Contents lists available at ScienceDirect International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho Can we feel like being neither alert nor sleepy? The electroencephalographic signature of this subjective sub-state of wake state yields an accurate measure of objective sleepiness level Arcady A. Putilov a,b, ⁎ , Olga G. Donskaya a , Evgeniy G. Verevkin a a Research Group for Math-Modeling of Biomedical Systems, The Research Institute for Molecular Biology and Biophysics of the Federal Research Center of Fundamental and Translational Medicine, Novosibirsk, Russia b Laboratory of Sleep/Wake Neurobiology, The Institute of Higher Nervous Activity and Neurophysiology of the Russian Academy of Sciences, Moscow, Russia ARTICLE INFO Keywords: EEG spectrum Sleep deprivation Objective sleepiness Alertness-sleepiness level Sleep-wake transition Karolinska Sleepiness Scale ABSTRACT Accurate measurement of objective level of sleepiness can have important implications for experimental and field studies of sleep deprived individuals. We proposed to accurately quantify changes in sleepiness level with single electroencephalographic (EEG) measures obtained from EEG spectra consisting of 16 spectral powers within the frequency interval from 1 to 16 Hz. The EEG signal was recorded every other hour from 19:00 of Friday to 19:00 of Sunday in 48 study participants. The differential spectra were calculated for the 1st minute with eyes closed as the differences between EEG spectra for pairs of distinct subjective sub-states (alert, neither alert nor sleepy, sleepy, and very sleepy sub-states scored on the Karolinska Sleepiness Scale as 3, 5, 7, and 9, respectively).The differential spectra were calculated for the sub-samples of those participants who succeeded and failed to succeed in completing all 25 EEG recording sessions (n = 25 and 23, respectively) and for the addition sample of 130 participants deprived from sleep for only one night. Single spectral EEG measures were then calculated by summation of 16 spectral powers weighted by a differential spectrum. The strongest corre- lation coefficients (0.981, 0.987, and 0.985) were attained between the time courses of subjective and objective measures when data on 130, 23 and 25 participants, respectively, were used for calculation of frequency waiting curve differentiating alert sub-sate either from sleepy sub-state or from neither alert nor sleepy sub-state. We recommended implementation of the proposed objective measure into experimental procedures requiring ac- curate estimation of objective sleepiness level. 1. Introduction Sleepiness is a cause of motor vehicle accidents and should be considered when fitness to drive is investigated (Bioulac et al., 2017). Accurate measurement of objective level of sleepiness can have im- portant implications not only for the driving but also for many other safety critical procedures and occupations (Kaplan et al., 2007; King et al., 2009; Horne and Burley, 2010). It has been recognized that the lack of practical tools for quick and objective testing of sleepiness has become one of the critical barriers to reducing the threats of sleep loss to safety, productivity, and public health (Czeisler, 2011; Quan et al., 2011). Although it is believed that sleep deprived individuals are able to monitor their own sleepiness, the vast majority of people seem poor in such a monitoring (Reyner and Horne, 1998; Berka et al., 2004; Smith et al., 2005; Tremaine et al., 2010). It is difficult to precisely define the physiological component of subjective feeling of sleepiness in scientific terms (Reyner and Horne, 1998), and, therefore, the question of what might be a reliable physiological marker of sleepiness remains a con- troversial issue. The consensus has not been yet reached on the question of what are the patterns of changes in brain activity that can reliably signal about alertness decrement. On the other hand, there exists a solid experimental evidence for significant association of subjective sleepi- ness level with higher electroencephalographic (EEG) spectral power in low (delta and theta) frequency ranges and with lower EEG power in high (high alpha and low beta) frequency ranges, especially when EEG is recorded in eyes closed rather than eyes open condition (Lorenzo et al., 1995; Strijkstra et al., 2003; Leproult et al., 2003; Marzano et al., 2007; Putilov and Donskaya, 2014). We previously proposed to estimate the current objective level of alertness-sleepiness by using a single EEG measure based on a differ- ential spectrum that is a difference between EEG spectra for a pair of https://doi.org/10.1016/j.ijpsycho.2018.11.005 Received 13 October 2018; Received in revised form 18 November 2018; Accepted 19 November 2018 ⁎ Corresponding author at: 11, Nipkowstr., 12489 Berlin, Germany. E-mail address: putilov@ngs.ru (A.A. Putilov). International Journal of Psychophysiology 135 (2019) 33–43 Available online 22 November 2018 0167-8760/ © 2018 Elsevier B.V. All rights reserved. T