the hypnagogic state: a critical review of the literature

Psychological Bulletin1976, Vol. 83, No. 3, 452-481

The Hypnagogic State: A Critical Reviewof the Literature

Daniel L. SchacterVeterans Administration Hospital, Durham, North Carolina

In this paper, literature concerned with the hypnagogic state is reviewed. Theterm hypnagogic state is used in a descriptive sense to emphasize that nu-merous phenomena characterize the drowsy interval between waking andsleeping. Major methodological issues are considered, psychological andphysiological aspects of the hypnagogic state are reviewed, and theories con-cerned with the genesis and function of hypnagogic phenomena are criticallyevaluated. Possible directions for future research are outlined, with emphasisplaced on elucidating the patterns of psychological and physiological phe-nomena which characterize the hypnagogic state.

Leaning (1926) prefaced her review on thehypnagogic state by noting that no sys-tematic treatment of the topic had yet sur-faced. She felt that investigation of the hyp-nagogic state would be fruitful for psycholo-gists and urged researchers to systematicallyexplore "this little cultivated area" (p. 290).The 40 years following Leaning's studyshowed less research on the subject than hadthe years previous to its publication. Holt(1964), in his paper concerning the banish-ment and subsequent return of imagery re-search, pointed out that no information waslikely to be found concerning the hypnagogicstate in either introductory or advanced psy-chology textbooks.

The appearance in the last few years ofexperimental studies exploring the hypnagogicstate indicates that widespread interest inhypnagogic processes is beginning to emerge.Unfortunately, information regarding hypna-gogic phenomena is widely scattered and un-known to most psychologists. For instance, myown search of the literature has uncoveredonly four authors (Freedman, Grunebaum,

This work was supported by Veterans Administra-tion Grant 1620-01.

My thanks to Herbert F. Crovitz for encourage-ment and advice throughout this project, to EdwardF. Kelly for a critical reading of the paper, and toNina Jensen and Kenneth Schacter for their valuableaid in the final preparation of the manuscript.

Requests for reprints should be sent to Daniel L.Schacter, Veterans Administration Hospital, 508Fulton Street, Durham, North Carolina 27705.

Stare, & Greenblatt, 1962; McKellar & Simp-son, 19S4; Oswald, 1962; Tart, 1969) whocited the Leaning (1926) study, which standsas one of the most thorough phenomenologi-cal treatments of the hypnagogic state avail-able today. Indeed, if one constant can beidentified in the hypnagogic literature, it isthat most investigators have worked in ig-norance of one another's efforts. The presentpaper is an attempt to at least partially rec-tify this discouraging situation by providinga systematic review of the existing literatureon the hypnagogic state.

The hypnagogic area differs from most top-ics addressed in contemporary psychology.There are no standard experimental proce-dures, comprehensive theoretical systems, orwell-known empirical controversies. Accord-ingly, I have organized this review with twoprincipal goals in mind: (a) to group themyriad topics considered in the hypnagogicliterature such that the issues and trendswhich do exist are clearly exposed, and (b) topreserve maximal informational value withinthe context of the groupings—that is, not allpoints in the literature can be touched upon,but an effort has been made to eliminate onlythose findings which appear least related tothe body of observations.

Definition of Terms and Discussion of CriteriaTo what exactly do we refer when we in-

voke the phrase hypnagogic state? The lit-erature to be reviewed shows that a variety ofpsychological phenomena have been labeled

452

HYPNAGOGIC STATE 453

hypnagogic: spontaneously appearing visual,auditory, and kinesthetic images; qualitativelyunusual thought processes and verbal con-structions; tendencies toward extreme sug-gestibility; symbolic representation of ongoingmental and physiological processes; and so on.Additionally, specific physiological indexeshave been observed in conjunction with thesephenomena: A slowing of frequency and de-pression of amplitude of the electroencephalo-gram (EEC) and the appearance of slow eyemovements are fairly well documented; andrecent studies suggest that a low rate offrontalis muscle activity and changes in res-piratory patterns may accompany the abovepsychological phenomena.

The factor common to these various psycho-logical and physiological phenomena is thatthey occur in a similar context, the drowsyinterval between waking and sleeping. In thepresent paper, the term hypnagoglc stale isused to refer to that period between wakingand sleeping when some or all of these phe-nomena occur conjointly. It should be notedthat the state concept is invoked here in adescriptive rather than an explanatory sense(Hilgard, 1969); the point to be emphasizedis that a multiplicity of phenomena should beconsidered when discussing the hypnagogicstate.

Considering the concept of the hypnagogicstate in a descriptive sense, we must clarifysome fundamental issues. Specifically, the in-dividual criteria which are relevant to a de-scription of the hypnagogic state need to beidentified. The literature as it exists providesa general description of the hypnagogic statein terms of the psychological and physiologi-cal criteria listed above, but these parametersrequire a more precise formulation, and theyare only a few of the important parameterswhich require study. For instance, little isknown about the nature of memory storageand retrieval processes in the hypnagogicstate; about the response capacity of physio-logical systems in the hypnagogic state (seeJohnson (1970) and Williams, Hammack,Daly, Dement, and Lubin (1964) for discus-sions of the importance of this parameter fordescribing a state of consciousness); about therelative contributions of the left and right

hemispheres to hypnagogic mentation; andso on. Moreover, the problem of elucidatingthe precise patterns of these criteria whichconverge to describe the hypnagogic state,and the degree to which these criteria main-tain a consistent relation to one another is ofcritical importance and has not yet been ade-quately confronted. There is some evidencewhich suggests that there are a succession oftemporally ordered stages within the hypna-gogic interval that can be described by dif-ferent patterns of psychophysiological criteria(Foulkes & Vogel, 196S; Vogel, Foulkes, &Trosman, 1966); further work on this prob-lem is needed before the patterns of criteriawhich adequately describe the hypnagogicstate become clear.

In the literature, the problem of criteriahas largely been addressed in the context ofdifferentiating the waking state from the on-set of the hypnagogic state and differentiatingthe dream state from the termination of thehypnagogic state.

Vihvelin (1948) distinguished between theinternal and external methods of differentiat-ing the hypnagogic state from the dream state.The internal view is espoused by Leroy(1926) and Burdach (cited in Vihvelin), whohold that an analysis of the phenomenon it-self is sufficient to decide whether the episodeis hypnagogic. The external view is advocatedby Schultz and Weygandt (both cited inVihvelin), who argue that the subject's degreeof awareness of the external world is decisive.Weygandt's principal criterion is that theperson must show continuing consciousnessof the environment if the episode is to beconsidered hypnagogic. Another way of mak-ing this distinction is through use of theEEC. Davis, Davis, Loomis, Harvey, andHobart (1937), Dement and Kleitman(1957), and Liberson and Liberson (196S)all reported that the appearance of 12-14-Hzspindles in the EEC is generally a reliable in-dication that a person has lost consciousnessof the external situation.

Considering the problem of ascertaining theonset of the hypnagogic state, Vihvelin pos-ited that two stages characterize its initiation.First, there is a stage of increasing fatigue,marked by a quantitative reduction of thought

454 DANIEL L. SCHACTER

processes; next, the proper hypnagogic state,characterized by qualitative changes in men-tal content. Davis et al. (1937), Foulkes andVogel (196S), Green, Green, and Walters(Note 1), and Liberson and Liberson (1965)suggested that the onset of occipital alphablocking and the appearance of low-voltage,4-7-Hz theta waves in the EEG signal thebeginning of the hypnagogic state.

Although little systematic research has beendirected toward this problem, it is clear thatspecification of the moment of hypnagogiconset or termination is impossible. Rapaport(1967b) has noted that "we are dealing withcontinuous transitions where one cannot drawany hard and fast dividing lines. I cannot saythat up to here these are the hypnagogichallucinations and here begin the daydreams. . ." (p. 395). We can only say that a par-ticular phenomenon is more or less in corre-spondence with the general criteria of thehypnagogic state. What future research can dois to elucidate the patterns of psychologicaland physiological phenomena arising withinthe general borders of the hypnagogic state.Then this more refined information can befed back into the construction of a moreprecise definition of the limits of the state.

Historical Background

One of the earliest printed references tohypnagogic phenomena is found in the auto-biography of astrologer Simon Forman, writ-ten in 1600 (cited in Ellis, 1911). As a childForman noted, "So soon as he was always laiddown to sleep he should see in his visionsalways many mountains and hills come rollingagainst him, as though they would overrunhim . . ." (p. 30). Forman believed that thesepresleep visions were sent to him from heav-enly sources as a harbinger of later difficul-ties. Similarly, Emanual Swedenborg, the 18thcentury scientist-turned-mystic is alleged tohave reported encounters with angels uponentering the hypnagogic state (Van Dusen,1972, p. 94). Another early reference to hyp-nagogic occurrences is found in Hobbes'(1651) Leviathan: "A man shall in the dark,though awake, have the images of lines andangles before his eyes; which kind of fancy

hath no particular name, as being a thingthat doth not commonly fall into mens dis-course" (p. 6).

The formal study of the hypnagogic stateappears to have commenced with Baillarger(cited by Vihvelin, 1948), who published anaccount in 1846. Soon after, Maury (1848)invoked the term hypnagogic, derived fromthe terms hypno (sleep) and agogos (induce).The term kypnopompic was coined some yearslater by Myers (1904) in reference to thosephenomena which appear as one makes thetransition from sleep to wakefulness. Hypno-pompic phenomena are not dealt with here.

A large number of the early studies con-cerned themselves with the question of howmany people have experienced at least onehypnagogic image. Muller (1848), the re-nowned physiologist, concluded from ques-tionnaires completed by groups of medicalstudents that the images were extremely rare;however, Leaning (1926), Maury (1848),and Taine (1883) all concluded that hypna-gogic phenomena are relatively common. Thescanty available figures, derived from ques-tionnaire studies, support the latter conclu-sion. McKellar (1972) reported that his owninvestigations using students yielded an inci-dence of 76% and that the investigation ofOwens (Note 2) placed the figure at 77%;Buck and Geers (1967) offered an estimateof 72%.

Figures representing the percentage of peo-ple who have experienced at least one hypna-gogic image are likely to be misleading. Ifone replies negatively when queried as topersonal experience with hypnagogic images,this does not necessarily indicate that theydo not occur. As Gallon (1883) pointed out,people are characteristically reluctant to ad-mit that they experience visions or hallucina-tions of any sort, due to the negative con-notations which society attaches to suchepisodes. Holt (1964) offered an identicalobservation:

In interviewing our subjects about their hypnagogicimagery, I was struck by the discomfort some ofthem showed in confessing that they experienced it;one of them commented on his nocturnal visualiza-tions, "If they were any more real, I'd know I wascrazy." (p. 262)

HVPNAGOGIC STATE 455

McKellar (1972) has pointed out that thetendency toward "false negatives" in hypna-gogic recall is common:

it [hypnagogic imagery] can be overlooked for avery long time even by those who subsequentlycome to realize that they have the experience fre-quently. . . . False negatives seem to occur by aprocess of ignoring what one is not alerted to notice,as well as from emotional blockage, (p. 43)

An anecdotal remark which appears con-sistently in the early literature is that hypna-gogic images surface often in childhood anddecline in frequency as a function of age.Some early data pertaining to this observa-tion, though not extremely dependable, sup-port the anecdotal record. Partridge (1898)asked 826 children to describe what theysaw at night after their eyes were closed butbefore they fell asleep. Hypnagogic imagerywas reported most frequently by the youngerchildren. De Manaceine (1897) found thathypnagogic images occurred in 8 out of 10of the 6-year-olds she studied, in 4 out of 10of the 8-15-year-olds, and still less frequentlyin adulthood. More recently, Foulkes (1971),as part of a longitudinal study of childrens'dreaming, found recall of hypnagogic phe-nomena on 61% of the interruptions of 9-10-year-olds and on only 18% of the interrup-tions of 3-4-year-olds. These data wouldappear to cast some doubt on the earlierspeculations.

