Behavioral states in humans. States of waking, NREM sleep, and REM sleep have behavioral, polygraphic, and psychological manifestations. In behavior channel, posture shifts (detectable by time‐lapse photography or video) can occur during waking and in concert with phase changes of sleep cycle. [Two different mechanisms account for sleep immobility: disfacilitation (during stages I‐IV of NREM sleep) and inhibition (during REM sleep). In dreams, we imagine that we move but we do not.] Sequence of these stages represented in polygraph channel. Sample tracings of 3 variables used to distinguish state are also shown: electromyogram (EMG), which is highest in waking, intermediate in NREM sleep, and lowest in REM sleep; and electroencephalogram (EEG) and electrooculogram (EOG), which are both activated in waking and REM sleep and inactivated in NREM sleep. Each sample record is ∼20 s. Three lower channels describe other subjective and objective state variables.

Behavioral states in cat. Polygraph records show that distinctive features of awake, NREM sleep, and REM sleep states shown in human records of Fig. 1 are shared by head‐restrained cat. The EMG, cortical EEG (EEG^Ctx), and EOG undergo sequence changes: progressive attentuation of muscle tone; activation, deactivation, and reactivation of cortical tone; and presence, absence, and reemergence of eye movement. Intracerebral leads reveal other features not detectable from surface recordings. The EEG of lateral geniculate body (EEG^LGB) shows distinctively clustered biphasic waves that are synchronous with eye‐movement clusters of REM sleep; these waves are of considerably smaller amplitude in waking. Extracellular microelectrode recording (Cell) shows single‐cell action‐potential profiles of 3 types. Type A, most common, consists of lower discharge rate in NREM sleep than in waking or REM sleep; many neurons in cerebral cortex and cerebellum are type A. In type B progressively higher rates across the 3 states are seen in small proportion of cells, often motoneurons, especially those of pontine brain stem. Type C shows progressive decreases of rate across the 3 states, often with total cessation of discharge in REM sleep; this type of pattern, which is least common of the 3, is seen only in aminergic nuclei of brain stem. Each record is ∼25 s. (W. Silva and J. A. Hobson, unpublished observations.

Circadian rhythms in humans. A: when human subject was isolated in underground bunker, period length of daily activity rhythm changed from 24 h after time cues were removed (day 3). This “free‐running” quality is cardinal characteristic of circadian rhythms and strongly suggests an endogenous origin. When time cues or zeitgebers are restored (day 21), rhythm was resynchronized to 24‐h period. B: 2 circadian rhythms may become dissociated from one another when both are allowed to run free. Sleep‐wakefulness rhythm has longer circadian period than body temperature rhythm. Thus more than 1 circadian clock must exist and be synchronized with one another by zeitgebers.

Circadian rhythms in rats. Circadian activity rhythms are released when normal animals, entrained to 24‐h zeitgebers (A), are deprived of light cues by blinding (B). These early records have been explained by discovery of retinal input to suprachiasmatic nucleus of hypothalamus.

Ultradian sleep cycle of NREM and REM sleep shown in detailed sleep‐stage graphs of 3 human subjects (A) and REM sleep periodograms of 15 human subjects (B). In polysomnograms of A, note typical preponderance of deepest stages (III and IV) of NREM sleep in the first 2 or 3 cycles of night; REM sleep is correspondingly brief (subjects 1 and 2) or even aborted (subject 3). During the last 2 cycles of night, NREM sleep is restricted to lighter stage (II), and REM periods occupy proportionally more of the time with individual episodes often exceeding 60 min (all 3 subjects). Same tendency to increase REM sleep duration is seen in B. In these records, all of which begin at sleep onset, not clock time, note variable latency to onset of first (usually short) REM sleep epoch. Thereafter inter‐REM period length is relatively constant. For both A and B time is in hours.

Biological rhythms and brain stem clocks. Three rhythms interact to determine cyclic order of sleep and waking states. Circadian rhythms are endogensus fluctuations of many bodily functions, including rest and activity, with periods of ∼24 h. As seen in schematic sagittal brain sections, suprachiasmatic nucleus of hypothalamus is key part of this control system that serves to synchronize internal processes with external forces. Ultradian sleep cycle, with its 90‐ to 100‐min period of NREM and REM sleep, is one of the physiological functions whose expression is circadian. It is controlled by reciprocal interaction of cholinergic and aminergic pontine reticular neurons, which oscillate out of phase with one another. This clock determines behavioral state (wake, NREM sleep, and REM sleep) of the organism. Mechanism by which circadian clock sets threshold of sleep‐cycle clock is unknown. Many homeostatic regulatory functions, including respiration, are influenced by circadian rhythm and sleep‐waking cycle. Respiratory oscillator is similar in neuronal design to sleep‐cycle clock but has shorter period (3 s) determined by reciprocal inhibition of expiratory and inspiratory neurons in medulla.

Afferent projections to midbrain reticular formation (MRF) of cat. A: retrogradely labeled neurons in thalamic, subthalamic, and hypothalamic structures (2) after horseradish peroxidase injection into MRF (1). SC, superior colliculus; CG, central gray; RN, red nucleus; SN, substantia nigra; PP, pes pedunculi; LGd and LGv, dorsal and ventral lateral geniculate; OT, optic tract; VB, ventrobasal complex; PUL, pulvinar; CM‐PF, centrum medianum‐parafascicularis complex; RFB, retroflex bundle; FF, forel field; ZI, zona incerta; MTB, mamillothalamic bundle. B: reciprocal connections between MRF (recording) and CM‐PF (stimulation). Antidromic field (f) responses followed by synaptically elicited unit (u) discharges. Arrowhead, stimulus. Antidromic field response could follow 3 shocks at 250/s; its graded character is revealed by progressively decreasing stimulation intensity; unit discharges no longer appeared at lower intensity. C: latency histograms of synaptically (Syn) evoked discharges in MRF neurons to stimulation of CM, ZI, and preoptic area (POA). Coded neurons in histograms antidromically identified to project toward indicated sites. CL, centralis lateralis n. D: convergent synaptic excitation in MRF cell from bulbar (B) reticular formation and POA. Slow‐speed dotgrams (bottom) show dissimilar periods of suppressed firing after initial excitation induced by B and POA. E: graphs depicting percentage of MRF cells with various degrees of synaptic convergence in 2 neuronal populations (which could not be or have been antidromically identified from structures outside the MRF); 0 indicates neurons that have not been synaptically excited, and 1–4 indicate number of stimulated sites that induced synaptic excitation; relative segregation between nonprojection and projection elements in terms of degree of synaptic inputs is highly significant.

Ascending projections of rostral reticular formation in cat. A: drawings of selected parasagittal autoradiograms showing site of injection in cuneiform nucleus (bottom) and labeled projections ascending on ipsilateral side. Approximate laterality of sections in mm from median plane, ac, Anterior commissure; Bac, bed n. of anterior commissure; Bst, bed n. of stria terminalis; Ca, caudate n.; CD, central dorsal n.; CL‐PC, centralis lateralis‐paracentralis complex; dha, dorsal hypothalamic area; En, entopeduncular n.; IC, internal capsule; LD, laterodorsal n.; LP, lateralis posterior n.; lpa, lateral preoptic area; MD, mediodorsal n.; NPC, n. of posterior commissure; R, reticularis thalami n.; SI, substantia innominata; VL‐VA, ventralis lateralis‐ventralis anterior complex; VM, ventralis medialis n. B: diagrammatic chartings comparing patterns of anterograde labeling in 4 closely spaced transverse sections through intralaminar complex of thalamus (1) and mediodorsal n. (2) in cases of injections of tritium‐labeled amino acids into paramedian pontine tegmentum (1) and raphe‐interpeduncular complex (2). Hbl, habenula; NCM, n. centralis medialis. C: diagram representing percentages of antidromically identified MRF neurons from total number of tested elements. Depicted stimulating electrodes inserted into CM‐PF, CL, ZI, POA, paramedian pontine (P), and bulbar (B) reticular formation. Left: examples of CM‐evoked and ZI‐evoked antidromic discharges. S, collision with spontaneously occurring discharge.

