OUP user menu

The process of awakening: a PET study of regional brain activity patterns mediating the re‐establishment of alertness and consciousness

Thomas J. Balkin, Allen R. Braun, Nancy J. Wesensten, Keith Jeffries, Mary Varga, Paul Baldwin, Gregory Belenky, Peter Herscovitch
DOI: http://dx.doi.org/10.1093/brain/awf228 2308-2319 First published online: 1 October 2002


Awakening from sleep entails rapid re‐establishment of consciousness followed by the relatively slow (20–30 min later) re‐establishment of alertness—a temporal dissociation that facilitates specification of the physiological underpinnings of each of these facets of the awakening process. H215O PET was used to assess changes in regional cerebral blood flow (rCBF) upon awakening from stage 2 sleep. Cerebral blood flow (CBF) was most rapidly re‐established in centrencephalic regions (e.g. brainstem and thalamus), suggesting that the reactivation of these regions underlies the re‐establishment of conscious awareness. Across the ensuing 15 min of wakefulness, further increases in CBF were evident primarily in anterior cortical regions, suggesting that the dissipation of sleep inertia effects (post‐awakening performance and alertness deficits) is effected by reactivation of these regions. Concomitant shifts in correlation patterns of regional brain activity across the post‐awakening period [in particular, a waning negative correlation between prefrontal cortex and mesencephalic reticular formation (RF) activity, and a waxing positive correlation between prefrontal cortex and ventromedial caudate nucleus (CAUD) activity] suggest that the post‐awakening reversal of sleep inertia effects may be mediated by more than mere reactivation—it may also involve the functional reorganization of brain activity. Conversely, stable post‐awakening correlations—such as those found between the anterior cingulate cortex (ACC) and most other brain regions—may denote the pattern of functional connectivity that underlies consciousness itself.

  • Keywords: alertness; consciousness; sleep inertia; rCBF; PET
  • Abbreviations: ACC = anterior cingulate cortex; CAUD = ventromedial caudate nucleus; CBF = cerebral blood flow; rCBF = regional cerebral blood flow; RF = mesencephalic reticular formation


The neurophysiological basis of several aspects of human consciousness has recently been explored using functional brain imaging techniques. For example, insights into the regional mediation of brain processes underlying perceptual awareness have resulted from studies of hallucinations (Ffytche et al., 1998), subjective perceptual shifts when dissimilar images are concurrently presented to the two eyes (Lumer et al., 1998) and implicit versus explicit awareness of visual stimuli (e.g. by studying patients who exhibit ‘blindsight’; Sahraie et al., 1997). The strategy employed in these functional brain imaging studies is straightforward; scans obtained during normal, conscious functioning are contrasted with those obtained during an abnormal or otherwise altered state (such as a shift in the level of perceptual awareness). Thus, functional brain imaging techniques have been used to highlight the brain regional activation/deactivation patterns that mediate even the most subtle nuances of human conscious experience.

In the present study, a comparable strategy was used to determine the pattern of brain activity that underlies alertness—the aspect of waking conscious experience that reflects extant sleep/wake tendency—in an attempt to distinguish alertness from consciousness [in its broadest sense, the amalgamation of the mental processes that distinguish wakefulness from sleep, i.e. ‘the ability to perceive, interact and communicate with the environment and others in an integrated manner’ (Zeman, 2001)].

Prior functional brain imaging studies of sleep deprivation (e.g. Thomas et al., 2000) suggest that alertness varies primarily as a function of brain activation levels in the thalamus and prefrontal cortices—with reduced activity in these regions foreshadowing the more robust pattern of deactivation that characterizes sleep itself (e.g. Maquet et al., 1996; Braun et al., 1997).

However, more precise specification of the neurophysiological substrate of consciousness—and an enhanced ability to differentiate those processes mediating consciousness from those mediating alertness–may be gleaned from functional brain imaging studies in which the awakening process—i.e. the transition from sleep to full alertness—is examined. This is because the first several minutes of wakefulness constitute a state of reduced alertness like that produced by sleep deprivation, except that there is an initial mismatch between alertness level and underlying sleep debt. This mismatch resolves over the first ∼20 min of continuous wakefulness—a time frame that allows characterization of any underlying changes in regional brain activity using H215O PET methods.

