**3. Effects of sleep loss on cognition**

Although what constitutes sufficient quality and quantity of sleep per day remains a subject of debate due to its the wide interindividual variability and age-related differences, it has been elucidated that sleep loss deteriorates various cognitive functions, whether due to partial or total sleep deprivation and chronic or acute sleep disturbance. Loss of sleep also impairs the activities of various cerebral regions or neural networks associated with ongoing cognitive performance. A recent study in rats suggested that sleep loss often elicits periods of local sleep, in which some neurons often go 'offline' briefly in one cortical area but not in another during long periods of wakefulness (Vyazovskiy, 2011). Several basic cognitive functions are vulnerable to sleep loss in humans. These include simple response speed (Buysse et al., 2005; Frey et al., 2004; Koslowsky & Babkoff, 1992), psychomotor vigilance (Blatter et al., 2005; Doran et al., 2001; Drake et al., 2001; Van Dongen et al., 2003), mental arithmetic (Frey et al., 2004; Stenuit & Kerkhofs, 2008; Van Dongen et al., 2003;), response inhibition (Drummond et al., 2006; Stenuit & Kerkhofs, 2008), problem solving (Killgore et al., 2008; Nilsson et al., 2005), and short-time perception (Soshi et al., 2010). However, the performance of executive functions, one of the higher cognitive functions that includes divided attention (Drake et al., 2001; Frey et al., 2004; Lim & Dings, 2010; Stenuit & Kerkhofs, 2008) and working memory (Bartel et al., 2004; Binks et al., 1999; Choo et al., 2005; Frey et al., 2004; Lim & Dings, 2010; Tucker et al., 2010; Wimmer et al., 1992) varies among studies; some report a significant effect of sleep loss (Bartel et al., 2004; Choo et al., 2005; Drake et al., 2001; Frey et al., 2004; Stenuit & Kerkhofs, 2008), while others report no such effect (Binks et al., 1999; Lim & Dings, 2010; Tucker et al., 2010; Wimmer et al., 1992).A discrepancy has also been seen in the influence of sleep loss has on behavioral performance versus its influence on neural activity; functional neuroimaging has revealed that sleep loss deteriorates not behavioral performance but neural activity (Choo et al., 2005). Such a discrepancy could point to a difference in neural substrates between basic and higher cognitive functions and/or possible personal differences in vulnerability of executive functions to sleep loss. We describe here two of our studies conducted using NIRS in order to explore the influence of sleep loss on basic and higher cognition associated with the frontal functions mentioned above.

#### **4. Influence of sleep loss due to total deprivation of a night's sleep on time perception**

#### **4.1 Short-time perception**

26 Infrared Spectroscopy – Life and Biomedical Sciences

and deoxy-Hb, and the scalp and skull are high permeable to near-infrared light (Obrig et al., 2000). When such light is locally irradiated from an irradiation probe, it diffuses in the cerebral tissue up to a depth of 20-30 mm. A detection probe located 30 mm from the irradiation probe can detect the light diffusely reflected by the oxy- or deoxy-Hb, making it possible to estimate local changes in oxy-, deoxy- and total-Hb concentrations (Ferrari et al., 2004). For high-resolution detection of oxy- and deoxy-Hb concentrations, multiple channels of wavelengths (2 or 3) of near-infrared light (700-1000 nm) are usually simultaneously

NIRS has been widely used for several years in medical and biological studies of the brain. Although NIRS uses an accessible, non-invasive neuroimaging device, it should be applied to measure local cerebral metabolic rate of oxygen consumption with consideration given to