Neocortical projections of intralaminar thalamic neurons and their monosynaptic excitation from midbrain reticular core in cat. A: calvarium with last recording thalamic microelectrode (Th) and chronically implanted stimulating electrodes in pericruciate motor cortex (M), parietal association cortex (P), and MRF. EOG and EEG, silver balls for recording eye movements and EEG rhythms; H, electrodes for recording hippocampal rhythms. B: lesion of ipsilateral pontine tegmentum for chronic degeneration of ascending systems coursing through MRF. BC, brachium conjunctivum; BP, brachium pontis; CS, n. raphe centralis superior; IC, inferior colliculus; LC, locus coeruleus; PG, pontine gray; RPO, n. reticularis pontis oralis. C: array of stimulating electrodes into MRF; most lateral electrode track was found in an anterior section. D: location of precruciate cortical stimulating electrodes within deep layers of medial parts of areas 8 and 6. Black dots, whole territory of pericruciate and anterior suprasylvian gyri (various cytoarchitectonic areas are indicated), covered with stimulating electrodes by changing their position from one experiment to another. E: location of 28 CL‐PC neurons found between anterior planes 9 and 10 and studied statistically for spontaneous and evoked activities in waking‐sleep states. Anti and Syn, antidromic and synaptic responses; CeM, n. centralis medialis; Pc, n. paracentralis; Rh, n. rhomboidalis; VPM, n. ventralis posteromedialis. Arrowheads indicate CL. F: physiological identification of CL‐PC neurons. Two different cells, antidromically activated from internal capsule (IC), motor cortex (MC), or parietal cortex (PC), and synaptically driven from MRF. Arrowheads, stimulus artifacts. 2: Only first stimulus of MC 3‐shock train at 250/s is marked; arrow, fractionation of antidromically elicited discharge to last stimulus in MC train. Collision between cortically elicited antidromic spikes and MRF‐evoked synaptic discharge shown in right superimposition (1) and in 10‐sweep sequence (2).

Retrograde labeling of intralaminar thalamic neurons after neocortical injections of horseradish peroxidase (HRP) in rat. A: distribution of labeled cells in both intralaminar CL‐PC n. and LP n. after injection in parietal cortex. B: distribution of labeled cells in both intralaminar CL n. and VL n. after injection in precentral agranular motor area. Thalamic nuclei: AD, anterodorsal; AM, anteromedial; AV, anteroventral; Ce, central medial; MV, medioventral; Pom, medial division of posterior complex; Pva, anterior paraventricular; Pvp, posterior paraventricular; Sm, submedial; VMb, basal part of ventromedial.

Neocortical layer I projection from nucleus VM in cat. A: drawings of anterogradely and retrogradely labeled structures after ipsilateral HRP injection into VM. Dorsal view of cortex. Stippled area, cortical extension of intense granular reaction in layer I. In sagittal section (at ∼2.3 mm from midline), HRP reaction product indicative of anterogradely transported HRP in layer I marked by heavy line; heavy dashed line, liminal grain density. Dots, presence of retrogradely labeled cells in layer VI; density of dots was drawn in close proportion to density of labeled neurons, but number of dots is drastically lower than number of neurons. Light dashed lines, border between gray and white matter. Location of cytoarchitectonic areas indicated by numbers. F, fornix; G. Splen, splenial gyrus; IC, inferior colliculus; OB, olfactory bulb; S. cru, cruciate sulcus; V3, third ventricle. B: retrogradely labeled neurons in layer VI of area 6 (∼2.3 mm from midline); cru, cruciate sulcus; pre and post, precruciate and postcruciate gyri. Neurons taken at greater magnification are from crown of precruciate gyrus. Calibrations in mm. C: 1 and 2 show pattern of VM‐ and VL‐evoked cortical response in medial part of area 6 to single‐ and 10‐Hz shocks recorded at surface and depth of 1.0 mm. Note recruiting (initially surface‐negative, depth‐positive) and augmenting (initially surface‐positive, depth‐negative) responses to stimulation of VM and VL, respectively. 3: Suppression of VM‐evoked surface‐negative wave during cortex superfusion with Mn²⁺ to reversibly block synaptic transmission. Control waves in deep layers unaffected (not depicted; see details and Fig. 9 in ref. 251).

Distribution of brain stem monoamine neurons and ascending serotonergic pathways. A: drawings of transverse hemisections through brain stem of cat to illustrate distribution of monoamine‐containing neurons. Open circle, serotonergic cell bodies; filled circles, catecholaminergic cell bodies. B: representation of major organizational features of ascending serotonergic systems of rat brain as revealed by light‐microscope radioautography after intraventricular administration of tritiated serotonin. BO, bulbus olfactorius; CP, cerebral peduncle; CS, n. centralis superior; CT, corticospinal tract; DR, dorsal raphe n.; F, columna fornicis; FR, fasciculus retroflexus; GP, globus pallidus; GPO, griseum pontis; HI, hippocampus; IC, inferior colliculus; IP, interpeduncular n.; L, n. linearis rostralis; LL, n. of lateral lemniscus; MFB, medial forebrain bundle; MH, n. medialis habenulae; ML, medial lemniscus; MLF, medial longitudinal fasciculus; nVII, n. of facial nerve; PBC, n. parabrachialis; PVS, periventricular system; PY, pyramidal tract; RB, restiform body; RMA, n. raphe magnus; RP, n. raphe pallidus; SC, n. subcoeruleus; SL, n. septi lateralis; SM (caudal), stria medullaris thalami; SM (rostral), n. septi medialis; SNc, substantia nigra, pars compacta; SNr, substantia nigra, pars reticulata; TTS, transtegmental system; VTA, ventral tegmental area.

Early and late electrographic signs induced by kainic acid (KA) injection into MRF of chronically implanted, nonanesthetized cat and histological aspect of kainic‐induced lesion. A: the 3 ink‐written traces depict (before and after kainic injection) EEG waves, ocular movements (EOG), and EMG of neck muscles. Note normal fluctuations of EEG desynchronization‐synchronization periods before injection (top traces) and continuous EEG desynchronization (associated with highly aroused behavior) after injection. Picture (5 h after kainic injection) began ∼30 s after onset of injection and lasted for 24 h. This period corresponds to early excitation induced by KA. B: 2–4 days after kainic injection, corresponding to period of neuronal destruction, long‐lasting (∼30 s) EEG desynchronization elicited by an arousing (auditory) stimulus in control period before injection (top traces) is replaced by phasic EEG desynchronization, despite similar peripheral signs elicited by arousing stimulus. C: histological aspect of lesion produced by unilateral (left) injection (2.5 μg) of KA in MRF. Frontal plane 3.2. Animal was chronically implanted on both sides with guide cannulas into superior colliculi (SC) (vertical arrow in 1) and sacrificed 10 days after kainic injection. Territory of total neuronal loss on left side delimitated by dots (1). Limits of lesioned area may roughly indicate limits of initially excited area. Details at higher magnifications of left depicted in 2. Oblique arrow, blood vessel (permits localization of areas photographed in 2). Arrowheads, normal ganglion‐type cells of mesencephalic trigeminal nucleus, surviving within completely depopulated region (for details on selective and absolute resistance of these ganglion‐type cells, see refs. 154,177). Lesion affected almost entire MRF, leaving intact only a limited area dorsolateral to RN. Lesion encroached on lateral part of CG and lateral part of deep layers of SC. With exception of ganglion‐type trigeminal cells, there was total loss of neurons within lesioned area. Sections stained with unsuppressed Nauta method showed integrity of axons in fields with complete neuronal depopulation. FTC, central tegmental field; III, oculomotor nucleus.