Post‐awakening deficits in alertness (called ‘sleep inertia effects’; Lubin et al., 1976) include decrements in psychomotor performance and cognitive performance, marked hypovigilance (Tassi and Muzet, 2000) and sometimes bewilderment (Kleitman, 1963). Prior behavioural studies of sleep inertia indicate that the greatest deficits occur shortly after awakening, and that post‐awakening improvements accrue in a decelerating, asymptotic manner (Jewett et al., 1999).

These post‐awakening deficits are qualitatively similar to those resulting from sleep deprivation (Balkin and Badia, 1988)—despite the fact that actual sleep debt should be lowest immediately upon awakening (since sleep debt should accumulate with every minute of wakefulness). Therefore, it is likely that sleep inertia effects reflect the intrusion of residual (and waning) sleep maintenance mechanisms into the waking state.

The present study is the first to characterize changes in regional cerebral blood flow (rCBF) during the post‐awakening period. We perform contrasts that capitalize on the naturally occurring differential time courses for the re‐establishment of consciousness versus alertness, and also examine functional connectivity in an effort to specify the patterns of regional brain activity that underlie each.

Material and methods


Subjects were 27 healthy male volunteers (age 21–32 years). On the basis of medical history, physical examination and baseline laboratory evaluation, all subjects were free of neurological and psychiatric illness. Subjects with a history of sleep disorders or who had used prescription medications within 30 days preceding the study were excluded. The study was conducted using a protocol approved by the NIH NINDS review board and the U.S. Army Surgeon General’s Human Subjects Review Board. Informed consent was obtained from all subjects in accordance with the Declaration of Helsinki after all potential risks, discomforts, and hazards had been explained.

Pre‐scan sleep schedule

To help ensure that sleep would be obtained in the scanner, all subjects underwent a 3‐day partial sleep deprivation procedure prior to scanning (for details, see Braun et al., 1997). Ambulatory polysomnographic recorders (Oxford Medilog 9000‐II; Oxford Instruments Medical, Hawthorne, NY, USA) were used to measure and record EEG from C3 and C4 sites (Jasper, 1958), submental EMG and electro‐oculogram (EOG) (from the outer canthus of each eye) to verify wakefulness during the sleep restriction periods. During scheduled sleep periods, signals from the ambulatory recorders were routed through a Nihon Kohden electroencephalograph (Model EEG‐4317B; Nihan Kohden America Inc, Foothill Ranch, CA, USA) for visual monitoring of the subjects’ sleep.

Scanning methods

Scans were performed on a Scanditronix PC2048‐15B tomograph (Uppsala, Sweden), which has an axial and in‐plane resolution of 6.5 mm. Fifteen planes, offset by 6.5 mm (centre to centre), were acquired simultaneously parallel to the cantho‐meatal line. Prior to placement in the scanner, indwelling arterial and venous catheters were inserted into the radial artery and antecubital vein of subjects’ right and left forearms, and a new set of electrodes (EEG, EOG, EMG) for polysomnographic monitoring in the scanner (using a Grass Model 8–10D polygraph; Grass Telefactor, West Warwick, Rhode Island, USA) were attached to the scalp and face at the same sites described above. Subjects’ eyes were patched and head motion was restricted for the duration of the study with an individually fitted thermoplastic face mask that was affixed to the frame of the scanner bed. A 30 mCi bolus of H215O was injected intravenously and scans were initiated automatically when the radioactive count rate in the brain (automatically detected by scanner) reached a threshold value of 50,000 per s (∼20 s after injection) and were continued for 4 min. Sixteen scan frames were collected (twelve 10 s scans followed by four 30 s scans). Arterial blood was sampled automatically throughout each scan, and arterial time–activity data and blood gas measures were used with the scans to produce quantitative pCO2‐corrected rCBF images (see Braun et al., 1997). Emission data were corrected for attenuation by means of a transmission scan obtained at the same levels.

As reported previously (Braun et al., 1997), scans were acquired prior to sleep and during sleep stages 2, 3–4 and rapid eye movement (REM) (Rechtschaffen and Kales, 1968).