Although what constitutes sufficient quality and quantity of sleep per day remains a subject of debate due to its the wide interindividual variability and age-related differences, it has been elucidated that sleep loss deteriorates various cognitive functions, whether due to partial or total sleep deprivation and chronic or acute sleep disturbance. Loss of sleep also impairs the activities of various cerebral regions or neural networks associated with ongoing cognitive performance. A recent study in rats suggested that sleep loss often elicits periods of local sleep, in which some neurons often go 'offline' briefly in one cortical area but not in another during long periods of wakefulness (Vyazovskiy, 2011). Several basic cognitive functions are vulnerable to sleep loss in humans. These include simple response speed (Buysse et al., 2005; Frey et al., 2004; Koslowsky & Babkoff, 1992), psychomotor vigilance (Blatter et al., 2005; Doran et al., 2001; Drake et al., 2001; Van Dongen et al., 2003), mental arithmetic (Frey et al., 2004; Stenuit & Kerkhofs, 2008; Van Dongen et al., 2003;), response inhibition (Drummond et al., 2006; Stenuit & Kerkhofs, 2008), problem solving (Killgore et al., 2008; Nilsson et al., 2005), and short-time perception (Soshi et al., 2010). However, the

irradiated and detected.

 **inexpensive high portability easy-to-use approach** 

its strong and weak points, which are listed in Table 1.

**independence of a specific measurement setting** 

 **difficulty in strict identification of anatomic locations narrow range of measurement (only cortical surface) relative quantitation (not absolute quantitation)** 

**Strengths (compared with MRI or PET)** 

 **high tolerance to body movements high temporal resolution (10 Hz or less) high tolerance to long-time measurements utility regardless of subject's posture** 

**Weaknesses (compared with MRI or PET)** 

Table 1. Strengths and weaknesses of NIRS

**3. Effects of sleep loss on cognition** 

**low spatial resolution** 

When elapsed time is comparatively brief (within several minutes), humans can typically perceive the passage of time accurately without referring to an artificial time keeping device such as a wristwatch (Ivry, 1996; Rammsayer, 1999; Treisman, 1963). Human short-time perception is modulated by a robust neural basis consisting of subcortical structures, such as the cerebellum and basal ganglia, together with the right prefrontal cortex (Harrington et al., 1998; Pouthas et al., 1999). Moreover, a circadian pacemaker located in the suprachiasmatic nucleus of the hypothalamus, which is driven by a self-sustaining oscillator with a period of about 24 h and provides the time of day, participates in short-time perception (Aschoff, 1998; Ashoff & Daan, 1997; Kuriyama et al., 2003). As such, short-time perception is not independent of the influence of the circadian pacemaker; under a condition where zeitgebers are strictly controlled, short-time perception fluctuates on around a 24-h cycle and correlates with circadian markers such as core body temperature and melatonin, and consequently shows diurnal variation (Kuriyama et al., 2005). It has been confirmed that short-time perception shortens from morning into night, and is prolonged again from night to the next morning under a 30-h constant routine (Kuriyama et al., 2005; see Fig. 1). For sleep deprivation on the other hand, it has been reported that there is less diurnal variation

Fig. 1. Diurnal fluctuation of short-time perception

Effects of Sleep Debt on Cognitive Performance and Prefrontal Activity in Humans 29

Fig. 2. NIRStation FOIRE-3000 (Shimazu Co., Tokyo, Japan)

Fig. 3. Schematic layout of NIRS probes with recorded channels on the frontal region

left anterior PFC (LAPFC) region, based on statistical analyses, were superimposed.

Behavioral data suggested that time perception fluctuates through the night to the morning in the SC condition; TP was significantly prolonged from night to the next morning. However, TP was not prolonged from night to the next morning in the SD condition (Fig. 4).

**4.4 Effects of sleep loss on short-time perception** 

window of 1.1 s. Data were normalized into z-scores to avoid the methodological ambiguity that changes in absolute values of Hb concentration for each recording channel would not be determined because the absolute path lengths of light through the cerebral cortex were not detectable. Concentration changes time-locked to trial onset were extracted from 5 s before to 27 s after the onset, covering a mean produced time of around 11 s and a mean rest interval of around 16 s. A total of 15 epochs were obtained for each experimental day (day 1 or day 2) in each condition (SC or SD). Before individual averaging, baselines were corrected with mean z-scores of 5 s before trial onset. Grand averaged concentration changes in the

in short-time perception as it is dissociated from endogenous circadian markers (Kuriyama et al., 2005; Soshi et al., 2010). This may be because short-time perception is modulated by the prefrontal cortex (PFC) along with subcortical structures including the circadian pacemaker; it is well known that the PFC is vulnerable to sleep loss, and this vulnerability presumably disturbs short-time perception (Soshi et al., 2010). To elucidate this issue, Soshi et al. (2010) utilized NIRS because it neither produces pulse scanning noise nor requires severe restriction of the subject's body posture nor movements, and thus is unlikely to seriously influence the sleep-deprived condition. It is also suitable for monitoring the subject's condition while performing the experimental tasks.