Firing rates and patterns of MRF neurons in cat. A: discharge rate as function of dorsoventral localization. 1: Location at frontal plane 2.5 of neurons whose firing rates were roughly estimated during recording and divided into 3 classes: silent, <3/s, and >3/s. AQ, aqueduct; 3, oculomotor n.; L, raphe linearis. 2: Dorsoventral position of 44 MRF neurons whose firing rates during quiet waking were measured in detail; Spearman rank correlation between high discharge rate and ventral location (r_a = 0.42) significant at P < 0.005; arrow, limit between deep layers of SC and MRF. B: discharge patterns of rostrally projecting MRF cell, antidromically identified from zona incerta and synaptically driven from preoptic area and paramedian pons. Note sustained regular firing regardless of movements in waking (W+) or quiet wakefulness (W‐). ISIH, interspike interval histogram; N, number of intervals; X, mean interval; M, modal interval; C, variation coefficient; E, proportion of intervals in excess of depicted time range. Note symmetric shape and interval density near mode. Bottom: autocorrelations (ACFs) of same neuron during states of W‐, slow‐wave sleep (S), and desynchronized sleep (D). Note, during W‐, rhythmic firing with peaks at multiples of mode, and flat contours in S and D states.

Midbrain reticular formation neurons increase discharge rates in advance of behavioral and EEG signs of activated states in cat. A: electrographic criteria of transitional state (SW) from slow‐wave sleep (S) to wakefulness (W). Abrupt (1) and progressive transition, with an intermediate SW period (2). B: percent cumulative histograms (1‐s bins) of 2 MRF neurons (neuron 2 antidromically identified from CM‐PF). Abscissas, real time of recording; arrows and vertical lines, earliest signs of reduced amplitude and/or increased frequency of EEG waves (as in A, pt. 2, arrow indicates time 0 of SW period). Inflection points are seen to occur 10–22 s in advance of any change in EEG; overt signs of wakefulness (eye movements and increased muscular tone) appeared several seconds after arrows (as in A, pt. 2). C: increase in firing rate of MRF neurons of cat before end of S epochs developing into W. Left: percent cumulative histogram of neuron belonging to sample analyzed in graph depicted on right; arrows, first change in fully synchronized EEG waves. Right: 25 cells whose global mean rate in S was at least 4/s were analyzed during last minute of S in 42 epochs leading to SW or directly to W. Mann‐Whitney test was used to compare reference rate during first 30 s for all cells with their respective rates in the last 6 5‐s bins. Note significantly increased rates in the 3 5‐s bins before end of S (arrow) compared with discharge rate in the first 30 s.

Midbrain reticular formation neurons decrease discharge rates in advance of first electro‐graphic signs during transition from W to S. A: electrographic criteria of transitional state from W to S (WS). Graph shows median of firing rates in sample of 52 MRF cells during W, WS, and S; note that major and significant decrease in firing rate occurs from W to WS. B: neuron antidromically identified from CM‐PF. Top 2 traces, original spikes and EEG waves simultaneously displayed on oscilloscope. Note decreased firing rate, leading to neuronal silence, before first EEG spindle sequence (between arrows) during transition from W to S. Bottom traces, same activities during repeated EEG desynchronization‐synchronization transitions (polygraphic recordings). C: 12 transitions of unit firing with respect to start of EEG spindle (time 0). Arrow, level of discharge (median rate) during W. Asterisks, bins with significant (<0.05) decrease in firing rate compared with median rate during W. Significantly decreased firing rate occurred 1 s before spindle onset.

Cortical and thalamic neuronal activity during waking‐sleep states. A: unit discharges from cell in motor cortex of monkey during W and S. Surface EEG rhythms from motor and visual cortices shown below discharges. Note that cell remained active during S. B: patterns of firing of lateral geniculate cell in cat during W and S. Upper beam swept from below upward; sweeps triggered by spikes. Positive deflections upward for continuous beam, to left for swept beam. Different time calibrations for upper and lower beam. Note bursting discharges during S. C: discharge frequency in antidromically identified pyramidal tract (PT) neurons of monkey and in neurons that were not antidromically identified (non‐PT) during W, S, and D. Ordinates, spikes/s in PT and non‐PT cells. The PT and non‐PT cells are quite different with respect to both total amount of activity and changes in amount of activity as a function of sleep and waking.

Discharge patterns of thalamocortical and corticothalamic neurons during waking‐sleep states in cat. A and B: intralaminar CL‐PC thalamic neurons antidromically identified from MC (neuron B was also synaptically activated from midbrain reticular core). Note high‐frequency spike clusters in S and their replacement by sustained discharge in both behavioral states of W and D. C: cortical cell recorded in area 5 and backfired from the CM thalamic n. (see inset at left with antidromic identification). Similarly sustained discharges in W and D; decreased firing rate in S. Bottom: polygraph traces depict unit spikes, focal EEG waves simultaneously recorded by microelectrode in area 5, EEG from depth of visual cortex, EMG of neck muscles, and EOG. Note tonically increased firing rate preceding onset of D (upward arrow) and during D, similar to discharge pattern in W (awakening marked by downward arrow). Two parts are separated by nondepicted period of 180 s. Tonic discharge of corticofugal neurons throughout D is very different from burst discharges of cortical interneurons related to REM epochs (see Fig. 20C).

Fast‐conducting and slow‐conducting PT neurons on arousal and subsequent steady waking in monkey and cat. A: histogram of antidromic response latencies of 67 precentral PT neurons after peduncular stimulation in monkey. Hatched columns, units in which spontaneous discharges stopped on arousal from synchronized sleep; white columns, neurons whose spontaneous firing increased on arousal. Inset: PT cells (a, b) that are represented in B. Arrowhead, stimulus; oblique arrow, lack of antidromic spike of cell b due to collision with spontaneous (S) discharge. B: 4 ink‐written traces depict a and b units, EEG waves, and eye movements. Note that fast‐conducting a neuron diminished firing rate on arousal but, afterward, during steady waking, increased its firing rate over value seen in synchronized sleep. Slow‐conducting b cell increased firing rate from beginning of arousal. C: intracellular recording of fast PT cell in midpontine pretrigeminal cat. Traces, from top to bottom: stimulation of MRF at 100/s, EEG from MC, hippocampal activity, zero potential level (broken line), cell activity (spike peaks off), spike rate/s, and cell activity at higher gain (spike peaks off). Bottom: antidromic responses to peduncular stimulation and spontaneous action potentials.

Discharge patterns of cortical interneurons during waking‐sleep states in cat. A: discharges during W, S, and D of putative short‐axon cell recorded from parietal association area 5, synaptically driven with high‐frequency spike barrage from LP thalamic n. (similar to interneuron depicted in C). B: similar type of cell in area 5. Polygraph traces, from top to bottom: unit discharges, EEG focal waves recorded by microelectrode, surface EEG waves, EMG, and ocular movements. Note diminution of spontaneous firing during arousal elicited by auditory stimulation (between arrows) and progressively increased occurrence of spike bursts with transition from W to S. C: another interneuron in area 5, driven by stimulation of LP thalamic n. with spike barrage at ∼500/s occurring in relation with onset of slow positive focal wave. Note stereotyped spontaneous spike bursts in all (W, S, and D) states. Note in polygraph traces (same activities as in B) that, during D (beginning at arrow), interneuronal discharge bursts are closely related to REM episodes and are dissimilar to sustained firing of corticofugal neurons during D, as shown in Fig. 18C.