As the focus of the present report, scans were also performed during the post‐sleep period as follows. After ∼3–5 h of sleep in the scanner, subjects were awakened from stage 2 sleep by an investigator who entered the scanner room and spoke the subject’s first name. When the subject responded, he was instructed to remain awake and motionless until the scanning procedures were completed. There was no further communication until all scans were completed and subjects were removed from the scanner. H215O was injected intravenously following 5 min of continuous, polysomnographically verified wakefulness, and again after 20 min of verified, continuous wakefulness (i.e. 15 min later).


PET scans were registered, normalized to a common stereotaxic space (Talairach and Tournoux, 1988) and smoothed using a Gaussian kernel of 20 × 20 × 12 mm in the x, y and z axes. Absolute pCO2‐corrected global flow rates were calculated for each subject by averaging grey matter pixel values. Global flow rates were compared across conditions. Global cerebral blood flow (CBF) was also used to proportionally normalize each image on a pixel by pixel basis, and normalized rCBF rates were compared in the pairwise contrasts. As reported previously (Braun et al., 1997), if significant differences in global flow rates were detected, results of the proportionally normalized contrasts are reported, but interpreted in context. Thus, if increases in absolute CBF rates only were observed in comparing scans from two time points, only those normalized comparisons revealing regional increases were considered indices of real change. Decreases in normalized flow rates were interpreted as identifying brain regions in which absolute values deviated the least—i.e. were associated with absolute invariance or possibly with minimal, non‐significant increases in absolute rCBF. On the other hand, when significant differences in absolute pCO2‐corrected global flow were not detected, normalized comparisons are simply reported as indices of relative change.

Differences between rCBF levels at 5 min post‐awakening versus 20 min post‐awakening, during stage 2 sleep versus 5 min post‐awakening and during stage 2 sleep versus 20 min post‐awakening were analysed using statistical parametric mapping (SPM) software (MRC Cyclotron Unit, London, UK). Of the 27 subjects from whom post‐awakening scans were acquired: 13 had post‐awakening scans at both 5 and 20 min; 11 had scans during both stage 2 sleep and at 5 min post‐awakening; and 11 had scans during both stage 2 sleep and at 20 min post‐awakening. The stage 2 sleep‐20 min post‐awakening contrast (highlighting the differences in regional activation that differentiate alert wakefulness from sleep) and the 5–20 min post‐awakening contrast (highlighting the differences in regional activation that differentiate sleep inertia from normal, alert wakefulness) were then compared. Because the stage 2 sleep‐20 min post‐awakening contrast revealed that all post‐awakening changes in CBF were increases, results from that contrast were used to mask those of the 5 min/20 min post‐awakening contrast on a voxel by voxel basis, so that local minima or maxima could be interpreted in that context.

Correlation analyses were also performed to evaluate regional interconnectivity patterns at 5 and 20 min post‐awakening. Rather than perform large‐scale correlations using an extended set of brain regions, connectivity patterns were characterized in three potentially important regions. These were selected on the basis of their previously demonstrated involvement in both sleep stage transitions and sleep/wake transitions, and grounded in findings from the present study:

(i) The mesencephalic reticular formation (RF) [selected because the earliest studies point to this region as an important mediator of arousal (Moruzzi and Magoun, 1949)].

(ii) The ventromedial caudate nucleus (CAUD) [a stage REM sleep‐specific disparity between activation levels in the caudate nucleus and the prefrontal regions—with which neuronal connections are extensive—suggests a key role for this region in sleep state mediation (Braun et al., 1997)].

(iii) The anterior cingulate cortex (ACC) [selected because of its role in the mediation of attention (e.g. Davis et al., 2000) and its especially robust sleep stage‐dependent changes in activity (Braun et al., 1997)].

Selection of regional coordinates was based on local Z‐score minima or maxima from the present study when these exceeded threshold (i.e. in the RF and CAUD). When threshold was not exceeded (i.e. in the ACC), coordinates selection was based on the results from comparable sleep stage contrasts reported previously (Braun et al., 1997).

The PET images processed using SPM software were also used in the correlation analyses. Normalized rCBF values for voxels throughout these images were correlated across the cohort of subjects, with values derived from the seed voxels of interest at 5 and 20 min post‐awakening, utilizing software written in MATLAB (Horwitz et al., 1998). This routine produces a normalized output image with Pearson product–moment correlation coefficients assigned to each pixel in the image. Coefficients were transformed to standard scores and thresholded at Z > 2.0 in absolute value.