De Manaceine (1897) was also the firstinvestigator to suggest that cultural differ-ences might influence the hypnagogic state.One third of her subjects were Finnish, andshe claimed to have uncovered a propensityfor hypnagogic phenomena among them:

as all of them presented a very marked and a verylong state of half-awakening, I cannot look on thisas a pure accident, but I think we have here anational peculiarity. Perhaps this liability to thestate of half-awakening is explained by the fact thatamongst the natives of Finland it is nearly impossibleto find a person of sanguine temperment. (p. 204)

The possibility of cultural factors influencingthe hypnagogic state has not yet receivedserious treatment. The point raised by Gal-ton (1883), that the demands of society mayserve to either inhibit or encourage the emer-

gence of hypnagogiclike material, might beprofitably researched in a cross-cultural con-text.

Before leaving the topic of historical back-ground, it should be noted that the latter halfof the 19th century can be characterized asthe heyday of reports of hypnagogic phe-nomena. A list of the scientists and philoso-phers who considered aspects of the hypna-gogic state is impressive: Ellis (1897, 1911),James (1890), Galton (1883), Kekule (citedby Koestler, 1964, p. 118), Ladd (1892),Muller (1848), Titchener (1909), and Wundt(cited by Leaning, 1926); all offered obser-vations concerning hypnagogic processes. His-torically, this is understandable, as the intro-spective method was practiced widely duringthis period, and Watsonian behaviorism hadnot yet begun its rise to paradigmatic stature.

Methodology

In this section, two major issues are ad-dressed. First, the methods which have beenused to collect data from the hypnagogicstate are critically evaluated; second, thetechniques which have been developed forinducing and prolonging the hypnagogic stateare discussed.

Data collection. Four major modes of ob-taining data from the hypnagogic state arefound in the literature: spontaneous self-ob-servation, systematic self-observation, ques-tionnaire sampling, and controlled experi-mental investigation. These methods varywidely in the degree to which they yield reli-able data, and the present discussion attemptsto expose the strengths and weaknesses ofeach.

Spontaneous self-observation refers to thosereports of hypnagogic phenomena offeredretrospectively by a person who did not spe-cifically intend to systematically observe thephenomena. Leaning (1926) presented thehypnagogic experiences of numerous peoplewhich are of this sort; similarly, Hanawalt(1954), Hebb (1968), Hollingworth (1911),Slight (1924), Varendonck (1921), and War-ren (1921) offered discussions of the hypna-gogic state based largely on their own spon-taneous experience of the phenomena.


One advantage of the spontaneous methodis that it may produce relatively "pure"hypnagogic phenomena. It can be argued thatmethods such as systematic self-observationor experimental investigation may signifi-cantly affect the type of phenomena whichare observed, through processes of self-ful-filling prophecy, experimental demand, andthe like. Thus, a positive aspect of the spon-taneous phenomena is that they may be rela-tively unaffected by these influences.

However, the spontaneous method has amajor flaw. It is difficult to know howthorough the observer has been in reportinghis own hypnagogic experiences. Althoughthere currently exists no systematic studyregarding memory of hypnagogic experiences,it has consistently been noted that hypna-gogic phenomena are extremely difficult toremember if they are not recorded immedi-ately after they occur. The possibility israised that the reported spontaneous phe-nomena are only those most impressive to theobserver. Thus, the validity of spontaneousexperiences as evidential data is lessened inthe absence of a more complete sample ofthe observer's hypnagogic experiences.

Several sources in the literature have usedthe method of systematic self-observation inexploring the hypnagogic state. The majordifference between this and the spontaneousmethod is that here the observer has decidedbefore entering the hypnagogic state that hewill systematically scrutinize his hypnagogicprocesses upon entering it. Two kinds of sys-tematic self-observation are found: one inwhich the author reports on his own hypna-gogic experiences (Maury, 1848; Rapaport,1967a, 1967b; Sartre, 1972; Silberer, 1951;Van Dusen, 1972), and one in which a smallgroup of people are trained to observe theirown hypnagogic processes (Leroy, 1926;Vihvelin, 1948). In each case, the observeracts both as subject and experimenter.

The major value of this method is that anin-depth look at these individuals' hypna-gogic processes is made possible. As Singer(1966) has pointed out, the trends and con-sistencies which are observed on an individualbasis can then be investigated with largernumbers of subjects. Singer has also recog-

nized the major problem with such an ap-proach:The execution of such a fairly complex experimentwith oneself as subject proved difficult, if engrossing.It seems impossible, of course, to avoid biases andgreat self-consciousness, which necessarily alter re-sults from what they might be if there were noawareness of an ongoing experiment, (p. 40)

Thus, the major drawback of the systematicself-observation method concerns the ques-tionable validity of the introspective account.

The questionnaire method of investigatinghypnagogic phenomena has been used fromtime to time in the literature. As noted earlier,several investigators have attempted to specifythe percentage of people who have experiencedat least one hypnagogic image through theuse of a questionnaire (Buck & Geers, 1967;McKellar, 1972; Muller, 1848). The ques-tionnaire has also been used to assess the rela-tive importance of each sensory modality inhypnagogic experience (McKellar, 1957) andto correlate the incidence of hypnagogic phe-nomena with the incidence of other cognitivephenomena, for example, dreams, waking im-agery, and the like (Buck & Geers, 1967;Holt, 1972).

The major difficulty with the questionnairemethod concerns the latency between thetime of the hypnagogic experience and thetime that the experience is assessed. Due tothe problems mentioned earlier concerningthe difficulties in remembering hypnagogicphenomena, which would appear to be ac-centuated when using the questionnaire, datagenerated by the questionnaire methodshould be viewed cautiously. However, thequestionnaire may prove useful in either oftwo ways. First, a questionnaire which as-sesses the major dimensions of hypnagogicexperience can be used to identify subjectssuitable for research purposes. Second, sucha questionnaire might be used to generate hy-potheses which could then be more specifi-cally investigated experimentally.

Systematic experimental investigation ofhypnagogic processes has only recently beenundertaken. With the exception of earlystudies by De Manaceine (1897) and Daviset al. (1937) and sleep studies not primarilyconcerned with the hypnagogic period (Aser-


insky & Kleitman, 1955; Dement & Kleitman,1957), the first detailed experimental investi-gations of the hypnagogic state were reportedby Bertini, Lewis, and Witkin (1964), Greenet al. (1970, Note 1), Liberson and Liberson(1965) and in a series of studies by Foulkesand Vogel and their colleagues (Foulkes,Spear, & Symonds, 1966; Foulkes & Vogel,196S; Vogel, Barrowclough, & Giesler, 1972;Vogel, Foulkes, & Trosman, 1966). Less de-tailed investigations have been provided byBraud and Braud (Note 3), Budzynski(1972), Dement (1965), Schacter and Kelly(1975), and Stoyva (1973).

One major advantage accrued by experi-mental investigation of the hypnagogic stateis that the recalling of hypnagogic eventsover long temporal intervals is minimized byon-the-spot elicitation of internal activity.Additionally, the experimenter is afforded theopportunity to systematically sample the sub-jects' hypnagogic mentation according to aschedule of interruption based on the criteriahe desires.

Another important advantage offered byexperimental investigation of hypnagogicprocesses is the opportunity to address animportant methodological issue—the empiri-cal worth of introspective accounts—throughthe use of converging operations (Stoyva &Kamiya, 1968). Stoyva (1973) argued thatby combining verbal reports with physiologi-cal indicators, the same type of validity thatthe discovery of rapid eye movements be-stowed upon dream research can be conferredupon hypnagogic work. If specific physio-logical indicators are consistently linked withreports of certain types of mental content,while other physiological signals are accompa-nied by systematically different types of con-tent, then the use of these converging opera-tions serves both as a proof of the validityof the introspective account and as a tech-nique for exposing new relationships betweenpsychological and physiological events. Thework of Foulkes and Vogel (1965) is theclearest example of the possibilities suggestedby the technique of converging operations.

The strategy of converging operationspromises to be a fruitful mode of assaying thehypnagogic state. However, there remain

problems in the experimental investigation ofhypnagogic processes which have not beensolved. For instance, the methodological con-siderations in obtaining the verbal report mustbe more adequately worked out. While it hasbeen shown that abrupt interruptions aremore effective than gradual ones for obtain-ing dream reports (Goodenough, Lewis, Sha-piro, Jaret, & Sleser, 1965), it is not clearthat the same holds true for hypnagogic re-search. It can easily be seen that an exces-sively abrupt interruption might prevent asubject from reestablishing the hypnagogicstate. Also, the appropriate form and contentof the questions posed during interruptionshave not yet been adequately formulated.Clearly, questions should be framed so as tounambiguously extract the desired informa-tion. However, they should not be so directiveas to help the subjects "fill in" missing partsof fleeting and difficult-to-reconstruct hypna-gogic episodes.

In sum, it appears that the methods ofspontaneous self-observation, systematic self-observation, and questionnaire sampling areof limited value in testing hypotheses concern-ing the hypnagogic state. Controlled on-the-spot sampling of hypnagogic mentation,though in need of refinement, is clearly themethod of choice for hypothesis testing. How-ever, the nonexperimental methods may proveuseful in generating hypotheses which canthen be experimentally tested.

Induction and prolongation. Green et al.(Note 1) and Stoyva (1973) have observedthat a delicate balance must be achieved be-tween drowsiness and conscious awareness ifthe hypnagogic state is to be maintained. Onemust be relaxed enough to permit the hypna-gogic content to emerge spontaneously, yetone must be sufficiently aware to observe whathas appeared. If one is excessively drowsy,the transition into sleep will be swift andunaccompanied by hypnagogic phenomena; ifone is excessively alert, the hypnagogic ma-terial is effectively blocked. Stoyva (1973)has specified five conditions which encourageentry into the hypnagogic state; the first twoare adapted from West's (1962) theory ofhallucinations: (a) reduction in the level andvariety of sensory input to the brain, (b)


maintenance of an arousal state sufficient topermit awareness, (c) reduction of proprio-ceptive input, (d) ability to shift to a stateof "passive volition," and (e) shift fromdominance of sympathetic to parasympatheticnervous system, "resetting the hypothalamicbalance."

Some recent research has explored the pos-sibility of systematically inducing and pro-longing the conditions which are conduciveto the maintenance of the hypnagogic state.The first attempt at systematically inducingthe hypnagogic state was reported by Kubie(1943), in a clinical setting. He constructedan apparatus which feeds back to the sub-ject his own breath sounds, using a micro-phone, amplifier, and headphones (see Kubieand Margolin, 1944, for a detailed descrip-tion). The monotony of the respiratory feed-back was successful in helping to induce thehypnagogic state. A more developed versionof this strategy has been employed by Ber-tini et al. (1964). These investigators putsubjects in a state of relative sensory isola-tion, using the ganzfeld procedure, and in-structed them to talk continuously. Theyfound that the combination of homogenousvisual stimulation and auditory white noisedid facilitate the appearance of hypnagogic-like material. Braud and Braud (Note 3) andHonorton and Harper (1974) used similarprocedures and found the ganzfeld to be use-ful in producing hypnagogiclike phenomena.My qualification of the term hypnagogic isimportant, as it points to one of the draw-backs of the ganzfeld procedure: It is notclear that this is specifically hypnagogic ma-terial which is surfacing, or anything otherthan free-associative thought. This point isconsidered in more detail later in the paper.