Antidromic responsiveness of thalamocortical cells during waking‐sleep states in cat. A: inhibition during EEG spindles of antidromic responses in thalamocortical VL neurons. Two simultaneously recorded cells. Neuron b was superimposed on negative hump of neuron a; note that neuron b was antidromically activated in absence of discharge of neuron a (5). Figures on EEG record (1–6) indicate application of testing shock trains, corresponding to those in traces depicting evoked discharges. Abolition of antidromic responses occurs during EEG spindles (3 and 6). Evoked discharge of cell a disappeared 500 ms prior to EEG spindle (5). B, top: antidromic responses to area 5 stimulation in typical CL cell of this sample during W, S, and D. Arrowheads, stimuli. Bottom: median (arrowheads), mean (columns), and standard deviation for percentage of antidromic responsiveness (R) in sample of 19 CL‐PC intralaminar thalamic neurons backfired from pericruciate (areas 6 and 4) or anterior suprasylvian (area 5) gyri during W and S. Five of these neurons were also recorded during D. Asterisk, collision with spontaneous discharge.

Antidromic responsiveness of corticofugal cells during waking‐sleep states in cat and monkey. A: facilitation of antidromic invasion of PT cell of cat during EEG desynchronization. Full responsiveness to a 4‐shock train during control period of spontaneous EEG desynchronization (1), depressed responsiveness during progressively developing EEG synchronization (2 and 3), and recovery of antidromic responsiveness during EEG desynchronization elicited by brief conditioning pulse train to MRF (4). B, top: corticothalamic cell in area 5 of cat, antidromically activated from n. CM. Antidromic responsiveness was investigated with a 4‐shock train. Arrowheads, stimuli. Bottom: percentage responsiveness (R) to 1st, 2nd, and 4th shock during W, S, and D. Below each state is mean rate of spontaneous firing (during another waking‐sleep cycle). C: pattern of antidromic invasion of pre‐central PT neurons during behavioral S (left) and arousal (right) in monkey. Dotted line and arrow indicate arousal. Fast‐conducting neuron has 0.5‐ms antidromic response latency to peduncular stimulation. 1 and 2, Responses to trains of 5 shocks (350/s) and 3 shocks (110/s), respectively, during 2 passages from sleep to arousal. Third trace: EEG desynchronization. Bottom trace: eye movements. Note spike fragmentation of first antidromic spike (arrowheads) during sleep and full recovery on arousal as well as diminution of spike fragmentation of successive responses.

Primary synaptic excitation in thalamus and neocortex during waking‐sleep states in cat. Evoked potential studies (A‐C) and extracellular unit recordings (D‐F). Filled circles, testing shocks. A: simultaneous recording of field potentials evoked in lateral geniculate n. (LG) and at surface of visual cortex (VC) by optic tract stimulations (monopolar recordings). Response of LG consists of presynaptic [tract (t)] positive component and monosynaptically relayed (r) negative component. Different components of VC response numbered 1–5. Note, during 300/s stimulation of MRF, enhancement of monosynaptic (r) component in LG without alteration in presynaptic (t) deflection; note also increased amplitude of VC response. B: surface VC potentials evoked by stimulation of white matter beneath VC (1) or to deep layers in VC (2). Note that, with white matter stimulation, postsynaptic components of VC response are enhanced without changes in presynaptic component. C: simultaneous recording of field potentials evoked in the VL thalamic n. and MC by stimulation of cerebellothalamic pathway during W, S, and D. Note in S, compared with both W and D, obliteration of monosynaptic thalamic wave (r) and marked reduction of cortical response, without alteration of component t that reflects activity in afferent fibers to VL n. (in this case, bipolar recording in VL). D: MRF‐evoked synaptic discharges in intralaminar CL thalamic neuron antidromically identified from parietal association area 5. Note decreased latency, increased probability of discharges in early bins, and shorter duration of spike bursts during W compared with S. E and F: unit recordings in SI cortex (SC) and white matter stimulation in animals with complete lesion of VB thalamic complex. The MRF shock train preceded testing shocks by 3–5 ms (not depicted). Note MRF‐induced facilitation of evoked discharges. F: MRF potentiation is seen by decrease in latency of discharges to first stimulus and increased discharge probability to second stimulus.

Effects of natural arousal and midbrain reticular stimulation on inhibitory processes in thalamus and neocortex of cat and monkey. A: periods of suppressed firing (a, b, and c) and postinhibitory rebound excitation of neuron recorded in rostrodorsal part of LP thalamic n. after single‐shock stimulation (arrowhead) of cortical area 5 in cat. Preceding shock train to MRF reduced duration of periods a and b and abolished silent period c. B: patterns of VL‐evoked events in precentral neuron of monkey during S and W. Note that first inhibitory period evoked by VL stimulus (dots) persists in W, but subsequent rhythmic inhibition‐rebound sequences seen during S are abolished in W on background of increased spontaneous discharge. C: method of testing recurrent inhibition acting on antidromic discharges elicited in cat PT neurons by pes peduncular (PP) stimulation. Midbrain lesion included medial lemniscus to avoid stimulation of afferent fibers. Conditioning volley (C) was delivered at 13 V [minimal voltage required to elicit inhibitory effects on testing (T) response induced by shock at 5 V, which is minimal voltage required to evoke 100% antidromic invasion]. At paired C‐T stimulation, complete inhibition of T response or spike fragmentation (arrow). Graph depicts much longer inhibition with 3 PP shocks than with single shock. With both conditioning procedures (1 PP and 3 PP), recovery was slower during S than during W. D: inhibition of synaptic discharges evoked by stimulation of posterior part of VL in precentral PT neuron of monkey. Left: field positive (inhibitory) wave evoked by first VL stimulus and facilitation (during W) of evoked discharges by second stimulus at 75‐ms interval toward end of inhibition. Right: percentage responsiveness of discharge evoked by first stimulus (time 0) and by second stimulus at 3 time intervals (15 ms, 27 ms, and 75 ms) during W and S.

Secondary excitation and incremental responses during waking‐sleep states in cat. A: focal slow waves recorded at depth of 0.5 mm in cortical area 5 after stimulation of rostrodorsal part of LP thalamic n. during W, S, and D (50 averaged sweeps). Note during S selective enhancement of second (b) depth‐negative component. B: responses of cortical SI neuron to VB thalamic stimulation during S and W. Note during W increased probability of VB‐evoked early discharge and suppression of late repetitive discharges. C: simultaneous recording of field potentials (1: 50 averaged sweeps) and unit discharges (2) at depth of 1 mm in suprasylvian area 5, evoked by 2 0.1‐s‐delayed stimuli to LP thalamic n.; stimuli indicated by dots in 1 and by vertical bars in 2. Secondary depth‐negative wave b associated with repetitive discharges is selectively enhanced at second stimulus. D: poststimulus histograms of augmenting responses in sample of 7 cortical SI neurons, elicited by white matter (WM) stimuli at frequency of 10/s in VB‐lesioned preparation. S, stimulus number; R, responsiveness (total number of discharge to 100 shocks). Note that augmentation elicited by S‐2 (the second 0.1‐sdelayed shock) consists of increased secondary excitation (10–15 ms, arrow) simultaneous to decreased probability of primary synaptic excitation. Effect of a conditioning stimulation of MRF consists of an increased probability of primary excitation and a change in pattern of augmenting response to second stimulus into a primary one. E: augmenting field potentials at depth of 0.7 mm in area 5 to 10/s stimulation of LP thalamic nucleus during W, S, and D. Arrowheads in A, B, C, and E indicate stimuli.

Cortically evoked rhythmic hyperpolarization‐rebound sequences. Intracellular recording of VL thalamic relay cell, antidromically invaded from precruciate MC and monosynaptically driven from brachium conjunctivum (BC). Resting membrane potential: −55 mV. 1: Typical response sequence. 2 to 4: 40 Averaged sweeps. Note in 3 and 4, analyzed at same speed, powerful rhythmic activity within frequency of spindle waves (∼7.5/s) evoked by MC, contrasting with lack of such activity after BC stimulation. Arrowheads, MC stimulation; filled circle, BC stimulation.