Changes in CBF

Global changes

The awakening process was characterized by a global increase in absolute CBF levels. During stage 2 sleep (the sleep stage from which all awakenings were initiated), the global CBF rate was 39.8 ± 2.9 ml/100 mg/min. At 5 min post‐awakening, the global CBF rate had increased significantly to 45.5 ± 2.5 ml/100 mg/min, with no substantive further increase at 20 min post‐awakening when the global CBF rate was 45.8 ± 2.0 ml/100 mg/min.

Changes in rCBF from stage 2 sleep to 5 min post‐awakening

Awakening‐mediated increases in CBF were not homogeneous; pairwise contrasts revealed that initial post‐awakening increases in rCBF occurred primarily in centrencephalic regions: brainstem, thalamus and basal ganglia (see Table 1 and Fig. 1A), whereas few significant changes were evident in the anterior cortical areas. Since sleep inertia effects are typically manifest at 5 min post‐awakening, this pattern of regional differences reveals the changes in brain activity patterns that characterize stage 2 sleep versus wakefulness with impaired alertness.

Fig. 1 Brain map depicting increases in rCBF between stage 2 sleep and 5 min post‐awakening (top row) and between stage 2 sleep and 20 min post‐awakening (bottom row). The SPM {z} map illustrating these differences is displayed on a standardized MRI scan, which was transformed linearly into the same stereotaxic (Talairach) space as the SPM {z} data. Planes of section relative to the anterior commissural–posterior commissural line are indicated (z‐axis coordinates in mm). Values are Z‐scores representing the significance level of changes in proportionally normalized rCBF in each voxel when scans acquired at 5 or 20 min post‐awakening are contrasted with those acquired during stage 2 sleep as baseline. The range of scores is coded in the accompanying colour table, with red designating Z‐scores of ≥+4.0. Locations of local minima and maxima for Z‐scores are summarized in Tables 1 and 2.

View this table:
Table 1

Stage 2 sleep versus 5 min and 20 min post‐awakening

Stage 2 sleep/5 min post‐awakeningStage 2 sleep/20 min post‐awakening
Region of interestBrodmann areaZ‐score x y z Z‐score x y z
 Heteromodal association
   Orbital operculum472.472628–12*4.112628–12
   Dorsal operculum453.15382620
   Lateral orbital cortex103.922442–8
   Dorsolateral prefrontal cortex463.18363416
   Medial prefrontal cortex93.00–84816
   Anterior cingulate24/322.69–828163.31–63220
   Anterior insula3.1530280
   Midbrain reticular formation3.03–10–2402.008–140
 Basal ganglia
   Caudate 3.6512683.31–14228
 Basal forebrain
   Caudal orbital cortex253.341826–12

Regions in which pCO2‐corrected rCBF levels at 5 and 20 min post‐awakening differ from stage 2 sleep are tabulated along with Z‐scores (representing maxima and associated Talairach coordinates). For tabulation, increases or decreases in CBF exceeding threshold in a single hemisphere were considered truly lateralized only if contralateral values failed to reflect a trend (defined as Z > 1.64 in absolute value) in the same direction. *No corresponding increases in contralateral hemisphere.

Changes in rCBF from stage 2 sleep to 20 min post‐awakening

Pairwise contrasts revealed that regional increases tended to occur in anterior cortical, paralimbic–limbic and subcortical regions (see Table 1 and Fig. 1B), whereas no significant changes were evident in the posterior cortical areas. Since the awakening process is known to be relatively complete after 20 min of continuous wakefulness (i.e. sleep inertia effects have typically dissipated to a considerable extent by this time), the regional differences revealed by these contrasts constitute the changes in brain activity patterns that distinguish stage 2 sleep from normal wakefulness.

Changes in rCBF from 5 min post‐awakening to 20 min post‐awakening

Although the awakening process was associated with a global increase in CBF, there were no significant changes in global CBF over the first 20 min of wakefulness. Therefore, both increases and decreases in normalized rCBF values are reported as indices of relative change across the first 20 min of wakefulness. These specify the pattern of rCBF changes that underlie dissipation of sleep inertia effects—i.e. the ascent from consciousness with impaired alertness to consciousness with relatively normal alertness—and are depicted in Fig. 2.