Recent research by several investigatorshas indicated that the technique of biofeed-back training can be a powerful tool for in-ducing and prolonging the hypnagogic state.Green et al. (1970, Note 1) were the first re-searchers to employ this mode of assay.Green et al. noted that theta rhythms in theEEG are often accompanied by reports ofhypnagogic phenomena. They constructed atwo-step training procedure designed to im-prove their subjects' ability to voluntarily in-

duce the hypnagogic state. In part one, bio-feedback training is preceded by several min-utes of relaxation and breathing exercises.The subject then undergoes IS minutes ofalpha-theta feedback, followed by 30 min-utes of theta feedback only. The presence ofalpha is signaled by a 900-Hz tone, thepresence of theta by a 400-Hz tone. In parttwo of the training, the same procedure isfollowed, but now the subject is encouragedto report instances of hypnagogic phenomena.The subject himself may relay hypnagogicepisodes as they occur or may be interruptedby the experimenter, on the basis of the EEG,and asked to describe mental content at thatmoment. The training lasts for 10 weeks, five45-minute feedback sessions held per week.Although Green et al. have not presentedquantitative analyses of their results, theynoted thatalmost all subjects have been able to consciouslyincrease their alpha production . . . and that theta,though more difficult, has also increased. . . . Whatis more significant . . . is an increase in reportedhypnagogic imagery as alpha-theta practice pro-gresses. (Green et al., Note 1, p. 8)

Green et al. have also introduced severalingenious methods of prolonging the hypna-gogic state. For instance, subjects can practiceby themselves using "repeat" alarm clocks. Atthe beginning of the practice session, thesubject sets the alarm to ring immediatelyand then resets it by depressing a bar at thetop of the clock. If he remains sufficientlyaware during the session to depress the barevery 5 minutes, the alarm will not ring.Each time he resets the clock, or is awakenedby it, he records the hypnagogic content thathe is able to recall. Another alerting device isa mercury switch finger ring which acts as anomnidirectional tilt detector. If the subject'slevel of awareness diminishes below a certaincritical point, the forearm will begin to tilt.This closes the mercury switch circuit andsounds a chime, which nudges the subjectback to above-threshold consciousness.

Budzynski and Stoyva (Budzynski, 1972;Stoyva, 1973) have employed a slightly dif-ferent mode of hypnagogic-biofeedback train-ing. These investigators first became inter-ested in the hypnagogic state when using an


electromyographic (EMG) feedback tech-nique to treat tension headaches (Budzynski,Stoyva, & Adler, 1970). They observed strik-ing qualitative changes in cognitive processesas their subjects entered states of deep relaxa-tion, and also noted that low levels of fron-talis muscle activity were accompanied by anabundance of theta in the EEC. Their outlinefor a hypnagogic-inducing biofeedback pro-cedure also calls for a two-step process. First,an EMG procedure which feeds back fron-talis muscle activity is used to facilitate entryinto the desired low-arousal state. Once thesubject has experienced some success inachieving a state of deep relaxation, directtheta feedback commences. Although thestructure of the training procedure suggestedby Budzynski and Stoyva largely parallelsthat of the Green group, one important dif-ference is that Budzynski and Stoyva used ananalog auditory feedback procedure, whileGreen et al. employed a digital one. Budzyn-ski and Stoyva (1969) noted that in thismanner, the subjects are presented with con-tinual information regarding their state ofconsciousness; in the digital procedure, thesubjects know whether they are in or out ofthe desired region, but they do not know howclose they are to their goal. While no large-scale analyses are offered, Budzynski andStoyva reported positive first results using theabove procedure.

The evidence reviewed in this section sug-gests that a basic methodological frameworkfor inducing and prolonging the hypnagogicstate is beginning to emerge. Further devel-opment of this methodology appears to becontingent on several variables. More exten-sive analyses of the biofeedback procedureare called for. It is not apparent in the stud-ies reviewed here precisely what effects wereencountered. Also, the design of these studiesdoes not permit one to attribute all of thereported effects specifically to the biofeedbackprocedure; all of the conditions for hypna-gogic activity listed by Stoyva are met bythe conditions of quietude and relaxation inwhich the biofeedback experiments are con-ducted. Studies comparing "correct feedback,""incorrect feedback," and "no feedback" areneeded in order to more critically evaluate the

efficacy of a hypnagogic-inducing biofeedbackprocedure.1

Finally, it should be noted that techniqueswhich intentionally induce or prolong thehypnagogic interval may yield qualitativelydifferent patterns of phenomena than arefound in the natural transition to sleep. Re-search aimed at discriminating between naturaland induced or prolonged hypnagogic periodswould be highly desirable.

PHENOMENOLOGICAL AND PSYCHOLOGICALASPECTS OF THE HYPNAGOGIC STATE

Most of the literature dealing with phe-nomenological and psychological parametersof the hypnagogic state is concerned exclu-sively with elucidating properties of hypna-gogic imagery; material on other propertiesof the hypnagogic state is less abundant. Theemphasis in this section is on providing awell-rounded view of the various phenomeno-logical and psychological characteristics of thehypnagogic state.

Hypnagogic Imagery

Mode of presentation. One of the most re-marked-upon features of hypnagogic images istheir seemingly spontaneous emergence intoconsciousness. Leroy (1926) noted this, com-menting that hypnagogic images appear in anintrusive manner. Maury (1848) first calledattention to this quality of hypnagogic im-ages, likening them to suddenly appearingfaces in the dark; the images appear "fullblown." Numerous observers (Ardis & Mc-Kellar, 1956; Leaning, 1926; Leroy, 1926;McKellar & Simpson, 1954; Oswald, 1962;Sartre, 1972) have compared the experiencingof hypnagogic images to being a passive spec-tator at a play or movie. Alexander (1909),Horowitz (1970), McKellar and Simpson(1954), Muller (1848), Myers (1957), andRichardson (1969) have all argued that the

1 Budzynski and Stoyva (1969) found that thesubjects receiving appropriate EMG feedback reacheddeeper levels of relaxation than did the subjectsreceiving either no feedback or irrelevant feedback.However, this study did not directly investigate theeffect of feedback on hypnagogic processes.


quality of autonomy is the distinctive featureof hypnagogic images.

Sartre (1972) exposed another feature inthe mode of presentation of hypnagogic im-ages similar to the above:

a radical distinction must be drawn between the waya face appears in perception and the manner thesame face appears in hypnagogic vision. In theformer case something appears which is then identi-fied as a face . . . consciousness must focus on theobject. . . .In hypnagogic vision this discrepancy does not exist.There is no focusing. Suddenly knowledge appeal s,as vivid as a sensory manifestation: one becomesaware of being in the act of seeing a face. (pp. 55-56)

An additional feature of hypnagogic imagesconcerns their localization; that is, are hyp-nagogic images recognized as appearing ininternal space, or are they localized "outthere"? Many of the cases cited by Leaningattributed a quality of externality to theimages. Such examples appear sporadicallythroughout the literature (e.g., Lukianowicz,1959), especially in cases involving children'simagery (McKellar, 1965). Vihvelin (1948)instructed his observers to note whether theirimages appeared in "external" or "internal"space. Unfortunately, Vihvelin did not pre-sent the data pertaining to this distinction.The literature on the whole only rarely showsinstances of externally located images, andthese are largely of the auditory variety. How-ever, this distinction may be of more thantaxonomic import. Liddon (1967), McKellarand Simpson (1954), and Roheim (1952)have noted that the spontaneous experienceof seemingly externally located hypnagogicimages may have played a significant role inthe development of folklore and mythology.

Content of the images. While a scan of theliterature suggests that the range of subjectmatter found in hypnagogic images is bound-less, most of the reported images can be seento fall into a limited number of categories.The visual modality is examined first, fol-lowed by treatments of the auditory and tac-tile-kinesthetic variety. Most of the literaturederived from spontaneous and systematic self-observation indicates that visual images occurmost frequently, followed by auditory and

tactile-kinesthetic.2 McKellar and Simpson(1954), in their questionnaire sample, foundthat auditory images were the most frequentlyexperienced. However, Foulkes and Vogel(1965) and Green et al. (Note 1), in experi-mental studies, reported that visual imagesoccurred most often, in accordance with theself-observational literature. More detailedstudies examining the relations between thesensory modalities in the hypnagogic state arecalled for.

Numerous investigators have noted a con-sistent pattern of sequential development ofvisual hypnagogic images: flashes of color,light, and geometric patterns, followed byfaces and static objects, followed by land-scapes and more complex and prolonged scenes(Critchley, 1955; Foulkes & Vogel, 1965;Gurney, Myers, & Podmore, 1886; Leaning,1926; Leroy, 1926; Sartre, 1972; Green etal., Note 1).

The light flashes and geometric patternsare often viewed as resulting from entopticphenomena (phosphenes, intraocular fluid,self-light of the retina, etc.; see Marshall,1935, for a detailed descriptive treatment ofthese phenomena). Ladd (1892), McKellarand Simpson (1954), and Sartre (1972) pre-sented thorough discussions of the relationshipbetween the entoptic lights and the forma-tion of visual hypnagogic images. The ma-terial gathered in studies of electrically stim-ulated phosphenes (Knoll, Kugler, Eichmeier,& Hofer, 1962) closely resembles the designsand patterns reported in this stage of thehypnagogic state, thus lending support to thecontention that the entoptic phenomena playa major role in generating the geometriclikeimages.

Several writers have observed that the moreformed images of this early stage can betraced to activities pursued during the day(Hanawalt, 1954; Hebb, 1968; Leaning,

2 Leaning (1926) reported the occurrence of ol-factory and gustatory hypnagogic images. McKellar(1957) found evidence for olfactory and thermalimagery. Green et al. (Note 1) also encounteredolfactory and gustatory images. However, all ofthese investigators noted that these types of hypna-gogic imagery are extremely rare.


1926; Leroy, 1926; Vihvelin, 1948; Warren,1921). Thus, Hebb (1968) related:

A day in the woods or a day-long car trip after asedentary winter sometimes has an extraordinarilyvivid aftereffect. As I go to bed and shut my eyes—but not till then, though it may be hours since theconclusion of the special visual stimulation—a paththrough the bush or a winding highway begins toflow past me and continues to do so till sleep inter-venes, (p. 475)

There is some question in the literatureconcerning the relation of these aftereffects ofsensory stimulation to other types of hypna-gogic phenomena. Oswald (1962) argued thatthese experiences occur in relaxed wakefulnessand thus should not be considered hypnagogic.To the contrary, Hebb (1968) suggested thattrue hypnagogic images are delayed after-effects of sensory stimulation. Research ex-ploring this issue would be desirable.