Effect of membrane potential on excitability of VL thalamic relay neuron in cat; intracellular recording. A and B: antidromic and orthodromic activation by MC and BC stimulation, respectively. C: direct stimulation of cell by current pulse. At resting potential (0 nA) current pulse was subthreshold for spike initiation. During injection of 1 nA continuous hyperpolarizing current (2), same pulse remained subthreshold but break was followed by slow decaying response on voltage trace. Finally, under 2 nA hyperpolarizing current (3), pulse triggered “slow spike” and burst of action potentials. D: postanodal exaltation phenomenon obtained by passing hyperpolarizing current pulses of increasing intensities at resting potential. Note stereotyped burst response in D, pt. 3 compared with C, pt. 3. E: traces, same phenomenon as in D but at higher gain and slower speed. Current intensities similar to those in D. E, pt. 2: slow spike can be seen in isolation. Upper calibration bars apply to A and B.

Effects of rostral reticular stimulation on intracellulary recorded inhibitory processes in thalamic neurons. A: excitatory postsynaptic potential (EPSP) and inhibitory PSP (IPSP) sequences in VL neuron of cat during 7/s midline thalamic stimulation (1) and marked attenuation of hyperpolarizing potentials and increase in cell discharges during simultaneous low‐frequency thalamic stimulation and high‐frequency brain stem reticular stimulation (2). Upper traces of 1 and 2, evoked responses at motor cortical surface. B: VL relay cell in cat. 1: MC stimulation (arrowheads) evokes antidromic discharge followed by rhythmic inhibitory‐rebound sequences. 2: Conditioning 320/s shock train to MRF leaves intact the cortically elicited early IPSP but abolishes late phase of hyperpolarization and following rebound‐inhibition sequence. 3: Effect of MRF stimulation alone. C: slow (∼0.1 Hz) thalamic rhythm of hyperpolarizing episodes and its suppression during MRF stimulation. 1: Ink‐written intracellular recordings of VL relay cell. Note oscillations within frequency range of spindles (∼7 Hz), consisting of phasic hyperpolarizations followed by rebounds, appearing during slow rhythm (−0.12 Hz) of hyperpolarizing episodes, each lasting 2.3 s. Postinhibitory rebounds consist of slow spike (see Fig. 27) superimposed by fast repetitive action potentials. 2: Suppression of slow rhythm of thalamic hyperpolarizations during MRF stimulation (brief shock train every second); note increased background firing that outlasted MRF stimulation period; first episode of hyperpolarization after MRF stimulation lasted only 1.3 s, compared with 2.3‐s duration of these episodes before MRF stimulation.

Activities of RE neurons during waking‐sleep states in cat. Two neurons recorded in rostral pole of nucleus reticularis thalami, monosynaptically driven from precruciate gyrus with high‐frequency spike barrages, as shown for neuron A in top poststimulus histogram. T, number of trials; X, mean latency (ms); M, latency mode (ms); C, coefficient of variation; R, responsiveness (total number of spikes to 100 stimuli during depicted time). Traces in A: unit activity, surface cortical EEG waves, focal EEG waves recorded from electrode pair that induced synaptic activation of the neuron, EMG, and time (1 s). Note spike bursts closely related to EEG spindles and tonically increased firing rate during arousal and sustained waking. B: peristimulus histograms depict activities evoked from precruciate cortex during S and W. Left histograms (2‐ms bins) show details of early excitation; right histograms (25‐ms bins) show late inhibition‐rebound phases. Note in left histograms that probability of early evoked discharges (first 2‐ms bin) doubles in W compared with S. In S evoked burst mostly extends within latency range of secondary excitation (∼15–40 ms). Also note (right histograms) 2 evoked rebound sequences in S, whereas such events are lacking in W (see similar phenomena in Figs. 24A and 28B for thalamic relay cells).

Facilitation of monkey's parietal visual neurons by attentive fixation. A: comparison of responses of parietal light‐sensitive neurons to visual stimuli in no‐task and task modes. Absence of responses during no‐task mode compared with strong responses in task mode is obvious. Summing histograms: standard error of mean calculated for each bin of the histograms; value shown by dotted line. Corresponding bin pairs within the histograms tested for significant differences (t test); bin pairs marked with diamonds differed at 5% level of significance. Overall response in no‐task and task states compared in following way. Rate of impulse discharge in prestimulus period was subtracted from that in poststimulus period for each trial, and populations of remainders were tested for significant differences (P ≪K 0.05 required) and used to form facilitation ratio for each neuron. Ratios for neurons with significant differences plotted in B. Facilitation ratio is that between net increment in response evoked by light stimulus in state of interested fixation over that evoked by physically identical and retinotopically similar stimulus delivered in no‐task or intertrial states. Fifty‐one neurons showed ratios of ≥1.0, and of these, difference was significant at 5% level (t test) for 38; values plotted in histogram. Ratio was fractional for 4 neurons indicated at left: for them, response was significantly greater in no‐task state than during interested fixation.

Facilitatory action of acetylcholine (ACh) in cerebral cortex. A: intracellular record from neuron in motor area of cat shows delayed depolarizing effect and prolonged firing evoked by iontophoretic application of ACh (140 nA). B: brief and instantly reversible depolarization and strong firing of same neuron caused by short intracellular current injection (monitored on lower trace). C: same neuron depressed after treatment with dinitrophenol; no longer fired in response to application of ACh (as in A); however, during continued ACh application, identical intracellular current injection (cf. B) now induced particularly powerful and prolonged discharge. The ACh thus greatly facilitates and prolongs any depolarizing input received by same cell. D and E: magnitude and time course of changes in potential and resistance induced by ACh. Open circles, resting potential: note slow and prolonged depolarizing effect; open triangles, resting resistance: note marked increase in resistance synchronous with depolarization; closed symbols, corresponding data recorded during IPSPs: they show relatively little change, except some possible reduction of inhibitory effect. F: summary of probable mechanism of action of ACh. In contrast to situation at neuromuscular junction (and other sites of nicotinic cholinergic action) where depolarization is mediated by enhancement of Na⁺ permeability, muscarinic ACh action in cerebral cortex tends to reduce membrane K⁺ permeability, causing an increase in overall resistance and facilitating depolarizing effect of other depolarizing inputs, such as EPSPs, which probably act by increasing Na⁺ permeability.

Eye movements (EM), EEG, systolic blood pressure (SBP), respiration (resp), pulse, and body movements (BM) in a 100‐min sample of uninterrupted sleep over successive minutes of typical sleep cycle. Entire interval from minute 242 to 273 is considered to be REM period, even though eye movements (heavy bars) are not continuous.

PGO waves and relation to REM sleep and eye‐movement direction. A: NREM‐REM transition showing PGO waves in LGB (types I and II). During transition periods from NREM to REM sleep, biphasic (PGO) waves in LGB first appear as large single events (type I waves). Waves become clustered and of diminished and decrementing amplitude (type II waves) as signs of REM sleep become more prominent: atonia (EMG), desynchronization of cortical EEG (Cx), hippocampal‐θ (HIP), and REMs (EOG). B: side‐to‐side alternation of primary waves. Once REM period is well established, PGO wave amplitude primacy alternates from one geniculate to the other according to lateral direction of eye movements. When there is rightward movement of the eyes (EOG‐R), corresponding PGO wave cluster is larger in right LGB (dots) than in left. Conversely when there is leftward movement of the eyes (EOG‐L) waves are larger in left LGB (dots).

Behavioral state and neuronal discharge rate. Mean rates of brain cell populations shown as function of behavioral state. A: D‐on cells. Most cells of the brain have higher rates in D and W than in S. Cbm, cerebellar Purkinje cells; VN, vestibular n; Thal, thalamic n.; PRF, pontine reticular formation; Ctx, cerebral cortex; Hypo, hypothalamus. B: D‐off cells. Minority of cells that decrease rate in D are all found in or near aminergic zones of pontine brain stem. Ret N, reticular n. subpopulation; RN, raphe n.; PN, peribrachial neurons.