Fig. 2 Brain map depicting changes in rCBF between 5 and 20 min after awakening from stage 2 sleep. Data are processed and displayed as for Fig. 1. Values are Z‐scores representing the significance level of changes in proportionally normalized rCBF in each voxel when scans acquired at 20 min are contrasted with those acquired at 5 min as baseline. Positive scores represent increases in relative blood flow from 5 to 20 min; negative scores represent concomitant, relative decreases. The range of scores is coded in the accompanying colour table, with red designating Z‐scores of ≥+4.0 and purple designating Z‐scores of ≤–4.0. Locations of local minima and maxima for Z‐scores are summarized in Table 3.

Those regions in which CBF increased across the 5–20 min post‐awakening period were primarily heteromodal neocortical areas, i.e. orbital and dorsolateral prefrontal cortices, frontal opercular cortex, middle temporal gyrus and superior temporal sulcus. In addition, significant increases in rCBF from the 5th to the 20th minute were evident in the anterior insula and the caudal orbital region of the basal forebrain (both of which have extensive interconnections with prefrontal cortices) and in portions of the auditory cortices.

Unlike the stage 2 sleep‐20 min post‐awakening contrast, no increases were evident in brainstem, basal ganglia or thalamus across the 5–20 min post‐awakening period. This indicated that reactivation of these brain regions was relatively complete by 5 min post‐awakening. In fact, as indicated by negative Z‐scores in Table 2, relative decreases were evident in several of these regions during the post‐awakening period. Changes in relative rCBF across the first 20 min of wakefulness are listed in Table 2 and illustrated in Fig. 2.

View this table:
Table 2

Changes in rCBF from the 5th to the 20th min after awakening

Region of interestBrodmann areaZ‐score x y z
 Heteromodal association
   Orbital operculum473.34–1824–16
   Dorsal operculum44/ 455.0244128
   Dorsolateral prefrontal cortex463.70383616
   Lateral orbital cortex112.75–2438–12
   Middle temporal gyrus /STS213.5056–5412
 Unimodal sensory
   Posterior superior temporal gyrus223.38–54–3416
   Fusiform gyrus19, 37–3.72–28–60–8
   Lateral occipital cortex19, 18–3.36–36–860
   Anterior insula3.9840108
   Midbrain reticular formation–3.48–6–30–4
 Basal ganglia
   Ventromedial caudate–2.84–142–8
 Basal forebrain
   Caudal orbital cortex253.22–1426–16

Regions in which normalized rCBF levels differ from the 5th to the 20th min post‐awakening are tabulated along with Z‐scores, representing local minima or maxima and associated Talairach coordinates. Positive Z‐scores designate increases and negative Z‐scores designate decreases in relative rCBF. For tabulation, increases or decreases exceeding threshold in a single hemisphere were considered truly lateralized only if contralateral values failed to reflect a trend (defined as Z > 1.64 in absolute value) in the same direction. *No corresponding changes in contralateral hemisphere.

Changes in functional connectivity patterns Mesencephalic reticular formation (RF)

At both 5 and 20 min post‐awakening, activity in the RF (x = –6, y = –30, z = –4; coordinates specifying the maximal difference between 5 and 20 min; Table 2) was correlated with activity in a number of cortical and subcortical regions. Positive correlations were evident with the cerebellar hemispheres, putamen, thalamus, auditory cortices, ACC, hippocampus and the parahippocampal gyrus. Negative correlations were evident with pre‐ and post‐central gyri, and ventral visual cortices.

In contrast, activity levels in the RF and prefrontal cortices—including the medial, dorsolateral, lateral orbital and opercular cortices—were negatively correlated at 5 min post‐awakening, but as listed in Table 3 and depicted in Fig. 3B, these correlations were no longer significant by 20 min post‐awakening. This pattern was also evident in middle temporal gyrus, auditory cortices and insula, but was especially striking in the operculum, where the correlation with RF activity was particularly robust at 5 min post‐awakening. This constituted the strongest correlation between the RF and all other brain regions at this earlier post‐awakening time point and the site of maximal change between 5 and 20 min (see Table 2).