Faces reported in the hypnagogic intervalare notorious for their vividness, detail, andnovelty (Leaning, 1926; McKellar, 1957;Maury, 1848), One of the observers quotedby Leaning asserted, "The faces I see aremore vivid than nature . . . absolutely fullof life and movement . . . always possess viv-idness of expression" (p. 317). Remarks simi-lar to the above are abundant in all sectors ofthe literature. It has been noted by McKel-lar (1965) that the quality of lifelikenesscombined with the tendency for hypnagogicfaces to assume a grotesque appearance oftenresults in children's experience of fear in reac-tion to their hypnagogic imagery. The pointto be stressed here is that hypnagogic facesare almost without exception reported as be-ing unrecognizable to the observer. Staticobjects, conversely, are the images whichseem to be the most readily identifiable bythe observer.

Landscapes are reported as possessing quali-ties of unusual grandeur and beauty. Leaning(1926) noted that while landscapes accountfor a large percentage of the adult images,they are infrequently reported by children.The more complex scenes which arise towardthe end of the hypnagogic interval are gen-erally characterized by an abundance of ac-tivity and movement. As is the case with hyp-nagogic faces, the scenes are typically expe-

rienced as foreign and unrelated to personalexperience.

McKellar and Simpson (1954) have dis-tinguished between the perseverative and im-personal varieties of hypnagogic images. Theperseverative images are those which are read-ily explainable in terms of past experience,while the impersonal variety "are not easilyexplainable in terms of past experience, andmay seem interestingly or alarmingly foreignto the personality" (p. 272). Regarding theabove content categories in these terms, cer-tain of the geometric and object images maybe seen as perseverative, while the scenes andfaces can be subsumed under the impersonalcategory.

Auditory hypnagogic images break downinto three general areas of content. First,there are the instances in which a personhears his own name "called" (Leaning, 1926;Maury, 1848; Oswald, 1962; Green et al.,Note 1). Green et al. provided the exampleof an experimental subject who, thinking thathe had been solicited, unhooked himself fromthe biofeedback equipment and ran into thehallway, only to realize that he had experi-enced an auditory hypnagogic image. Also,there are the times the figures in visual hyp-nagogic images are heard to speak; Leaningoffered numerous examples of this. Finally,there are those auditory images in whichmusic and chimelike sounds are heard. Mc-Kellar (1957) related that one of his subjectsheard distinct auditory hypnagogic images ofRachmaninoff's Second Piano Concerto witha clarity that rivaled the sensory experience.

Tactile-kinesthetic images assume variousforms in the hypnagogic state. McKellar(1957) offered an example of kinestheticimagery from one of his subjects: "I oftenimagine myself out hunting on horseback andcan feel myself bracing to take jumps, checkmy horse, urge him on, etc." (p. 40). Distor-tions of the body are experienced as kines-thetic images in the hypnagogic state(Critchley, 1955; Isokower, 1938; Kliiver,1942; McKellar, 1957; Green et al., Note 1).Both Kliiver and McKellar have offered ex-amples of polyopia, dysmegalopsia, and dys-morphosia of the somatosensory sphere oc-curring in the hypnagogic state. Kliiver


speculated upon the possible role of the ves-tibular apparatus in these distortions, but noexperimenter has yet empirically evaluatedthis suggestion.

Isokower noted that the mouth seems totake on great importance in the hypnagogicstate: "A hypercathexis of the oral zone isestablished: 'I am all mouth'" (p. 338).Lewin (1946, 1948) offered similar observa-tions, and both of these psychoanalyticallyoriented investigators interpreted this formof kinesthetic hypnagogic imagery as evidenceof "regressive" tendencies.

A final type of kinesthetic hypnagogic ex-perience is that of "falling" (Bertini et al.,1964; McKellar & Simpson, 19S4; Roheim,1952). Bertini et al. and Roheim have sug-gested that the experience of falling is sym-bolic of the descent into sleep, or "sinkinginto oneself." Oswald (1962) has noted thatthe experience of falling may be accompaniedby sometimes severe involuntary jerks of thebody.

Relation of Hypnagogic Imagery to OtherTypes of Imagery

Dream imagery. An issue which has sur-faced continually throughout the hypnagogicliterature is the relation between hypnagogicimages and dreams. The feature which mostconsistently differentiates between the two isthat one does not hallucinate one's ownactivities in hypnagogic images, while one isalmost always a participant in one's dreamimagery (Leaning, 1926; Leroy, 1926; Os-wald, 1962; Rapaport, 1967a, 1967b; Sartre,1972; Taine, 1883). Foulkes and Vogel(1965), in contradistinction to this widelyheld notion, found that their subjects didhallucinate their own activity in hypnagogicepisodes and that participation increased asthe subjects moved deeper into the hypna-gogic state.

A second difference which consistentlyemerges concerns mode of presentation: hyp-nagogic images appear as disconnected "snap-shots," while dreams are characteristicallylonger and better organized (Dement & Kleit-man, 1957; Kubie, 1943; Leaning, 1926; Le-roy, 1926; Maury, 1848; McKellar & Simp-

son, 1954; Oswald, 1962). Foulkes and Vogcl(1965) presented both supporting and con-flicting results. They observed that while thetime of nocturnal dreams typically mirrorsthe time course of the events represented inthe dreams, hypnagogic episodes tend to oc-cur more quickly than the incidents in theimagery, thus lending support to the previousaccounts. However, they also found that hyp-nagogic episodes are as well organized andcomplex as dreams; Foulkes (1966) referredto hypnagogic images as "condensed hiero-glyphics" (p. 132). Archer (1935), on thebasis of self-observation, also stated thathypnagogic imagery can be as complex andwell organized as dream imagery.

The results of the Foulkes and Vogel studyalso suggest that the presence of affect candistinguish hypnagogic images from dreamimages. They found hypnagogic imagery to be"affectively flat"; in instances in which emo-tion was manifest it was rarely intense. Incontrast, the authors reported a 50% inci-dence of experienced emotion in a previousstudy of rapid eye movement (REM) dream-ing (Foulkes, 1960).

The most outstanding resemblance betweenhypnagogic and dream imagery concerns thebizarre and generally fantastic qualities whichcharacterize both (Bertini et al., 1964;Foulkes & Vogel, 1965; Foulkes et al., 1966;Leroy, 1926; Maury, 1848; Oswald, 1962;Varendonck, 1921). Foulkes et al. (1966) hadtwo judges rate hypnagogic and REM dreamreports on a 0-7 continuum of DreamlikeFantasy, where 0 indicates no content, 4 indi-cates nonhallucinatory content, 7 indicatesbizarre content, and so on. Ratings on thisscale initially revealed no difference betweenthe quality of hypnagogic and dream content.However, when a "strict" criterion of imagerybizarreness was used (7 by both judges onthe Dreamlike Fantasy Scale), it was foundthat significantly more dream reports metthis standard.

Vogel et al. (1972) conducted a detailedstudy comparing hypnagogic, nonrapid eyemovement (NREM), and REM mentation interms of their "composite regressivity" (de-gree of bizarre, perceptual, hallucinatory, etc.,content). Judges were used in a variety of


tasks in which they attempted to classify theprotocols of 28 subjects, which had been ob-tained from hypnagogic, NREM, and REMperiods in two nights of laboratory study,into the correct category based on the contentand quality of the protocol. Two additionaljudges rated protocols on the Dreamlike Fan-tasy Scale. Ratings on the Dreamlike FantasyScale showed NREM reports to be signifi-cantly less regressive than hypnagogic re-ports, which in turn were significantly lessregressive than REM reports. In the at-tempted classification of protocols into thethree categories, the judges correctly identi-fied 64.5% of the hypnagogic reports (sig-nificantly above chance), 40.3% of the REMreports, and 44.3% of the NREM reports(neither significantly above chance). Whendiscriminations were made between each pairof stages, the hypnagogic-REM pair was cor-rectly discriminated 62.9% of the time, thehypnagogic-NREM pair 61.2%, and theREM-NREM pair 67.1% (all significantlyabove chance). Thus, the results indicate bothdifference and overlap in the composite re-gressivity dimension of the three types ofmentation.3

The qualitative resemblance between hyp-nagogic and dream imagery led a number ofthe early writers (Hicks, 1924; Ladd, 1892;Leroy, 1926; Maury, 1848; Muller, 1848) toa conclusion similar to the one reached byEllis (1911): "Hypnagogic imagery presentsus with the germinal stuff of dreams. If it isnot identical with the fully formed dream, itis still the early stage of dreaming" (p. 32).Similarly, Oswald (1962) has argued thathypnagogic images should be characterized as"micro-dreams" (p. 98). The work reviewedhere does not support the equatability ofhypnagogic and dream imagery. Indeed, itmust be pointed out that with the exceptionof the Foulkes groups (who compared hypna-gogic images to REM dreams) and the Vogelet al. study, no distinction is made in theliterature between REM and NREM dreamswhen dreams are compared to hypnagogicimages. Given the differences between REMand NREM mentation (Foulkes, 1962; Giora& Elam, 1974), it is clear that a distinction

must be made between REM and NREMdreams in future attempts to investigate hyp-nagogic versus dream imagery. Investigationsof this sort are necessary before a clearer pic-ture of the phenomenological-psychologicalrelation between hypnagogic and dream im-agery can be drawn.

Drug-induced imagery. Several authors havecalled attention to the relation between hyp-nagogic and mescaline imagery (Ardis & Mc-Kellar, 1956; Kltiver, 1942; McKellar &Simpson, 1954; Mitchell, 1896; Rawcliffe,1952). Mitchell (1896) was the first writer toexpose this similarity, emphasizing the au-tonomous nature of both imagery types; healso postulated a possible interaction betweenhypnagogic and mescaline imagery, notingthat his mescaline imagery intensified as heneared sleep. Ardis and McKellar (1956, pp.23-27) offered the most extensive compari-son between hypnagogic and mescaline im-

3 The Vogel et al. (1972) study raises an interest-ing problem. Their hypnagogic reports were obtainedfrom (a) sleep onset after all-day wakefulness, (b)sleep onset after a brief awakening from REM sleep,and (c) sleep onset after a brief awakening fromNREM sleep. DeStrooper and Broughton (1969),Fiss, Klein, and Bokert (1966), Lavie (1974), andLavie and Giora (1973) have provided evidencethat some cognitive and physiological parameters ofa sleep stage may persist into the wakefulnessimmediately following awakening from that sleepstage. Thus, it is possible that hypnagogic episodeswhich follow a REM awakening may contain char-acteristics of REM mentation and that hypnagogicepisodes which follow an NREM awakening containcharacteristics of NREM mentation. Vogel et al.investigated the possible influence of this factor onhypnagogic-REM discrimination. They found thatthe percentage of errors in hypnagogic-REM dis-crimination were about the same when the hypna-gogic report had been obtained at initial sleep onsetor after a REM awakening. However, there wereonly half as many errors in hypnagogic-REM dis-crimination when hypnagogic reports followed anNREM awakening.

This finding suggests that there are differences be-tween the hypnagogic state following all-day wake-fulness, with which the present paper is concerned,and the transition to sleep following an awakeningfrom a sleep stage. Understanding the nature of thesimilarities and differences between these hypnagogicintervals would appear to depend on precise specifica-tion of the patterns of psychological and physiologi-cal phenomena which characterize each.


agery. These authors delineated four principaldimensions of similarity:

1. Mode of presentation—autonomy andrapid rate of change are common toboth types of imagery.

2. Quality—three dimensions of similaritywere noted: unusual vividness, brilliantcolors, and microscopic clarity of detail.

3. Content—some or all of Kliiver's "formconstants" (see below) are common toboth; strange faces and scenes of re-markable grandeur also occur in bothforms of imagery.