### **4.2 Study design**

Fourteen healthy male university students participated in a crossover design study conducted over a 4-day period. Subjects performed a 10-s time production (TP) task in sleep controlled (SC) and sleep deprived (SD) conditions, scheduled in random order with a 1-day interval (Fig. 4). On the first day (day 1) NIRS probes were attached to the surface of the scalp. The 15-min TP session in either the SC or SD condition started at 21:00. After the session, in the SC condition, subjects rested without sleep or exercise until 0:00 and then stayed in bed under complete darkness (> 0.1 lux) until 08:00 on day 2; in the SD condition, subjects stayed awake quietly under room light (100 lux) until 08:00 the next morning while being monitored by video. On day 2, the TP session started again at 09:00. All the experiments were performed at a time isolation facility, and the ambient temperature and humidity were maintained constant throughout the study.

TP tasks were arranged in an event-related design to detect the hemodynamic response for a single trial. TP sessions were conducted at 21:00 on day 1 and 09:00 on day 2, corresponding to the expected nadir and peak period of the diurnal variation of TP in subjects with a regular sleep-wake cycle. Each TP session consisted of 15 trials with 30-s inter-trial intervals. Subjects were asked to produce a 10-s interval and to begin and end each trial by pressing a key button (Kuriyama et al., 2003, 2005). Duration from the first to the second button presses was defined as the perceived time.

#### **4.3 NIRS recording and data analysis**

Regional hemodynamic changes in brain tissue were monitored throughout the TP sessions by a continuous wave-type NIRS system (FOIRE-3000; Shimazu Co., Tokyo, Japan; Fig. 2) which outputs near-infrared light at three wavelengths (780, 805 and 830 nm). All transmitted intensities of the three wavelengths were recorded every 130 ms at 22 channels in order to estimate concentration changes in oxy-Hb, deoxy-Hb, and total-Hb, based on the modified Beer-Lambert equation as a function of light absorbance of Hb and pathlength. A set of 3×5 probes were utilized, in which light detectors and emitters were alternately positioned at an equal distance of 30 mm. The 22 channels (see Fig. 3) covered the middle and superior PFC regions (BA9, 46, 10).

Oxy-Hb data was chosen to examine event-related responses in the PFC since it is an optimal index for changes in regional cerebral blood flow (Hoshi et al., 2001). We applied a high-pass filter to raw data, re-sampled at 10 Hz, using a low-cutoff frequency of 0.05 Hz. Smoothing was performed by the moving average method (boxcar filter) with a sliding time

in short-time perception as it is dissociated from endogenous circadian markers (Kuriyama et al., 2005; Soshi et al., 2010). This may be because short-time perception is modulated by the prefrontal cortex (PFC) along with subcortical structures including the circadian pacemaker; it is well known that the PFC is vulnerable to sleep loss, and this vulnerability presumably disturbs short-time perception (Soshi et al., 2010). To elucidate this issue, Soshi et al. (2010) utilized NIRS because it neither produces pulse scanning noise nor requires severe restriction of the subject's body posture nor movements, and thus is unlikely to seriously influence the sleep-deprived condition. It is also suitable for monitoring the

Fourteen healthy male university students participated in a crossover design study conducted over a 4-day period. Subjects performed a 10-s time production (TP) task in sleep controlled (SC) and sleep deprived (SD) conditions, scheduled in random order with a 1-day interval (Fig. 4). On the first day (day 1) NIRS probes were attached to the surface of the scalp. The 15-min TP session in either the SC or SD condition started at 21:00. After the session, in the SC condition, subjects rested without sleep or exercise until 0:00 and then stayed in bed under complete darkness (> 0.1 lux) until 08:00 on day 2; in the SD condition, subjects stayed awake quietly under room light (100 lux) until 08:00 the next morning while being monitored by video. On day 2, the TP session started again at 09:00. All the experiments were performed at a time isolation facility, and the ambient temperature and