Fig. 3 Correlations between rCBF rates in (A) CAUD, (B) RF and (C) ACC (coordinates for each region selected as outlined in the text) and regions within the prefrontal cortex. Values on the x and y axes represent normalized rCBF rates measured at 5 min (left) and 20 min (right) post‐awakening. Each plotted point therefore represents the activation levels from two voxels (one from each of the two specified brain regions) in a single subject, at a given point in time (5 or 20 min post‐awakening). Regression lines are shown for each statistically significant correlation. (A) Blood flow in CAUD (Talairach x = –14, y = –2, z = –8) and orbitofrontal cortex (x = –26, y = 42, z = –8) was uncorrelated at 5 min, but positively correlated (r = 0.81, Z = 4.7, P < 0.0001) at 20 min post‐awakening. (B) Blood flow in the RF (x = –6, y = –30, z = –4) and dorsolateral prefrontal cortex (x = –34, y = 48, z = 12) was negatively correlated at 5 min (r = –0.82, Z = 4.8, P < 0.0001), but uncorrelated by 20 min post‐awakening. However, as shown in C, blood flow in the ACC (x = –6, y = 40, z = –4) and orbitofrontal cortex (x = –12, y = 52, z = –8) was positively correlated at both 5 min (r = 0.64, Z = 3.1, P < 0.001) and 20 min (r = 0.70, Z = 3.6, P < 0.0005) post‐awakening.

View this table:
Table 3

Correlations between normalized rCBF values in selected regions

5 min post‐awakening20 min post‐awakening
Region of interestZ‐score x y z Z‐score x y z
  Medial orbital–2.66–2038–12*2.17–636–12*
  Medial prefrontal–3.18–185020*
  Dorsolateral prefrontal–4.00–344212***
  Ventral opercular–3.06–42204*
  Dorsal opercular–4.48–48820****
  Medial orbital4.67–1642–8***
  Medial prefrontal3.93–22444*
  Dorsolateral prefrontal3.29–384012*
  Ventral opercular3.233222–4**
  Dorsal opercular2.42442216*
  Medial orbital3.23–840–84.00–1240–8
  Medial prefrontal3.93–846162.52–104616
  Dorsolateral prefrontal–2.34–342824–2.13–322428
  Ventral opercular2.25382403.6640240
  Dorsal opercular–2.4342228–2.62361028

Correlations of normalized rCBF values in RF, CAUD and ACC with rCBF values in frontal cortical regions at 5 and 20 min post‐awakening. Z‐transformed correlation coefficients, designating local maxima or minima, are tabulated along with associated Talaraich coordinates. Asterisks signify differences between transformed coefficients, at the indicated coordinates. *ΔZ >±1.64, **ΔZ >± 2.33, ***ΔZ >±2.57, ****ΔZ >±3.09.

The dynamic range of rCBF rates did not vary significantly from 5 to 20 min post‐awakening [F(1,19) = 1.33, P = 0.53]. This suggests that the observed differences in correlation patterns at these two times truly reflect different patterns of functional connectivity rather than relative differences in the stability of the rCBF rate across the two time points.


At both 5 and 20 min post‐awakening, activity in the CAUD (x = –14, y = 2, z = –8; coordinates specifying the maximal difference between 5 and 20 min; Table 1) was significantly correlated with activity in several cortical and subcortical brain regions. Positive correlations were evident with activity levels in midbrain tegmentum, other portions of the basal ganglia (contralateral caudate, globus pallidus bilaterally, ipsilateral putamen), ventral precentral gyri, anterior auditory cortices, portions of the middle temporal gyri bilaterally, caudal orbital cortices, temporal pole, ACC, hippocampus and amygdala. Negative correlations with CAUD activity levels at both 5 and 20 min post‐awakening were evident in the dorsal precentral gyrus, post‐central gyri, posterior auditory association cortices and ventral visual cortices.

In addition, a positive correlation between activity in the caudate and the inferior insula was evident at 5 min, but not at 20 min post‐awakening—although positive correlations between the caudate and both the anterior and posterior insula subsequently emerged at 20 min.

In contrast, activity in the caudate was not correlated with activity in the majority of prefrontal cortices at 5 min post‐awakening (the single exception being a positive correlation between CBF rates in the caudate and medial prefrontal cortex—which was evident at both 5 min and 20 min post‐awakening), nor with activity in the thalamus. Nevertheless, by 20 min, activity in the caudate was positively correlated with activity in a wide array of prefrontal regions including orbital, medial and dorsolateral cortices, and opercular cortices (see Table 3, Fig. 3A); as well as in the functionally related dorsal thalamus.