4. Synesthesia—although found more com-monly in mescaline imagery, synestheticepisodes are found in hypnagogic im-agery.

Ardis and McKellar (1956) speculated thatthe processes involved in hypnagogic andmescaline imagery are fundamentally alike. Asimilar position was advocated earlier byKliiver. In attempting to specify hallucina-tory constants, Kliiver (1942, p. 177) iden-tified the following "form constants": (a)grating, lattice, honeycomb, fretwork, and soon; (b) cobweb; (c) tunnel, funnel, alley,cone or vessel; (d) spiral. Kliiver noted thatsome or all of these form constants are presentin both hypnagogic and mescaline experi-ences. He also posited that polyopia, dys-megalopsia, and dysmorphosia are evident inhypnagogic and mescaline imagery. Kliiveremphasized that these number, size, andshape distortions occur in the somatosensorysphere as well as the visual sphere in bothcases.

The similarity between hypnagogic imageryand LSD imagery is suggested in a passagefrom Caldwell (1968); a colleague tells himthat hypnagogic imagery is the best way ofhallucinating without LSD (p. 109). There islittle in the way of empirical evidence onwhich to test these various observations.Holt (1972) reported a .00 correlation be-tween previous experience with hypnagogicimagery and experienced intensity of experi-mental LSD imagery in 30 actors. However,this zero correlation does not directly pertainto the phenomenological relation between the

two. Aside from this inconclusive bit of evi-dence, the literature has nothing specific tooffer concerning the phenomenological rela-tion between hypnagogic and drug-inducedimagery or the effect of prior experience withone on subsequent production of the other inexperimental situations.

Sensory deprivation imagery. Several au-thors have attempted to relate the reportedvisual sensations (RVSs) of sensory depriva-tion experiments to hypnagogic imagery. Thishas been done either by asserting that RVSsessentially are hypnagogic images or by sug-gesting that previous experience with hypna-gogic imagery predicts frequency or qualityof RVSs in sensory deprivation situations.

Freedman and Greenblatt (1960) andFreedman et al. (1962) hold the position thatRVSs are most likely to occur in periods ofdrowsiness and are thus hypnagogic images.However, the empirical literature militatesagainst accepting this hypothesis. Leiderman(cited in Zuckerman, 1969) presented EECdata which show that RVSs occur in eitherwaking or drowsy states. Bexton, Heron, andScott (1954) and Short and Oskamp (1965),on the basis of EEC and electrooculogram(EOG) data, concluded that RVSs occur dur-ing periods of alertness; Zuckerman andCohen (1964) reported that RVSs emerge al-most exclusively in waking periods. Zucker-man and Hopkins (1966) noted that if theRVS-hypnagogic parallel were valid, then per-sons who doze and sleep frequently duringisolation should have more RVSs than personswho remain alert. Goldberger and Holt(1958) found no correlation between RVSsand time spent sleeping; Holt and Goldberger(1961) found a correlation of —.66 in theiractor sample; and Zuckerman, Albright,Marks, and Miller (1962) found a —.71 cor-relation between sleep ratings and RVS com-plexity. Murphy, Myers, and Smith (cited inZuckerman and Cohen, 1964) found thatthose subjects who reported frequent episodesof drowsiness also reported less complexRVSs than did alert subjects. Clearly, theliterature does not support the equatabilityof hypnagogic and sensory deprivation im-agery, hence the qualifying kypnagogiclike inreference to the ganzfeld work cited earlier.


Contradictory evidence exists concerningthe question of how previous experience withhypnagogic imagery relates to RVSs in experi-mental situations. Freedman et al. (1962) re-ported a significant (p — .05) correlation be-tween prior history of hypnagogic imageryand subsequent experience of imagery in sen-sory deprivation.4 Holt (1972) found an in-significant (r ~ .29) correlation between priorhistory of hypnagogic imagery and experi-mental isolation imagery in the actor sample.In a further study with a different sample, heagain found extremely low correlations be-tween history of hypnagogic experience andfrequency of isolation imagery (r — .16) andintensity of isolation imagery ( r — . O l ) . Itshould be noted in reference to all of thesestudies that the questionnaire method of as-sessing experience with hypnagogic imageryis highly unreliable, for reasons discussedearlier.

The evidence gathered from these variedsensory deprivation researches supports Holt's(1964) statement that "the attempt to referall the imaginal phenomena from these experi-ments to hypnagogic imagery was a consid-erable oversimplification . . ." (p. 261). Sen-sory deprivation phenomena might be moreprofitably viewed as a potpourri of varioustypes of thought and ideation, including day-dreams, fantasies, hypnagogic images, anddreams. Certainly, hypnagogic images can bea part of the sensory deprivation experience,but they do not appear to be a major com-ponent.

Waking imagery. Although the relationshipbetween hypnagogic imagery and waking im-agery has received practically no systematictreatment, interesting speculations are foundin the older literature. It should be kept inmind that these observations are largely theresult of spontaneous self-observation.

Leaning (1926) presented the cases of sev-eral observers who claimed to be exceptionallyweak imagers in the waking state, yet expe-rienced extremely vivid hypnagogic imagery.Taine (1883), who experienced feeble im-agery while awake found his hypnagogic im-agery to be intense, distinct, and lasting. My-ers (cited in Leaning, 1926) spoke of "otherbad visualizers besides myself deriving from

hypnagogic illusions some notion of the goodvisualizer's habitual capacity" (p. 304). Hol-lingworth (1911) noted a similar phenome-non, which he dubbed "the transformation ofthe imagery type" (p. 104). He suggestedthat modes of imagery weak in the wakingstate become dominant while drowsy. He of-fered himself as an example of an auditory-motor imager while awake and a vivid visualimager while drowsy. Hollingworth addition-ally noted that E. B. Titchener underwent atransformation from visual-auditory dominantwhile waking to kinesthetic dominant whiledrowsy.

Thus, one prediction of this older literaturewould be that a negative correlation shouldexist between waking imagery and hypna-gogic imagery. However, Holt's (1972) ques-tionnaire data show insignificantly positivecorrelations between hypnagogic imagery andboth waking thought imagery (r = .14) andimagery in daydreams (r = .19). The earlierliterature does not distinguish between dif-ferent types of waking mental imagery whenmaking comparisons to hypnagogic imagery.Recently, Holt (1972) and Starker (1974)have suggested that a distinction should bemade between active, volitional waking im-agery of the sort measured by the Betts scale(Sheehan, 1967) and passive, spontaneouswaking imagery as found in daydreams. Al-though the preliminary questionnaire data ofHolt do not show differential relations be-tween hypnagogic imagery and these twomodes of waking imagery, it does seem clearthat experimental investigations aimed atclarifying the relation between waking andhypnagogic imagery should bear the abovedistinction in mind.

Eidetic imagery. The comparisons betweeneidetic imagery and hypnagogic imagery aresparse. The only apparent similarity is that

4 Freedman and Marks (196S), in a photic stim-ulation experiment, found that the subjects whoresponded positively to the questionnaire item "Manypeople see quite vivid images when falling asleep orwaking up, usually (but not always) with their eyesclosed. Has this ever happened to you?" tended toreport a greater variety of visual forms in the photicstimulation situation than did the subjects whoanswered the question negatively.


both are reported to emerge frequently inchildhood. Gray and Gummerman (197S)concluded, after reviewing the literature oneidetic imagery, that "eidetic imagery differsfrom other types of visual imagery in degreeonly, not kind . . ." (p. 383). Viewed in thiscontext, there is little reason to expect thathypnagogic imagery would be correlated witheidetic imagery in a manner strikingly dif-ferent from the way in which it is correlatedwith normal waking imagery. Holt (1972)reported a correlation of —.27 between hyp-nagogic imagery and eidetic imagery as achild, which is the only quantitative assess-ment in the literature relating these twotypes of imagery.

Deja vu experiences. Though deja vu is not,strictly speaking, a type of imagery, deja vuand hypnagogic imagery are linked in severalinvestigations. Buck and Geers (1967), in aquestionnaire administered to introductorycollege students, found significantly positivecorrelations between frequency of occurrenceof hypnagogic imagery and frequency of oc-currence of visual (r = .35) and auditory (r= .27) deja vu. Similarly, Holt (1972) founda near significant ( r= .32) correlation be-tween hypnagogic imagery and "miscellaneousexperiences"—deja vu and synesthesia. Ellis(1897), in an article entitled "A note on hyp-nagogic paramnesia," claimed that the experi-ence of deja vu is common in the hypnagogicstate. Isokower (1938) called attention tothe similarity of feelings accompanying dejavu episodes and feelings arising during the"oral-cathexis" state mentioned earlier: "Thesubject knows or thinks he knows, exactlywhat is going to happen" (p. 336). One doesnot quite know what to make of these ob-servations, but studies exploring this issue arecalled for.

Quality of Thought in the Hypnagogic State

Bertini et al. (1964), Critchley ( 1 9 5 5 ) ,Foulkes and Vogel (196S), Green et al. (Note1), Hollingworth (1911), Myers (19S7), Os-wald (1962), Rapaport (1967a, 1967b), andVan Dusen (1972) have all observed that asone moves deeper into the hypnagogic state,thought and ideation become increasingly

bizarre, less amenable to conscious control,and reflect looser patterns of association.Unusual verbal combinations which are "fil-tered out" of waking awareness appearabundantly in the hypnagogic state. The fol-lowing illustration from the notebooks ofClark Hull (1962, p. 820) is typical:"Bill Hambra—Ju [sic] know him?"Note: A moment ago while passing in a mild lapseof attention, the above came into my mind withgreat distinctness. It seemed almost as if I werespeaking it. ... I have noticed this just as clearlyas the above many times while half asleep in bedbut never had paper handy and so never wrote itdown . . .

Question: Where in the world could that namehave come from? I haven't the slightest recollectionof anything like it.

This human machine is a queer thing!

Strikingly similar types of constructionsare reported by a number of observers.Archer (1935, p. 33) presented hypnagogicconstructions taken from his own experiencesuch as "A savoury pudding—raw in the mar-ket" and "A little management of Killie-krankie." Similarly, Oswald (1962) offeredexamples of hypnagogic thoughts such as "Heis as good as cake double" and "The pencilholds well. To the pavement with tell too."Froeschels (1949) studied this aspect of thehypnagogic state and postulated that rules ofassociation radically different from the rulesof the waking state govern the formation ofthought in the hypnagogic state. He concludedthat the unconscious plays a major role inhypnagogic thought:similarity seems to be the term that characterizesbest the basis upon which the subconscious works inthe state of transition. But this term evidently doesnot mean to the subconscious what it means to con-scious reasoning. The latter takes the feeling of simi-larity . . . for a stepping stone on the way tothorough differentiation and identification. The sub-conscious on the other hand frequently considerssimilarity identical with identity, and does notbother with further "research." (p. 24)

Rapaport (1967b), based upon processesobserved in his own hypnagogic state, postu-lated that in the hypnagogically constructedthought, the position normally given to whatis syntactically important in a sentence isreplaced by what is "psychologically out-standing" (p. 395). Although the phrase


psychologically outstanding is a rather vaguereferent, Rapaport's thesis is similar toFroeschels'; both saw qualitatively uniquerules governing the formation of thought inthe hypnagogic state.

