J, et al. Reference electrode standardization interpolation technique (RESIT): a novel interpolation method for scalp EEG. Brain Topogr. 2021 Jul;34(4):403–14. doi: 10.1007/s10548-021-00844-2 Epub 2021 May 5. PMID: 33950323; PMCID: PMC8195908.
     Google Scholar
  11. Singh J, Ebersole JS, Brinkmann BH. From theory to practical fundamentals of electroencephalographic source imaging in localizing the epileptogenic zone. Epilepsia. 2022;63:2476–90. doi: 10.1111/epi.17361.
     Google Scholar
  12. Kappenman ES, Luck SJ. The effects of electrode impedance on data quality and statistical significance in ERP recordings. Psychophysiol. 2010 Sep;47(5):888–904. doi: 10.1111/j.1469-8986.2010.01009.x Epub 2010 Mar 29.PMID: 20374541; PMCID: PMC2902592.
     Google Scholar
  13. Kersschot I. Unipolar versus bipolar leads. Aubert AE, Ector H, Stroobandt R Eds. Cardiac Pacing and Electrophysiology. In: Dordrecht: Springer, 1994. doi: 10.1007/978-94-011-0872-0_20.
     Google Scholar
  14. Milz P, Faber PL, Lehmann D, Koenig T, Kochi K, Pascual-Marqui RD. The functional significance of EEG microstates–Associations with modalities of thinking. Neuroimage. 2016 Jan 15;125:643–56. doi: 10.1016/j.neuroimage.2015.08.023 Epub 2015 Aug 15. PMID: 26285079.
     Google Scholar
  15. American Electroencephalographic Society. Guidelines for standard electrode position nomenclature. J Clin Neurophysiol. 1994;11(1):111–3. doi: 10.1097/00004691-199401000-00014.
     Google Scholar
  16. Luck SJ, Yee CM. Committee report: publication guidelines and recommendations for studies using electroencephalography and magnetoencephalography. Psychophysiol. 2014;51(1):1–21.
     Google Scholar
  17. Keil A, Debener S, Gratton G, Junghöfer M, Kappenman ES, Luck SJ, et al. Committee report: publication guidelines and recommendations for studies using electroencephalography and magnetoencephalography. Psychophysiol. 2014;51(1):1–21.
     Google Scholar
  18. Handy TC, Picton TW. ERP indices of target and distractor processing in visual search. Psychophysiol. 2001;38(5):841–57.
     Google Scholar
  19. Ollikainen JO, Vauhkonen M, Karjalainen PA, Kaipio JP. Effects of local skull inhomogeneities on EEG source estimation. Med Eng Phys. 1999 Apr;21(3):143–54. doi: 10.1016/s1350-4533(99)00038-7.
     Google Scholar
  20. Odabaee M, Tokariev A, Layeghy S, Mesbah M, Colditz PB, Ramon C, et al. Neonatal EEG at scalp is focal and implies high skull conductivity in realistic neonatal head models. Neuroimage. 2014;96:73–80. doi: 10.1016/j.neuroimage.2014.04.007.
     Google Scholar
  21. Ramon C, Schimpf PH, Haueisen J. Influence of head models on EEG simulations and inverse source localizations. Biomed Eng Online. 2006;5:10. doi: 10.1186/1475-925X-5-10.
     Google Scholar
  22. Badcock NA, Mousikou P, Mahajan Y, De Lissa P, Thie J, McArthur G. Validation of the Emotiv EPOC EEG gaming system for measuring research quality auditory ERPs. PeerJ. 2013;1:e38. doi: 10.7717/peerj.38.
     Google Scholar
  23. Badcock NA, Preece KA, DeWit B, Glenn K, Fieder N, Thie J, et al.Validation of the Emotiv EPOCEEGsystem for research quality auditory event-related potentials in children. PeerJ. 2015;3:e907. doi: 10.7717/peerj.907.
     Google Scholar
  24. Choong WY, Khairunizam W, Mustafa WA, Murugappan M, HamidA, Bong SZ, et al. Correlation analysis of emotional EEG in alpha, beta and gamma frequency bands. J Phys. 2021, Conf. Ser. 1997 012029.
     Google Scholar
  25. Bonita JD, Ambolode LCC II, Rosenberg BM, Cellucci CJ,Watanabe TAA, Rapp PE, et al. Time domain measures of inter-channel EEG correlations: a comparison of linear, nonparametric and nonlinear measures. Cogn Neurodyn. 2014;8:1–15.
     Google Scholar