As in the RF, comparison of the variances in rCBF rates for the CAUD revealed no significant differences in dynamic range at 5 versus 20 min post‐awakening [F(1,19) = 1.28, P = 0.59], again reinforcing the assertion that differences in correlations across these times reflect meaningful differences in the pattern of functional connectivity.

Anterior cingulate cortex (ACC)

Since ACC activity did not vary significantly over the post‐awakening period, selection of coordinates for this region was based on the peak difference evident during a prototypical sleep stage contrast [x = –6, y = 40, z = 8, Z = 3.42, stage 3–4 sleep versus REM, as reported previously by Braun et al. (1997)].

Positive correlations were evident at both 5 and 20 min between activity levels in the ACC and the midbrain tegmentum, caudate, putamen, thalamus, medial orbital cortex, inferior medial and dorsolateral prefrontal cortices, ventral operculum, anterior auditory association cortices, caudal orbital cortex, temporal pole, anterior and posterior insula, other portions of the ACC, hippocampus and amygdala. Likewise, the negative correlations with ACC that were evident at 5 min post‐awakening persisted at 20 min post‐awakening, including those with superior portions of the medial prefrontal and dorsolateral prefrontal cortices, dorsal operculum, pre‐ and post‐central gyri, and visual cortices.

Therefore, unlike the patterns of correlations reflecting the functional connections of the reticular formation and caudate, correlations between the ACC and other brain regions (including prefrontal and opercular cortices) remained relatively stable across the post‐awakening period (see Table 3 and Fig. 3C).

Fig. 4 includes brain maps illustrating the functional connectivity patterns of some frontal cortical regions with the CAUD, RF and ACC at both 5 and 20 min post‐awakening.

Fig. 4 Brain map illustrating correlations between rCBF values in (A) CAUD, (B) RF, (C) ACC and frontal cortical regions at 5 and 20 min post‐awakening. Maps illustrating Z‐transformed correlation coefficients are displayed on a standardized MRI scan using the methods outlined for Fig. 1. The ranges of positive and negative Z‐scores are coded in the accompanying colour tables. Locations of local minima and maxima are summarized in Table 2. In (A), +3 mm relative to the anterior commissural–posterior commissural (AC–PC) line, positive correlations between rCBF in the caudate (Talairach x = –14, y = –2, z = –8) and lateral prefrontal cortices and operculum are manifest after a delay of 20 min. In (B), +17 mm relative to the AC–PC line, blood flow in the RF (x = –6, y = –30, z = –4) and medial and dorsolateral prefrontal cortices was negatively correlated at 5 min and uncorrelated at 20 min post‐awakening. In (C), +1 mm relative to the AC–PC line, blood flow in the ACC (x = –6, y = 40, z = –4) was positively correlated with that in medial and lateral prefrontal and opercular cortices at both 5 and 20 min post‐awakening.


Unique changes in the pattern of regional brain activity across the first 20 min of wakefulness and concomitant changes in patterns of regional interconnectivity (reflected in the correlation analyses) were investigated in an effort to differentiate the neural substrate of consciousness (the post‐awakening re‐establishment of awareness of self and the environment) from that of alertness (as reflected by the post‐awakening dissipation of sleep inertia‐related hypovigilance). Since consciousness and alertness are re‐established at differential rates following awakening from sleep, PET scanning at two time points shortly following awakening was used to extricate the physiological underpinnings of these two aspects of the awakening process.