Singer (1966) studied his own hypnagogicthought processes by setting up for himselfspecific mental tasks and observing whatkinds of intrusive thoughts appeared as hebecame drowsy. The self-assigned mentaltasks varied along three major bipolar dimen-sions: simple-complex, reminiscent-planful(or theoretical), and personal-impersonal.Thus, a simple-personal-reminiscent taskwould be to recall personal activities of thepast winter; a complex-impersonal-planfultask would be to consider the future optionsof a university program; and so on. Singer(1966) found that "the frequency of intrusivethoughts was by far the greatest for com-plex, impersonal, and planful or theoreticalmaterial. Reminiscing proved to be the easiesttask . . ." (p. 41). He also noted that per-sonal reference was predominant in the in-trusive thoughts. Further work of this sort inan experimentally controlled context would beuseful, as would a comparison of these pro-cesses with the same type of task performedin an alert state.

Several authors have noted the similaritybetween hypnagogic and schizophrenicthought processes (Froeschels, 1949; Holling-worth, 1911; McKellar, 1957, 1972; Mintz,1948; Oswald, 1962; Vogel et al., 1972).Maury (cited by Oswald, 1962, p. 112) con-cluded that associations in both cases arebased on sound. Oswald paraphrased Kraep-lin's conclusion that the disordered thoughtfound in hypnagogic and schizophrenic condi-tions could be attributed to "the encroach-ment of associations, the basic thought becom-ing displaced because of an associated ideareplacing an essential link in the normal chainof thinking" (p. 113). Mintz (1948) sug-gested that schizophrenic speech and sleepyspeech are similar in that groupings of ele-ments with only vague relations to one an-other characterize both modes of construc-tion. Vogel et al. (1972), discussing theirfinding of the shared regressivity properties ofREM and hypnagogic mentation in the con-

text of previously suggested REM-schizo-phrenia parallels, concluded that "an adequatetest of the hypothesis that the physiologic cor-relates of dreaming are those of schizophrenicmentation requires the use of physiologic cor-relates of SO [sleep onset] regressive men-tation" (p. 449).

Systematic experimental treatment of hyp-nagogic thought processes has not yetemerged. This may be partially due to thetendency to ascribe all hypnagogic phenomenato the imagery domain. Whatever the reason,further elucidation of the variables contribut-ing to the marked contrast in this mode ofcognition between hypnagogic and wakingconsciousness is needed. A possible strategyfor such study is suggested by the work ofSalzarulo and Cipolli (1974). These investi-gators subjected verbal material gatheredfrom dreams in different sleep stages to lin-guistic analysis and found semantical andsyntactical differences between REM andNREM verbal constructions.

The Autosymbolic Phenomenon

Closely related to the nature of hypnagogicthought processes is the autosymbolic phe-nomenon, which was first described in apaper by Silberer (1951) appearing in 1911and has since been reported by numerousobservers (Hollingworth, 1911; Isokower,1938; Lowy, 1946; Neisser, 1967; Rapaport,1967a; Singer, 1966; Slight, 1924; VanDusen, 1972; Vihvelin, 1948). The pro-cess that Silberer labeled the autosymbolicphenomenon consists of a transformation ofabstract mental contents into concrete visualimages during the hypnagogic state. The phe-nomenon is occasioned by the struggle of twoantagonistic forces: drowsiness and the effortto think.

Silberer delineated three principal varie-ties of autosymbolic phenomena. The ma-terial phenomena are representations of thecontent of thought as hypnagogic images. Forinstance, Silberer is thinking about advising afriend not to carry out a dangerous resolu-tion; suddenly, he sees "three gruesome rid-ers on black horses storming by over a duskyfield under leaden skies" (p. 203). Functional


phenomena are autosymbolic representationswhich reflect the nature of the operations ofconsciousness at the moment of the repre-sentation. Rapaport (1967a) found most ofhis own autosymbolic transformations to beof this variety, and noted that they oftenreflected the search for fading thoughts whilestruggling to remain awake—"watching some-one trying to find a way out of a labyrinth;seeing somebody going toward a door only todiscover that it is not a door but a shadow;following somebody's increasingly frantic ap-proach toward a door as the door slowlyshuts, and so on" (p. 643). Finally, somaticphenomena are autosymbolic representationsof physiological conditions such as body posi-tion, temperature, sensations, and the like.For example, Silberer takes a deep breath,feels his chest expand, and there appears ahypnagogic image of Silberer and anotherperson lifting a table high.

Van Dusen (1972) has argued that auto-symbolic representation is a defining char-acteristic of the hypnagogic state, positingthat "the region is naturally autosymbolic,that is, it represents where one is at themoment" (p. 95). He also noted that a degreeof familiarity with one's own hypnagogicstate may be necessary before one is aware ofspontaneously occurring autosymbolic phe-nomena. Experimental investigation of auto-symbolic processes has not yet been under-taken, although the Bertini et al. (1964)study might be viewed as a start in this di-rection. Foulkes and Vogel (1965) failed toobserve autosymbolic processes in their study,but noted that "we did not obtain the secondof the two conditions which Silberer felt nec-essary for its occurrence: drowsiness and aneffort to think" (p. 242). Clearly, experi-mental investigation of autosymbolic phe-nomena poses major methodological diffi-culties. What types of "stimuli" does one pre-sent to the subjects? How and when does oneelicit reports of internal activities? Most im-portantly, how does one discern if a particularstimulus has been autosymbolically trans-formed? These difficulties notwithstanding, Iconcur with Neisser (1967) that experi-mental investigation is sorely needed to clar-ify the nature of autosymbolic processes.

Receptivity and Suggestibility in theHypnagogic State

Converging evidence from several lines ofinvestigation indicates that the hypnagogicstate may be characterized as an unusuallyreceptive one. De Manaceine (1897) was in-terested in the degree of suggestibility main-tained in the hypnagogic state. She gave hersubjects two types of suggestions while theywere in transit from waking to sleeping:"intellectual suggestions", for example, twotimes three equals five, and "emotional sug-gestions," for example, "The building is burn-ing down" (pp. 195-223). The results showthat among children under 15 years of age,the intellectual suggestions were 85% suc-cessful and the emotional suggestions 97%successful; among adults, the intellectualvariety were 25% successful and the emo-tional ones 45% successful. A suggestion wasconsidered successful if the person repeatedthe experimenter's statement or accepted itwithout protest. No baseline measures weretaken (How suggestible are they whileawake?), nor was detailed information givenregarding the experimental conditions, sothese data should be cautiously interpreted.Rosett (1939) posited that the drowsy stateis especially conducive to uncritical accept-ance of suggestions but did not offer empiri-cal support for his hypothesis. More recently,Barber (1957) suggested that subjects in adrowsy state are just as open to suggestion asare hypnotized subjects.

Several of the early authors have noted thatexternal stimuli are readily incorporated inthe hypnagogic state. Hollingworth (1911)offered an example of music being trans-formed into a hypnagogic image at a concertduring which he had become drowsy. Varen-donck (1921) implicated external stimuli asbeing the primary contributors to the genesisof hypnagogic mentation; however, the condi-tions under which Varendonck made most ofhis observations—-the trenches of World WarI—might be seen as biasing factors. Segal(1972), on the basis of her investigations ofthe Perky phenomenon, has concluded that"any currently available sensory input may berecruited toward the formation of the image"(p. 229). It would be interesting to know if


this phenomenon is accentuated in the hypna-gogic state, as would be predicted by thisolder literature.

In an experimental context, Ornitz, Ritvo,Carr, LaFranchi, and Walter (1967) studiedthe cortical averaged evoked response (AER)to auditory stimuli during wakefulness, sleeponset, and the nocturnal sleep stages. Theyfound a marked increment in the amplitudeof the second negative deflection (N2) of theevoked waveform in the few minutes preced-ing and just after sleep onset. Ornitz et al.noted that this amplitude increment occurredat the time associated with hypnagogic dream-ing, and concluded that there is "increasedcerebral responsiveness to auditory stimuli inthe environment at this time" (p. 340). Fruh-storfer and Bergstrom (1969) examined theAER to auditory stimuli at different levels ofvigilance. They too reported that the N2 com-ponent of the evoked potential shows amarked amplitude increase just before sleeponset, and further noted that other com-ponents of the AER show an amplitude de-crease at this time. Fruhstorfer, Partanen,and Lumio (1971) found enhanced N2 ampli-tudes in response to auditory stimuli duringthe hypnagogic interval, and also reportedthat spontaneous vertex potentials occurringjust before sleep onset are systematicallyrelated to heart action. Fruhstorfer et al. con-cluded that "in the transition from wakeful-ness to sleep there are epochs of enhancedcortical responsiveness during which intero-ceptor stimulation also may evoke large vertexresponses" (p. 616).

The controversial field of hypnopaedia pro-vides an additional source of information re-garding the incorporative nature of the hypna-gogic state. Several researchers have foundthat the drowsy state seems to be the mosteffective period for sleep learning (Rubin,1968). Simon and Emmons (19S6) did notfind evidence for sleep learning in the deeperstages of sleep, but did report encouragingresults in the drowsy state, and postulatedthatit may be in the drowsy state preceding sleep, theindividual is more susceptible to suggestion; perhapsone's attitudes or habits can be modified during thispresleep period when criticalness is minimized, (p.96)

For a fuller discussion of the ramifications ofthis topic, see Budzynski (1972).

An account of the receptive nature of thehypnagogic state would not be complete with-out reference to the parapsychological litera-ture. Gurney et al. (1886), in their censusof hallucinations, found that a disproportion-ately large number of their veridical cases oftelepathy occurred while the percipient was inthe "borderland" state between waking andsleeping (Vol. 1, pp. 389-456). They rea-soned that since such a small amount of timeper day is spent in this state and such a largenumber of paranormal episodes result fromit, further study of the borderland statewould prove fruitful for parapsychology.White (1964) reviewed methods of ESP re-sponse employed by several gifted percipients.It is evident from a reading of these accountsthat a number of these gifted percipients en-tered a hypnagogic state during their ESPefforts. Recent experimental studies conductedby several investigators have explored the ESPperformance of subjects who attempted toenter the hypnagogic state. Honorton and hisassociates (Honorton, Drucker, & Hermon,1973; Honorton & Harper, 1974) had theirsubjects attempt to enter a hypnagogic stateby means of ganzfeld procedures while aremotely located agent viewed a randomlyselected target slide. The subjects were suc-cessful, to a statistically significant degree,in correctly identifying the target out of apool of slides on the basis of their hypna-gogic mentation. Braud and Braud (Note 3),also using the ganzfeld and a similar free-response-forced-choice design, reported signifi-cant ESP effects. Schacter and Kelly (197S),in a pilot study, used a biofeedback pro-cedure to induce the hypnagogic state in twopercipients. They found statistically signifi-cant (p < .007) ESP effects in one percipientand nonsignificantly positive results in theother, using a similar design.

The material presented in this section in-dicates that the hypnagogic state may becharacterized by a high degree of suggesti-bility, receptivity, and responsivity. However,the literature reviewed here leaves importantquestions open. For instance, there has beenno research clarifying the range and form of


individual differences in hypnagogic receptiv-ity, how this is related to personality, cogni-tive, and perceptual variables, the effect ofsystematically varied stimuli, and so on. Also,it is not clear from the cited studies whichcortical mechanisms underlie the observedchanges in the AER. Moreover, much of theevidence which has emerged is tentative andneeds more substantial empirical support be-fore a clearer outline of receptivity and sug-gestibility in the hypnagogic state can beconstructed.