Relation of pontine burst cell firing, PGO wave amplitude, and eye movement lateral direction. A: burst cell and waves. Recording of cell in peribrachial region of pons that fires 4 clusters of spikes, each of which precedes PGO waves in LGB by 10–20 ms. Geniculate ipsilateral to cell (LGB₁) shows markedly greater wave amplitude than the contralateral geniculate (LGBc) in initial pair of series. Correlation between pontine burst cell activity and LGB wave amplitude is 0.99. B: efferent copy. Three‐way correlation between eye movement direction (EOG), geniculate wave amplitude (LGB), and pontine burst cell activity shown on drawings of ventral brain surface (above) and exemplified in oscilloscope tracings (below). When eye movement is toward same side as cell, wave amplitude is greater in LGB₁ and vice versa.

Pontine localization of REM sleep generator: lesion evidence. A: prepontine transection of the brain (with forebrain ablation) is followed by alternation of state resembling waking and REM sleep. Note suppression of muscle tone (EMG) and clustered eye movements (EOG). The EEG of brain stem shows low‐amplitude PGO waves occurring with eye‐movement clusters. Data show that REM sleep clock and trigger neurons are retropontine. B: in isolated pons preparation, C1 spinal cord transection is added to prepontine brain stem transection. Atonia is eliminated but REMs and pontine PGO waves appear with 30‐min periodicity characteristic of REM sleep in cat. Data show that REM sleep clock and trigger must reside in lower brain stem.

Pontine mechanisms of REM sleep generation: D‐on cells and executive actions. A: identification. Reticular neurons projecting to spinal cord can be identified by antidromic activation from axons at lumbar spinal cord levels via chronically implanted stimulating electrodes while recording somatic action potentials in brain stem of head‐restrained but unanesthetized cats with moveable microelectrode. At paramedian pontine site shown in 1, cell was recorded that showed a fixed‐latency response (2), fast following (3), and collision (4) with orthodromic spikes. Cell was typical in that it showed lowest firing rate in waking (5), a slightly greater rate in S (6), and intense bursting prior to REM bursts of D (7). B: selectivity. Tendency to concentrate firing in D quantified as ratio of mean rate in that state to rate in W or S. As shown in 1‐min episodes of spontaneous firing for each of these cells in the 3 sites, selectivity was highest in pontine reticular formation (1), next highest in pontomesencephalic reticular formation (2), and lowest in such precerebellar nuclei as tegmental reticular nucleus (TRC) (3). C: phasic latency. Tendency to fire prior to eye movements quantified by plotting sequential and cumulative histograms as shown and averaging across populations. Longest phase leads (up to 300 ms) and greatest increases (up to 100% of firing) were seen in pontine reticular formation. D: tonic latency. Pontine reticular neurons showed statistically significant increases in firing rate as early as 3.5 min prior to D epochs. These tonic rate changes were longer in gigantocellular tegmental field (FTG) than any other neuronal group in brain stem or forebrain. Note that rate increase occurs in episodic increments suggesting that phasic activation waves may spatially and temporally summate as recruitment spreads within the reticular pool. E: periodicity. Spontaneous discharge of single FTG neuron in 10 h of continuous recording. Discharge peaks occur at ∼30‐min intervals coincident with each REM period. F: proportion of firings by 4 single FTG neurons averaged in successive intervals of repeated sleep cycles. Each cycle was time normalized by establishing duration of D and dividing it into 5 equal parts; the 10 min before (D‐10) and after (D+10) each D period were also examined. For each cycle, percentage of total number of firings was determined for each of 25 bins.

Pontine mechanisms of REM sleep generation: D‐off cells and permissive action. In contrast to discharge pattern and activity profile on D‐on cells found in pontine reticular formation is behavior of neurons in aminergic nuclei such as LC cells. A: anatomical location. Histological reconstruction (drawing) and computer plot (inset) of microelectrode penetration and recording site, in n. LC (CAE), of D‐off cell. FTG, pontine reticular formation; 7G, genu, of facial nerve; SA, striae acousticae; VSL, lateral vestibular nucleus. B: spike trains. The LC cell recorded at site shown in A fired regularly in W, intermittently in S, and arrested discharge in D. Cell resumed firing, with no change in spike height, during subsequent cycle. C: dotgrams. One‐min periods of spontaneous activity in each of the 3 states shown by triggering Z‐axis of oscilloscope with action potentials shown in B. D: cell activity curves. As in D‐on cells, change in rate across states is progressive and continuous in D‐off cells but opposite in direction. Resumption of firing prior to end of D‐sleep epochs shown in insets of both graphs. SWS, slow‐wave sleep; PS, paradoxical sleep.

A: reciprocal activity of LC and giant reticular neurons during sleep cycle. Reciprocal discharge by cells in CAE 559,560,561,562 and FTG 563,564,565,566,567. Outline tracing, sagittal section of cat pontine brain at 2.5 mm lateral to midline, showing path of an exploring microelectrode passing through cerebellum into dorsal brain stem. Inset, location of the 7 successive recording sites in penetration; circle diameter is 5 mm. Cumulative discharge histograms of cells recorded at 7 sites shown in inset during transition period beginning 2 min after D onset (vertical line) and ending 1 min thereafter. Each activity curve shows cumulative percentage of discharge for as much of the epoch as was free of arousal. Units recorded in LC and subcoeruleus decrease discharge rate (negatively inflected histograms), whereas those in FTG increase discharge rate (positively inflected histograms) with approach and onset of D. B: pooled geometric mean rate for population of 21 LC and 34 FTG neurons. Abscissa is W, S, and D. C: selectivity ratios of geometric mean rate for pooled population data shown in B. D: pooled cumulative histograms of discharge rates in subsets of the 2 neuronal groups during transition period from S to D. Ordinate, percentage of discharge; abscissa, time in minutes, where t_o is onset of D and 2 min before and 1 min after are shown. Overall mirroring of the 2 curves suggests that activity of the population may be modulated by shared, perhaps mutual, process.

Reciprocal interaction model of REM sleep generation. A: anatomical maps. Locus of D‐on cells (hatched area) in reticular formation nuclei and D‐off cells (dashed line) in pontine aminergic nuclei is shown on sagittal section of brain stem. Four sites with neurons relatively indifferent to state are shown on whole brain schematic: cerebral cortex (Ctx), cerebellar Purkinje cells (Cbm), other reticular nuclei (Ret), and PG. CAE, locus coeruleus; FTP, L, posterior and lateral tegmental fields; Pbl, parabrachial zone; RN, raphe n. B: selectivity gradient. The D/W selectivity ratios, computed and pooled as described in Fig. 39, are plotted for some of nuclei shown in A. Continuous gradient from most positively selective cells in FTG to most negatively selective cells in RN. Converse gradient would describe propensity to fire in waking under same recording conditions. C: physiological model. D‐on cells are assumed to be cholinergic/excitatory (E); D‐off cells are assumed to be aminergic/inhibitory (I). Each group is assumed to have both feedforward and feedback interconnections giving structural and synaptic model of reciprocal interaction. Model can be described mathematically and tested pharmacologically.