First, with respect to alertness, the finding that global CBF rates were constant across the first 20 min of wakefulness indicates that post‐awakening improvement does not accrue simply as a function of generally increasing levels of brain activation. In this respect, these findings differ from those of prior functional brain imaging studies during sleep deprivation, from which it could be surmised that sleepiness is associated with global deactivation (e.g. Thomas et al., 2000). Rather than increases in global activity, focal differences—reflecting reactivation of critical brain regions and/or re‐establishment of the functional circuitry that typifies normal waking brain function—must account for post‐awakening increases in alertness. In the present study (and partially consistent with the findings during sleep deprivation reported by Thomas et al., 2000), the most notable regional changes (in terms of spatial extent and peak differences in activity) across the post‐awakening period were evident in the prefrontal association cortices. However, unlike prior sleep deprivation studies, which showed concomitant reductions in thalamic and prefrontal cortical activity, activity levels in these two regions were dissociated in the present study: Reactivation was complete in the thalamus at 5 min post‐awakening (a time when sleep inertia effects are typically manifest) and thus preceded reactivation of prefrontal cortices. Therefore, the present findings eliminate thalamic deactivation as a physiologically necessary component of hypovigilance. Instead, hypovigilance must ultimately be a function of reduced prefrontal cortical activity. Consciousness, on the other hand, must be a function of activity in those regions (or a subset of those regions) in which reactivation upon awakening is relatively rapid: brainstem, thalamus, basal ganglia and ACC.

However, just as it is unlikely that sleep can be distinguished from wakefulness solely on the basis of the activation level of individual brain regions, it is unlikely that any individual brain region mediates either the re‐establishment of consciousness that accompanies awakening or the reversal of alertness and performance deficits that occurs during the post‐awakening period. This is because brain regions do not function independently—they typically operate as elements in a series of networks distributed throughout the CNS. Accordingly, it is likely that the various sleep/wake states and consciousness itself are emergent consequences of functional interactions between brain regions.

The correlational findings from the present study are, for example, consistent with the notion that increasing alertness during the post‐awakening period is an emergent product of orchestrated interregional activation patterns, i.e. these analyses suggest that, in addition to regional reactivation, the post‐awakening process involves functional reorganization. In some instances, the reorganization involves re‐establishment of those functional circuits that are purported to characterize the coherent and orchestrated activity of the normal, alert brain. Thus, the correlations between caudate, prefrontal cortex and thalamus (that were not evident until 20 min post‐awakening) may signify re‐establishment of functional coherence in the prefrontal corticostriatal thalamocortical circuit (Alexander et al., 1986)—a circuit centred upon the caudate that, as previously suggested (Braun et al., 1997), may be functionally uncoupled during sleep.

The correlational analyses also revealed that post‐awakening re‐establishment of normal alertness is sometimes characterized by functional uncoupling of interregional activity, as suggested by the disappearance at 20 min post‐awakening of the initially significant correlation between activity levels in the RF and prefrontal cortices.

In contrast, some inter‐regional functional relationships (again, indicated by significantly correlated regional rCBF levels) were re‐established by 5 min post‐awakening and remained unchanged at 20 min post‐awakening. This suggests, by their relative stability across the sleep inertia period, that these functional relationships may underlie (and perhaps in some way constitute) consciousness itself. For example, the widespread and stable functional connectivity of the ACC, indicated by numerous and stable correlations between activity levels in this region and widespread other brain regions at both 5 and 20 min post‐awakening, suggests that it may be part of a functional circuit or network of brain regions in which inter‐connectivity subserves conscious wakefulness.

Lastly, as Revonsuo (2001) notes, some caution in the interpretation of functional brain imaging studies of consciousness is warranted since the extent to which current brain imaging techniques—including the PET H215O technique used in the present study—provide measurements at the critical level of physiological organization is unknown.

To summarize, it is suggested that those brain regions (e.g. the prefrontal cortices) for which activation levels and/or connectivity patterns change significantly across the first 20 min of wakefulness—when sleep inertia effects are known to dissipate—most likely mediate alertness. In contrast, those brain regions for which reactivation is maximal upon awakening (e.g. thalamus, caudate, brainstem) or in which interconnectivity patterns remain stable across the first 20 min of wakefulness (e.g. ACC) are more likely to participate in the mediation of consciousness itself.

Further studies of greater scope (e.g. with measures that include multiple levels of physiological organization) and involving contrasts of additional altered states of consciousness are needed to identify the full complement of regions, functional circuits and electrophysiological processes that subserve consciousness and its nuances.

US Department of Defense disclaimer

Human subjects participated in this study after giving their free and informed consent. Investigators adhered to AR 70–25 and USAMRDC Reg 70–50 on the use of volunteers in research. The opinions or assertions contained herein are the private views of the authors and are not to be construed as reflecting the views of the Department of the Army or the Department of Defense.


View Abstract