Personality Characteristics Associated withthe Hypnagogic State

Until recently, little information was avail-able linking personality traits and the hypna-gogic state. Foulkes and Vogel (1965) notedinformally that the subjects who manifestedlarge amounts of dreamlike hypnagogic ma-terial appeared less anxious and constrictedthan did the subjects whose mentation wasmore controlled and thoughtlike. Foulkes etal. (1966) systematically investigated therelation of personality variables to the emer-gence of hypnagogic dreamlike fantasy in 32subjects.

Awakenings were made at two points dur-ing sleep onset and at three different pointsduring the course of REM sleep. All reportsof mentation gathered upon each awakeningwere rated by judges in terms of aggressiveand sexual content, hedonic tone, and depthof sleep. Also, reports were rated on theDreamlike Fantasy Scale, and all subjectscompleted the California Psychological In-ventory (CPI) and the Thematic Appercep-tion Test (TAT). It was found that hypna-gogic Dreamlike Fantasy scores correlatedpositively and significantly with TAT results.The Dreamlike Fantasy scores of REM epi-sodes did not correlate with waking fantasyas measured by the TAT. Further, it wasfound that the CPI items which correlatedsignificantly with tightly controlled hypna-gogic episodes—rigidity, conventionality, in-tolerance—closely approximate the standardconception of the authoritarian syndrome.Conversely, those with less controlled hypna-gogic ideation tended to be more self-accept-

ing, tolerant, socially poised, and less rigidlyconforming.

These correlations led Foulkes et al. to con-clude that the emergence of hypnagogic ma-terial is associated with healthy, expansive,ego functioning. Budzynski (1972), discus-sing the Foulkes et al. findings, speculatedthat extended experience in entering the hyp-nagogic state might produce positive person-ality changes in those constricted individualsidentified with tightly controlled hypnagogicmentation. Further research is needed to as-certain the generalizability of the Foulkes etal. study and also to investigate the thera-peutic possibilities raised by Budzynski.

PHYSIOLOGICAL CORRELATES OF THEHYPNAGOGIC STATE

In this section, the discussion is limited to aconsideration of those studies which mostexplicitly link physiological indicators withhypnagogic phenomena. The more extensiveliterature on physiological parameters ofdrowsiness are not touched upon here, and thereader is referred to Kamiya (1961) andKleitman (1963) for excellent discussions ofthis topic. The results of EEG and EOG workare the focus here, as no other physiologicalindexes have received extensive systematictreatment in the hypnagogic literature. How-ever, it should be noted that Timmons,Salamy, Kamiya, and Girton (1972) foundthat EEG patterns characteristic of hypna-gogic phenomena were systematically ac-companied by specific abdominal-thoracicrespiratory patterns, suggesting that furtherstudy of this parameter will be useful. Also,the findings of Budzynski (1972) and Stoyva(1973) that a low level of frontalis muscleactivity may be related to the onset of thehypnagogic state require more extensive in-vestigation.

EEG CorrelatesDavis et al. (1937) were the earliest in-

vestigators to explore the relationship betweenthe EEG and sleep onset mentation. Davis etal. divided the progression from wakefulnessto sleep into three stages: the "A" stage,characterized by interrupted alpha patterns;


the "B" stage, characterized by low voltageactivity; and the "C" stage, signaled by theappearance of 14-Hz spindles, plus some deltaactivity. A total of 14 subjects were used in aseries of experiments designed to differentiatebetween these stages. Monopolar recordingswere taken from the vertex and occiput, ref-erenced to the mastoid. Sessions were held inthe afternoon and at night. The subjects wereinstructed to relax and to signal by squeezinga bulb when they realized that they had just"drifted" or "floated" for a moment.

Several correlations between EEC patternsand signals from the subjects were specified.The authors noted that the relations wereclearest in the subjects with alpha-dominatedEEGs, that is, 1Q% or more alpha in theirEEC. In all, 165 episodes of floating werecollected from the alpha-dominant subjects;all but 6 of these incidents were preceded bya depression of the alpha wave which lastedfrom 1.5 to 30 or 40 seconds. The authors didnot specify whether this effect is found in bothvertex and occipital leads. However, the EECexcerpts presented indicate that the alphadepression is more pronounced in the occipitallead, though present in the vertex.

The subjects agreed on two points concern-ing the state of consciousness associated withthe signals of floating: (a) the episodes werenot "real sleep," but an intermediate stage;and (b) when a signal was given, there hadbeen a depression of awareness. It is clearfrom the few protocols which are presentedthat the subjects were experiencing hypna-gogic phenomena in these periods of floating.One subject noted that "these things arepractically dreams, but I am awake enough tocatch them" (p. 32). Although little that isprecise can be said relating the EEC changesto particular properties of hypnagogic phe-nomena, the Davis et al. study provides thefirst general picture of what the EEC lookslike in hypnagogic episodes.

Dement and Kleitman (19S7) were inter-ested in the type of mental content to befound in the hypnagogic state, largely as itcompared to REM dreams. The EEGs wererecorded monopolarly from the frontal, parie-tal, and occipital leads, referenced to the ears.Dement and Kleitman named the sleep onset

period "descending stage one," and noted that"a low voltage, irregular, relatively fast pat-tern containing slow components in the 4-6c/sec. range similar to the Loomis et al. 'B'stage characterizes sleep onset EEGs" (p.678). Reports elicited in the presence ofdescending stage one EEG contained hypna-gogic activity, and the subjects additionallyindicated that they were not asleep duringthese episodes. Reports gathered from ascend-ing stage one (REMs) yielded full dreamsand the feeling that the person had emergedfrom deep sleep.

Foulkes and Vogel (1965) studied the psy-chophysiology of the hypnagogic state insomewhat more detail. Their study was com-prised of nine subjects, who contributed foursessions each. The sessions were scheduled atthe subjects' normal bedtime and typicallylasted 2-4 hours apiece. Standard Dement andKleitman (1957) electrode placements wereused. Foulkes and Vogel divided the hypna-gogic state into four stages: (a) continuousalpha, sporadic rapid eye movements; (b)discontinuous alpha, slow eye movements; (c)low-voltage descending stage one EEG, sloweye movements; and (d) descending stagetwo EEG.

The experimenter interrupted the subjectan approximately equal number of times ineach of these stages to obtain a report ofmental content. The subject first gave a spon-taneous description of what he had experi-enced, and then responded to questions con-cerning degree of alertness, volitional controlover mental content, awareness of the en-vironment, and the like.

Foulkes and Vogel reported several corre-lations between EEG-EOG stages and mentalcontent. A steady decline in volitional controlover ideation and in awareness of the immedi-ate surroundings was observed over successiveEEG-EOG stages. Additionally, an increasein dreamlike content was observed with suc-cessive stages. It was also found that whilevisual imagery was most frequently reported,no consistent association of imagery modalityand EEG-EOG stage was observed. (Theauthors did not look for associations of im-agery modality and activity in appropriateEEG leads; this potentially fruitful strategy


has not yet been employed.) In a later re-interpretation of this data, Vogel et al. (1966)further specified the correlations betweenEEG stages and mental content. Alpha-domi-nant EEC was associated with nonregressivecontent (ego intact), descending stage oneEEG was associated with regressive content(ego destructuralized), and descending stagetwo EEG was associated with a return to non-regressive content (restructuralized ego). Theauthors noted that an ideal psychophysiologi-cal parallel was not always observed but thateven in cases in which there was a displace-ment of mental content in relation to EEGstages, the order of progression was preserved.

Liberson and Liberson (1965) denned theonset of drowsiness as corresponding to theonset of EEG occipital alpha blocking. Intheir study, EEG was recorded both mono-polarly and bipolarly from vertex-ear andparieto-occipital leads. Samples of mental con-tent were obtained by asking the subjects topress a key as rapidly as possible when anauditory signal was heard and to report men-tal content at that time. Liberson and Liber-son, much like Davis et al. (1937) andFoulkes and Vogel (1965), found that presentoriented reports occur exclusively in alpha-intact segments and that the frequency ofvague statements and hypnagogic images re-ported increases through progressive stages ofdrowsiness.

Liberson and Liberson also found a delayin the recognition of drowsiness; that is, only50% of their subjects signaled the presenceof hypnagogic images or other indicators ofdrowsiness during the first 30 seconds afterthe onset of occipital alpha blocking. Theytermed this phenomenon "paradoxical drowsi-ness." However, this finding is paradoxicalonly if the onset of occipital alpha blockingis taken to be a foolproof indicator of drowsi-ness. Liberson and Liberson noted that dif-fuse theta activity, a reliable indicator ofdrowsiness, often does not appear until wellafter the onset of occipital alpha blocking.This suggests that the onset of occipital alphablocking may be a valid indicator of the on-set of drowsiness, or the hypnagogic state, inonly a limited portion of the population.

Green et al. (Note 1), in the biofeedbackstudy discussed earlier, linked the occurrenceof hypnagogic phenomena to the presence oflow-voltage theta waves in the EEG. Noquantitative analyses clarifying this proposedrelationship have yet been presented. Simi-larly, Budzynski (1972) and Stoyva (1973)reported that the hypnagogic state is char-acterized by increased theta activity butfailed to offer analyses demonstrating the pre-cise relationship between hypnagogic phe-nomena and the EEG.

The evidence accumulated in the abovestudies suggests that the gross EEG character-istics of the hypnagogic state are known. Itappears reasonable to hypothesize that thelow-voltage activity of Davis et al., the de-scending stage one of Dement and Kleitman,(and Foulkes and Vogel) and the diffuse,low-voltage theta activity of Liberson andLiberson, Green et al., Budzynski, and Stoyvaare all descriptively different ways of char-acterizing a similar phenomenon. In all cases,a loss of amplitude, frequency, and synchro-nous repetition describe the EEG which ac-companies the hypnagogic state. However, itshould be pointed out that in addition to theslow activity which is reported to replace thealpha rhythm, Dement and Kleitman (andFoulkes and Vogel) also found that fast com-ponents in the beta (14-21 Hz) range ac-company hypnagogic episodes. In my ownpilot work (Note 4), I have found that abasic wave in the theta range, with consider-able low-voltage fast activity superimposed onit, characterizes hypnagogic episodes. Theanalysis techniques employed in the citedstudies do not allow a precise profile of theEEG to be constructed; thus, it is not clearwhat the proportion of slow to fast compo-nents in the hypnagogic EEG signifies.

Eye MovementsThe first hypothesis linking eye movements

with hypnagogic activity was offered by Ladd(1892), on the basis of systematic self-obser-vation. Miles (1929) reported experimentalevidence of slow eye movements occurring indrowsy states. Miles was interested in theways in which saccadic eye movements differin waking and drowsy states. The subjects


were instructed to alternately fixate two dotsseparated by 40 degrees visual angle. Milesthen photographed the eye movements andfound them to become much slower and lessaccurate as drowsiness set in. However, mostinteresting for the purposes of this paper isa phenomenon "accidentally" observed byMiles:In two records which are reproduced, a subject,although motivated to remain awake, went to sleepduring the eye movement test. The eyes remainedopen for a few seconds, the typical saccadic move-ments which had been very slow now changed intoa rolling movement gliding back and forth as if fol-lowing a slow pendulum, and all evidence of fixa-tion disappeared, (p. 140)

Although Miles' observation does not spe-cifically relate the eye movements to hypna-gogic phenomena, the implication is clear.