TP tasks were arranged in an event-related design to detect the hemodynamic response for a single trial. TP sessions were conducted at 21:00 on day 1 and 09:00 on day 2, corresponding to the expected nadir and peak period of the diurnal variation of TP in subjects with a regular sleep-wake cycle. Each TP session consisted of 15 trials with 30-s inter-trial intervals. Subjects were asked to produce a 10-s interval and to begin and end each trial by pressing a key button (Kuriyama et al., 2003, 2005). Duration from the first to the second button presses

Regional hemodynamic changes in brain tissue were monitored throughout the TP sessions by a continuous wave-type NIRS system (FOIRE-3000; Shimazu Co., Tokyo, Japan; Fig. 2) which outputs near-infrared light at three wavelengths (780, 805 and 830 nm). All transmitted intensities of the three wavelengths were recorded every 130 ms at 22 channels in order to estimate concentration changes in oxy-Hb, deoxy-Hb, and total-Hb, based on the modified Beer-Lambert equation as a function of light absorbance of Hb and pathlength. A set of 3×5 probes were utilized, in which light detectors and emitters were alternately positioned at an equal distance of 30 mm. The 22 channels (see Fig. 3) covered the middle

Oxy-Hb data was chosen to examine event-related responses in the PFC since it is an optimal index for changes in regional cerebral blood flow (Hoshi et al., 2001). We applied a high-pass filter to raw data, re-sampled at 10 Hz, using a low-cutoff frequency of 0.05 Hz. Smoothing was performed by the moving average method (boxcar filter) with a sliding time

subject's condition while performing the experimental tasks.

humidity were maintained constant throughout the study.

was defined as the perceived time.

**4.3 NIRS recording and data analysis** 

and superior PFC regions (BA9, 46, 10).

**4.2 Study design** 

Fig. 2. NIRStation FOIRE-3000 (Shimazu Co., Tokyo, Japan)

Fig. 3. Schematic layout of NIRS probes with recorded channels on the frontal region

window of 1.1 s. Data were normalized into z-scores to avoid the methodological ambiguity that changes in absolute values of Hb concentration for each recording channel would not be determined because the absolute path lengths of light through the cerebral cortex were not detectable. Concentration changes time-locked to trial onset were extracted from 5 s before to 27 s after the onset, covering a mean produced time of around 11 s and a mean rest interval of around 16 s. A total of 15 epochs were obtained for each experimental day (day 1 or day 2) in each condition (SC or SD). Before individual averaging, baselines were corrected with mean z-scores of 5 s before trial onset. Grand averaged concentration changes in the left anterior PFC (LAPFC) region, based on statistical analyses, were superimposed.

#### **4.4 Effects of sleep loss on short-time perception**

Behavioral data suggested that time perception fluctuates through the night to the morning in the SC condition; TP was significantly prolonged from night to the next morning. However, TP was not prolonged from night to the next morning in the SD condition (Fig. 4).

Effects of Sleep Debt on Cognitive Performance and Prefrontal Activity in Humans 31

(Harrington et al., 1998; Pouthas et al., 1999) the present study failed to detect any significant changes in right PFC activity (Soshi et al., 2010). It has also been argued that increased activation of the PFC after sleep deprivation is associated with neural compensation for cognitive function, although TP on day 2 in the SD condition was different from that in the

NIRS has a serious shortcoming in that it cannot determine whether or not a subcortical network including the cerebellum and the basal ganglia contributes to attenuating the diurnal fluctuation of short-time perception. If subcortical activities may be altered by sleep deprivation, attenuation of short-time perception possibly reflects subcortical vulnerability,

A temporary decline in the diurnal variation of short-time perception may be important for surviving a crisis, such as in an emergency situation. Time perception in humans should be fundamentally synchronized to the physical state to allow for constant adaptation to the regularity of daily life; however, in times of severe stress, time perception must desynchronize from regular physical homeostasis and be shortened, to enable time expansion and presumably allow us to adopt suitable strategies for coping with the stressful environment by thinking and acting more rapidly than usual. In-depth consideration of the adaptive nature of the PFC function in humans (Duncan, 2001; Miller & Cohen, 2001) suggests that the PFC might play a switch-like role in short-time perception as a situation