Reciprocal interaction model: physiological and mathematical aspects. A: structural model of interaction between FTG and LC cell populations. Plus sign implies excitatory influences; minus sign implies inhibitory influences; a‐d correspond to constants associated with strength of connection and included in text equations. B: theoretical curve derived from model that best fits FTG unit in Fig. 37E. In this fit a = 0.3029, c = 0.1514, and the initial conditions (amplitude unsealed) were x(0) = 1 and y(0) = 4.5. C: histogram is average discharge level of another FTG unit over 12 sleep‐waking cycles, each normalized to constant duration. Cycle begins with end of D. Arrow, bin with most probable time of D onset. Solid curve, FTG fit; dotted curve, LC fit derived from model with values a 0.5490, c = 0.2745, x(0) = −1, and y(0) = 3.0 for the 2 populations. Dot, equilibrium value for the 2 populations. D: geometric mean values of discharge activity of 10 LC cells before (S), during (D), and after (W) a D episode. Each time epoch is equal to one‐quarter of D period. Note that discharge‐rate increase begins in last quarter of D, at same time that FTG activity is falling as predicted by model.

Modulation of pontine neuronal excitability over sleep‐waking cycle. The finding that pontine reticular neurons fire in association with movement during waking in unrestrained cats is shown to be consequence of strong excitatory drive from other central motoneurons and afferent input from periphery. That different mechanisms could lead to same net output in W and D may be appreciated by comparing synaptic models (A) with activity pattern (B). In W powerful afferent excitation of FTG (++++) overcome tonic inhibitory restraint coming from aminergic neurons (—) and summates with recurrent collateral excitation (++) to produce clustered firing. In absence of afferent drive, cell is silent in W. In D less powerful afferent input (++) is capable of driving cells because level of inhibitory restraint has decreased (−). For same reason collateral excitation is more capable of sustaining firing within reticular pool. Result is repeated clustered firing that drives eye movements and muscle twitches. Because of tonic inhibition, motoneurons of large skeletal muscles do not respond; hence there is no major movement and no feedback from movement to central motor pattern generators in D. Because of phasic presynaptic inhibition of la cutaneous afferents there is also no change from peripheral sensory stimuli to drive system as is present in W. In slow‐wave sleep the system is in condition intermediate between the 2 extremes just discussed: disinhibition increases progressively and facilitation is markedly reduced so that cell firing is sparse.

Pharmacology of REM sleep generation: comparison of results with predictions of reciprocal interaction model. REM sleep is favored (black postsynaptic neurons) by either enhancement of cholinergic transmission or by blockade of noradrenergic and serotonergic neurotransmission. Top: REM sleep increases are seen when neurotransmission is directly enhanced by agonists such as carbachol, which binds and stimulates acetylcholine receptors; REM sleep is also enhanced by eserine, which prevents enzymatic breakdown of endogenously released acetylcholine by cholinesterase. In contrast, REM sleep decreases are seen when muscarinic acetylcholine receptor is competitively occupied by atropine. Middle: REM sleep is decreased by any drug action that increases effective amount of this neurotransmitter. Monoamine oxidase (MAO) is inhibited (MAOI), leading to increase in NE and, via an inhibition of REM generators, a decrease in REM. In contrast, when β‐receptors are blocked by propranolol, postsynaptic REM generator cells are disinhibited and REM sleep is increased. Bottom: reuptake of aminergic neurotransmitter (e.g., by amitriptyline) may also lead to effective increase in serotonergic inhibition of REM sleep. On other hand, when reserpine depletes amount of serotonin (and NE) in vesicles, there is less inhibitory restraint and REM sleep increases.

Cholinergic REM sleep: enhancement effects of carbachol. A: predictions of reciprocal interaction model. Under physiological conditions, REM sleep occurs when D‐on cells are exposed to high levels of cholinergic excitation and low levels of aminergic inhibition. Carbachol increases cholinergic excitation of D‐on cells and thereby tips balance of system toward REM sleep generation. B: polygraph records. Qualitative identity, in all channels of physiological and pharmacological D. In both cases, EMG is flat, indicating atonia; cortical EEG (CX) is desynchronized, indicating activation; there are PGO waves in EEG of LGB; there is θ‐activity in hippocampal EEG (HIP); and REMs are indicated by deflections of EOG. C: individual time course data. Compared with control (above) note onset (<10 min) of prolonged (>60 min) D episodes seen after each of 3 carbachol cannula placements in pontine reticular formation. D: pooled time course data. Consistent enhancement of D seen across carbachol trials and subjects. At 1 h, peak value of 75% is 4–5 times greater than control and the effect lasts for 3 h.

Cholinergic REM sleep enhancement: ionotophoretic effects of carbachol. A: chemical microstimulation. Microiontophoretic drug delivery system. Head restraint implant placed stereotaxically so that glass microiontophoretic pipette can be directed at pontine brain stem under chronic recording conditions. Electrographic D with carbachol (D‐carb) can be produced with currents as low as 300 nA for 10 min, and action potentials of single giant cells can be recorded through same pipette before and after delivery of drug. B: microiontophoretically induced D‐carb. Electrographic signs after delivery of carbachol by passing current through glass micropipette indistinguishable from those of physiological D and from D‐carb induced in cannula diffusion experiments. Injection was histologically localized to rostral pole of paramedian pontine reticular formation. C: iontophoretically induced PGO waves. At site different from that of B, note stereotyped groups of 4–8 waves that begin to occur, in absence of other D signs, at 33–52 min and are still present at 13 h, 42 min postinjection. Data suggest that it may be possible to chemically microdissect neuronal ensembles responsible for D‐state components. D: cross‐correlation of iontophoretically induced PGO waves and cell activity. Average PGO waveform and activity of single cell cross‐correlated in tape‐recorded segments of data shown in C. Note identity of PGO waveform (upper trace) to that seen in physiological D. Lower trace: histogram in which peak of PGO waves is taken as time zero and frequency of unit firing is counted for 500 ms before and after wave peak. Unit activity shows sharp increase in 40 ms prior to wave peak. This neuron histologically localized to PGO burst cell zone (see Fig. 35A).

Cholinergic REM sleep enhancement: intrabrain differentiation of enhancement and suppression stimulation sites. A: enhancement sites. Left: anterodorsal pontine reticular formation from which pure REM sleep is evoked. Middle: posteroventral pontine reticular formation from which REM sleep with stereotyped side‐to‐side alternation of eye movements (EMs) and PGO waves are evoked. Right: peribrachial pontine tegmentum from which state‐independent PGO waves are evoked. B: suppression sites. Left: midbrain reticular formation from which arousal with persistent motor circling is evoked. Middle: medullary reticular formation from which arousal with axial trunk rolling is evoked. Right: periabducens pontine tegmentum from which arousal with persistent oscillating eye movements (OEMs) is evoked. In each diagram parasagittal distance (in mm) is indicated. For sections at 1.2 mm: MLB, medial longitudinal bundle; TV, ventral tegmental n.; 6N, abducens nerve; TG, genu of facial nerve; 6, abducens nucleus; PH, praepositus hypoglossi. For section at 3.7: FTP, paralemniscal tegmental field; 5M, motor division of trigeminal nucleus; 7N, facial nerve; SO, superior olive; 7, facial nucleus; SA, striae acousticae; VL, lateral vestibular n.; S, solitary tract; VS, superior vestibular n.; EMG, nuchal EMG; EEG, cortical EEG; LGB, lateral geniculate body EEG; EOG, interorbital EOG. Each record ∼20 s.