**5. Influence of sleep loss due to partial deprivation of a night's sleep on** 

As already noted in Section 3 concerning the effects of sleep loss on cognition, a discrepancy in the effects that sleep loss has on behavioral performance compared with the effects it has on neural activity has been reported in relation to working memory performance; sleep loss does not deteriorate behavioral performance itself, but rather the neural activity associated with the behavior. Possible interindividual differences in the vulnerability to sleep loss of executive functions, which also play crucial roles in working memory processing, have also been suggested. We explored this issue in a second study using NIRS (Honma et al., 2010).

Fifty-five healthy university students (26 males, 29 females) participated in the study. Subjects, who regularly slept 7–9 h in a night, participated in an overnight experiment in a laboratory setting, starting at 22:00 on day 1 and finishing at 10:00 on day 2. Subjects were deprived an average of 2.32 h (29.5%) of sleep by experimental manipulation. Subjects

A visual *n*-back working memory (WM) task (Callicott et al., 1998, 1999; Gevins & Cutillo, 1993; Kuriyama et al., 2008) with two separate load levels was utilized. For the 0-back task (low-load WM task), subjects had to respond whenever a single-digit number appeared on a screen. For the 2-back task (high-load WM task), they had to press a button on the right when the single-digit number on the screen was identical to that which had appeared last

retired to bed at 01:00 in the laboratory and were forcibly awakened at 07:00 am.

and thus the PFC activity change is possibly only a byproduct.

demands, helping us meet demands for adaptation.

**5.1 Working memory performance** 

**5.2 Study design** 

**individual differences in working memory performance** 

SC condition.

Fig. 4. Sleep deprivation attenuates short-time perception the following morning

It was previously shown that a short-time perception profile exhibits diurnal variation, reaching a peak (the longest produced time) around 09:00 and a nadir (the shortest produced time) around 21:00 with a regular sleep-wake cycle under experimental conditions (Kuriyama et al., 2005). Taken together, circadian oscillation in short-time perception under the SD condition is clearly attenuated.

#### **4.5 Influence of the PFC's vulnerability to sleep loss on short-time perception**

Oxy-Hb concentration measured by NIRS suggested that PFC activity in the SD condition, compared with that in the SC condition, was more enhanced in the left hemisphere on day 2. Moreover, enhanced oxy-Hb concentration changes on day 2 in the SD condition, compared with those in the SC condition, were observed in the LAPFC region of interest (ROI) at channels 17, 21, and 22 (Fig. 5).

Fig. 5. Left anterior PFC activity during the TP task was enhanced after sleep deprivation

A functional correlation was observed between increased activation of the LAPFC after sleep deprivation and short-time perception, although unlike in previous studies (Harrington et al., 1998; Pouthas et al., 1999) the present study failed to detect any significant changes in right PFC activity (Soshi et al., 2010). It has also been argued that increased activation of the PFC after sleep deprivation is associated with neural compensation for cognitive function, although TP on day 2 in the SD condition was different from that in the SC condition.

NIRS has a serious shortcoming in that it cannot determine whether or not a subcortical network including the cerebellum and the basal ganglia contributes to attenuating the diurnal fluctuation of short-time perception. If subcortical activities may be altered by sleep deprivation, attenuation of short-time perception possibly reflects subcortical vulnerability, and thus the PFC activity change is possibly only a byproduct.

A temporary decline in the diurnal variation of short-time perception may be important for surviving a crisis, such as in an emergency situation. Time perception in humans should be fundamentally synchronized to the physical state to allow for constant adaptation to the regularity of daily life; however, in times of severe stress, time perception must desynchronize from regular physical homeostasis and be shortened, to enable time expansion and presumably allow us to adopt suitable strategies for coping with the stressful environment by thinking and acting more rapidly than usual. In-depth consideration of the adaptive nature of the PFC function in humans (Duncan, 2001; Miller & Cohen, 2001) suggests that the PFC might play a switch-like role in short-time perception as a situation demands, helping us meet demands for adaptation.