Cholinergic REM sleep enhancement: effects of bethanechol. A: molecular configuration. Top: acetylcholine, naturally occurring neurotransmitter that is rapidly degraded by cholinesterase. Middle: carbamylcholine (carbachol), long‐acting synthetic agent that resists enzymatic degradation. Carbachol is mixed nicotinic and muscarinic agonist. Bottom: β‐methyl carbamylcholine (bethanechol), long‐acting synthetic agent but pure muscarinic agonist. B: polygraphic records. EOG, flurries of REMs; EMG, complete suppression of nuchal muscle tone; EEG, desynchronization of cortical electrical activity; LGB, biphasic waves in LGB. Bethanechol‐induced state appears to be an intensified version of naturally occurring D state. C: time course data. Top: bethanechol trials and matched saline control recordings for 3 cats. At latencies of 10–30 min, D periods of up to 55 min ensue and are virtually continuous for 2 h. In all cases, latencies are shorter, durations longer, and intervals shorter than control values. Bottom: D‐beth values shown in top were pooled with 30‐min bins. Peak (85%) was reached at 1 h followed by progressive decline to trough (30%) at 3.5 h but with subsequent peaking. First time at which control levels matched experimental values was 5.5 h postinjection. D: anatomical differences indicate site specificity of enhancement and suppression of D by bethanechol. Enhancement of D was seen only at pontine sites where 4 of 7 trials produced at least a doubling of saline control values and only 1 produced suppression. In contrast, suppression was rule at medullary and midbrain sites where 6 of 8 trials produced at least a halving of control values and none produced enhancement. Four of the 6 medullary trials produced complete suppression of D.

Antiaminergic REM‐sleep enhancement: effects of propranolol. A: predictions of reciprocal interaction model. Under physiological conditions, noradrenergic inhibition restrains cholinergic D generator. With β‐adrenergic blockade by microinjection or propranolol, D generator is disinhibited. B: polygraphic records show natural D onset and that seen after propranolol. There is no synchronized electroencephalographic activity in the cortex (CX) so that drug‐treated animal enters D directly from W. Channels as in Fig. 44. C: individual time course data. Enhancement of D by propranolol occurs by way of more episodes of normal length indicating that either threshold is not as effectively lowered as with carbachol and/or that serotonergic inhibition is capable of ending drug‐induced episodes. D: pooled time course data. Peak effect, at 1 h, is 2–3 times control. After return to base line at 3 h, there appears to be an undershoot or suppression that is not seen with cholinergic agonist enhancement.

Pathophysiology of narcolepsy. Narcoleptic patients have several abnormalities of sleep‐waking state control that appear to be result of changes in set point of REM state oscillator. During waking, REM sleep signs intrude as sleepiness, sleep attacks, and cataplexy. This increased propensity to REM is measured as decreases in multiple‐sleep latency test and may manifest itself in hypnogogic hallucinations or by sustained sleep‐onset REM period (normal subjects go quickly through stage I). On arousal from REM sleep, there may be corresponding persistence of mental phenomena of REM (hypnopompic hallucinations) and/or REM sleep motor inhibition (sleep paralysis). Reciprocal interaction model accounts for all these phenomena by hypothesizing decreased level of aminergic inhibition (light dashed line) and/or increased level of cholinergic excitation (light solid line) in pontine oscillator. During waking there is an increased propensity for REM generator to reach threshold. At sleep onset there is brief escape from aminergic restraint and sleep‐onset REM period is triggered. System then resets and cycles normally until end of REM when there is a lag in reinstitution of waking‐state conditions and REM phenomena again escape their normal temporal bounds. Clinical efficacy of aminergic agonists (e.g., amphetamine) or amine reuptake blockers (e.g., imipramine) may be due to their capacity to reset aminergic inhibition to normal level (heavy dashed line). By reciprocal interaction, this would also reset level of cholinergic generator (heavy solid line).

Pathophysiology of sleep apnea syndrome. During waking, respiratory oscillator of medulla receives tonic drive from other neural structures and can respond to voluntary and metabolic signals to change breathing pattern. Ventilation is assured by active maintenance of airway pathway via tonus of oropharyngeal musculature. In NREM sleep, central drive on both respiratory oscillator and peripheral muscles declines due to disfacilitation. Respiratory rate and amplitude thus diminish and airway is subject to collapse. If obstruction occurs, forced expiratory effort may actually aggravate airway construction and prolonged apneas with marked hypoventilation and hypoxia may occur. During REM sleep, activation of pontine generator neurons produces tonic and phasic driving of respiratory oscillator, which may desynchronize leading to hyperpnea and/or to apnea. In addition, medullary oscillator becomes unresponsive to metabolic signals. In patients with tendency to airway collapse, these processes may multiply deleterious effects of ventilation.

Motor activity in sleep: timing of posture shifts. A: posture shifts in video data. Movements are scored by making frame‐by‐frame comparisons of posture on replay of tape. Left, top and bottom: unambiguous posture shifts (>90% trunk rotation). Right, top and bottom: no such postural adjustments. Right, top: no movement is discernible; bottom: arm but no trunk movement. Drawings were made by tracing outlines of human subject in time‐lapse photographic study. B: individual records. Movement profiles characteristic of prompt (top) and delayed (bottom) sleep onset. Visual cross‐correlation between EEG data and posture shifts was established for each subject‐night. Movement ceases when NREM stages are progressive (top) but persists when waking and stage I alternate (bottom). Stereotyped coordination of posture shifts and sleep‐stage sequence followed sleep onset. Descending stages of NREM sleep, especially those reaching stages III and IV, were movement free. However, ascending NREM stages and REM (black areas) interruptions or terminations were accompanied by movements. C: pooled movement data. Immobility and sleep cycle phase. Top: average of 44 sleep cycles. Bottom: time of beginning (left curve) and ending (right curve) of 44 epochs of postural immobility related to average of all sleep cycles in which they occurred (top). Each curve is cumulative histogram of percentage of occurrences of immobility as function of percentage of cycle completed. Note steep and smoothly ascending curve of onsets indicating that immobility begins in association with early NREM sleep; curve of endings is by contrast inflected sharply at stage IV onset, which indicates that process controlling posture shifts is activated well before end of NREM sleep. D: neuronal activity curve. Actual activity of presumed REM generator neuron in cat brain stem (solid stepped curve) is compared with theoretical curves of on‐cell population (solid curve) and off‐cell population (dotted curve) as function of percent of cycle completed. Comparing D with C reveals inverse parallelism between timing curve of onsets in C of immobility and off‐cell trajectory in D; it is hypothesized that motor disfacilitation is occurring during first third of cycle. Later, curve of brain stem neuronal activation (D) directly parallels immobility offset in C. Both appear to be manifestations of phase shifts in motor systems such that immobility of REM sleep is produced by active peripheral inhibition in face of strong central facilitation.

Mental activity in sleep: psychophysiology of dreaming. A: systems model. As a result of disinhibition caused by cessation of aminergic neuronal firing, brain stem reticular systems autoactivate. Their outputs have effects including depolarization of afferent terminals causing phasic presynaptic inhibition and blockade of external stimuli, especially during bursts of REM, and postsynaptic hyperpolarization causing tonic inhibition of motoneurons that effectively counteract concomitant motor commands so that somatic movement is blocked. Only oculomotor commands are read out as eye movements because motoneurons are not inhibited. Forebrain, activated by reticular formation and also aminergically disinhibited, receives efferent copy or corollary discharge information about somatic motor and oculomotor commands from which it may synthesize such internally generated perceptions as visual imagery and sensation of movement, both of which typify dream mentation. Forebrain may in turn generate its own motor commands, which help to perpetuate process via positive feedback to reticular formation. B: synaptic model. Some directly and indirectly disinhibited neuronal systems, together with their supposed contributions to REM sleep phenomena. At level of brain stem, 4 neuronal types are illustrated. MRF: midbrain reticular neurons projecting to thalamus that convey tonic and phasic activating signals rostrally; PGO: burst cells in peribrachial region that convey phasic activation and specific eye movement information to geniculate body and cortex (dashed line indicates uncertainty of direct projection); PRF: pontine reticular formation neurons that transmit phasic activation signals to oculomotor neurons (VI) and spinal cord, which generate eye movements, twitches of extremities, and presynaptic inhibition; BRF: bulbar reticular formation neurons that send tonic hyperpolarizing signals to motoneurons in spinal cord. As a consequence of these descending influences, sensory input and motor output are blocked at level of spinal cord. At level of forebrain, visual association and motor cortex neurons all receive time and phasic activation signals for nonspecific and specific thalamic relays.