The occurrence of slow, rolling eye move-ments at the onset of sleep was subsequentlyaffirmed by Aserinsky and Kleitman (19SS).These investigators used the electrooculo-graph to record eye movements and foundlateral, "pendular" slow eye movements(SEMs) present at sleep onset. These SEMswere typically completed in 3-4 seconds andwere frequently binocularly asymmetrical.The SEMs were contrasted to REMs, whichare jerky, binocularly symmetrical, and exe-cuted in fractions of a second. Dement andKleitman (1957) specifically noted that noREMs were to be found at sleep onset. (Anexception to this general rule is found innarcoleptics, who frequently manifest REMsat sleep onset.5 See Liddon, 1967, and Recht-schaffen, Wolpert, Dement, Mitchell, andFisher, 1963.) Foulkes and Vogel (1965), asnoted earlier, classified the stages in the tran-sition from wakefulness to sleep on the basisof EEG-EOG characteristics. The REMs(not of the nocturnal variety) accompaniedthe intact alpha stage; SEMs covered the20-30 seconds of the record prior to the ap-pearance of discontinuous alpha; SEMs al-most always accompanied descending stageone EEG, and were "usually" found in de-scending stage two episodes.

Liberson and Liberson (1965) linked theSEMs to drowsy episodes as manifest in theEEG. They observed that SEMs started acertain time prior to the onset of occipital

alpha blocking; 50% of their subjects mani-fest SEMs approximately 5 seconds beforealpha blocking, while SEMs appeared in 33%of their subjects 15 seconds prior to -blocking.It was also observed that SEMs increasedduring the first 15 seconds of drowsiness andthen gradually subsided until they had dis-appeared entirely by the time that spindleactivity was initiated. Additionally, Libersonand Liberson found that SEM activity per-sisted between interrupted drowsy episodes,when the EEG indicates an alert state. Thissuggests that the EOG may be a more sensi-tive measure of drowsiness than the EEG.However, since subjective accounts are notlinked to the persisting SEM periods, thisquestion cannot be answered on the basis ofthe Liberson and Liberson data.

Dement (1965) has argued that the pres-ence of SEMs is an extremely accurate gaugeof the presence or absence of hypnagogic ma-terial. Dement interrupted subjects accord-ing to the presence or absence of SEMs. WhenSEMs were absent, the subjects invariablyclaimed to be awake and "thinking." How-ever, in the presence of SEMs, the subjectsconsistently reported hypnagogic images anddecreased awareness of the environment,though their EEG remained unchanged. Thisobservation lends support to the tentativeinterpretation of the Liberson and Libersondata offered above. Dement further observeda consistent tendency for greater subjectiveduration and complexity of hypnagogic im-agery to be reflected by a correspondinglylonger SEM period. He noted that the slowside-to-side swings found in SEM episodesare unrelated to the reported imagery; unlikeREM dreams, appropriate scanning move-ments are not manifest during SEM hypna-gogic activity. Oswald (1962) has suggestedthat a person experiencing hypnagogic imagesdoes not manifest appropriate sequential ocu-lar responses to them "because he quicklyand repeatedly re-establishes contact withreality after his brief fantasy creation" (p.99).

6 Hypnagogic images are part of the narcoleptictetrad, which also includes narcolepsy, cataplexy, andsleep paralysis.


The evidence adduced in this section sug-gests that eye movements may be at least assensitive an indicator of hypnagogic activityas the EEC. Although most of the studieslink SEMs with a general drowsy-hypnagogicstate, the Dement work shows that SEMs arepotentially useful tools for discriminating be-tween episodes on the basis of content. Addi-tionally, Dement has suggested that the lackof appropriate scanning movements in hypna-gogic imagery, in contrast to REM dreams,indicates that eye movements may be helpfulin understanding differences between the twomodes of ideation.

Neurophysiological MechanismsThere is little to be found in the literature

concerning neurophysiologic mechanisms un-derlying the hypnagogic state. Oswald postu-lated that hypnagogic phenomena are attribu-table to a decline in "cortical vigilance." Blissand Clark (1962) hypothesized that hypna-gogic phenomena "represent the way the mindacts in the twilight zone between alertnessand deep sleep, and reflect the dominance ofthe visual apparatus at that time" (p. 105).Kluver (1942) posited that fundamentalmechanisms involving various levels of thenervous system are responsible for hypna-gogic, mescaline, and other similar phenom-ena.

The most ambitious attempt to specify pos-sible neurophysiological underpinnings of thehypnagogic state is found in West's (1962)theory of hallucinations. The "perceptualrelease" theory of hallucinations rests on twobasic assumptions: (a) Life experiences affectthe brain in such a manner as to leave perma-nent neural traces, engrams, templates, andthe like, which form circuits that subserve theneurophysiology of memory, thought, andimagination, (b) In normal wakefulness, theinterplay of dynamic forces which characterizethe individual exert an organizing and inte-grating force upon these traces.

West cited work on the ascending reticularactivating system (ARAS) which implicatesthis brain stem structure as being responsiblefor the maintenance of the wakeful state, aswell as for exerting the integrating and inhibi-tory effects on consciousness which enable the

waking organism to reduce the amount of in-put impinging upon it to a manageable level.He then specified the two conditions necessaryfor hallucinatory activity: (a) Reduction orimpairment of sensory input to the brain, and(b) maintenance of a state of "residual aware-ness." The crux of the theory is that underconditions of impaired sensory input, theARAS no longer receives the stimulationnecessary to exert inhibitory effects on con-sciousness; perceptual traces normally in-hibited are now released. In order that theperson experience these released traces, aminimal degree of awareness must be main-tained.

In direct relation to the hypnagogic state,West noted that the transition from wake-fulness to sleep is accompanied by a reductionin sensory input, which impairs the inhibitoryfunction of the ARAS and permits the releaseof perceptual traces. Additionally, a suffi-cient degree of cortical arousal (descendingstage one) is maintained in the hypnagogicstate to permit awareness of these traces;thus occur the hypnagogic phenomena. Westalso observed that the level of stage one cor-tical arousal occurs in only one other phaseof sleep—the REM stage. He posited that onemay be "dreaming" throughout sleep but thatthe degree of cortical arousal needed to per-mit awareness of released perceptual tracesoccurs only in descending and ascending stageone episodes. No experimental evidence link-ing West's theory with hypnagogic phenom-ena has yet emerged; thus, his postulatesmust be regarded as tentative at this time.

THEORETICAL FORMULATIONS

As noted in the introduction, the literatureon hypnagogic phenomena is underdevelopedwith respect to theoretical systems. Most ofthe explanations which have been offered, ata psychological level, are typically directed toa small segment of the phenomena. In thissection, three ways of explaining the genesisand function of hypnagogic phenomena arepresented. These are not the only existingtheories, but they do represent the majormodes of thinking about hypnagogic phenom-ena found in the literature.


The Entoptic Explanation

Most writers in the hypnagogic literaturehave made at least cursory reference to therole of entoptic phenomena in hypnagogicmatters. A number of the early writers(Gruithusien, cited in Leaning, 1926; Hicks,1924; Maury, 1848; Miiller, 1848; Purkinjie,cited in Leaning, 1926) considered entopticphenomena to be of paramount importancein the genesis of hypnagogic images. Ladd(1892) and McKellar and Simpson (1954)offered the most developed form of this argu-ment. The main point made by Ladd andMcKellar and Simpson is that the backgroundof "retinal dust" which appears upon closingthe eyes is colorful, characterized by novelforms and shapes, ever changing, and whollyoutside the realm of voluntary control. Theseproperties of the entoptic lights explain threequalities consistently found in hypnagogicimages: (a) the seemingly "original" char-acter of the forms, shapes, colors, and so on;(b) their fast rate of change and seeminglyvoluntary behavior; and (c) the unrecogniza-bility of the images to the observer. Ladd(1892), reflecting upon his own experience,averred that "by far the purest, most brilliant,and most beautiful colors I have ever seen,and the most astonishing combinations ofsuch colors, have appeared to me with closedeyes in a dark room" (p. 300). Ladd went onto say that careful introspection shows thathypnagogic images are no more than combi-nations of the entoptic lights.

Clearly, entoptic phenomena do play a rolein the genesis of visual hypnagogic images;at this time the entoptic background furnishesthe whole of one's visual field. However, thereare myriad phenomena which a totally en-toptic theory cannot explain. Alexander(1909) noted that a purely retinal explana-tion could not account for the existence ofauditory images. Similarly, the quality of theverbal constructions, kinesthetic representa-tions, and autosymbolic phenomena discussedearlier do not fit a purely retinal explanation.Finally, a small bit of empirical evidencemilitates against acceptance of the entoptictheory. Holt (1972) found almost no correla-tion between awareness of phosphenes andincidence of hypnagogic phenomena in his

sample; an entoptic theory predicts that thosemost aware of phosphenes should reportgreater incidences of hypnagogic phenomena.

The "Creative Unconscious" Theory

This notion stands in direct opposition tothe entoptic theory and embodies the ideathat hypnagogic phenomena are manifesta-tions of processes blocked from awareness inthe waking state. Alexander (1909) and Lean-ing (1926) are the principal exponents ofthis theory, although the idea is implicit inthe work of Froeschels (1949), Green et al.(1970, Note 1), and Van Dusen (1972).

Interestingly, it is precisely those character-istics of hypnagogic mentation (vividness,originality, autonomy) which the entoptictheorists use as a basis for their ideas that the"unconscious" theorists (whose postulates aredirectly opposite to those of the entopticgroup) use as evidence of their formulations.Alexander, noting all of the above qualities ofhypnagogic phenomena, posited that "theseimages are the work of an agent that does notshare in my interests, aims, or feelings" (p.631). Some years earlier, Herschels (cited inLeaning, 1926) had remarked that hypna-gogic phenomena are manifestations of"a thought, an intelligence, working within ourown organization distinct from that of ourown personality" (p. 391), Green et al. (Note1) recently suggested that "usually 'unheardthings' . . . come to consciousness in the formof hypnagogic imagery" (p. 22). Van Dusen(1972) emphasized that a dialogue betweenconscious and unconscious processes character-izes the hypnagogic state.

Although the concept of the unconscioushas fallen on hard times in modern experi-mental psychology, there is abundant evidencethat mechanisms either not operative or notusually manifest in waking consciousness con-tribute to the formation of hypnagogic phe-nomena. For instance, in the autosymbolicprocess discussed earlier, the brain appears tobe making qualitatively different responses tointernal stimuli than are manifest in wakingconsciousness. Further, there are numerousreports of creative episodes emerging in thehypnagogic state. The most famous of theseconcerns the chemist Kekule, who derived the


idea of the benzene ring from a hypnagogicimage of a snake biting its own tail whichappeared to him while drowsy (in Koestler,1964, p. 118). Hollingworth (1911) notedthat Mark Twain, Edgar Allen Poe, and Rob-ert Louis Stevenson all appear to have capi-talized upon the "creative" aspects of spon-taneously appearing imagery while drowsy, byincorporating hypnagogic phenomena intotheir work. He also cited a study by Marsh(1906), who found that artistic-literary typesdid their most creative work in the earlymorning and late evening hours, while scien-tific types worked more during midday hours.Varendonck (1921), commenting on his ownexperiences, noted that "for years I used towonder why I seemed cleverer in bed thanout of it; the few ideas which I imagine my-self to have conceived as my original contri-bution to science having come just beforesleep" (p. 26). Green et al. (Note 1) positedthat hypnagogic imagery and creativity arerelated, and one of the major goals of theirresearch is to enhance creative ability bydeveloping awareness of hypnagogic imagery.

If one insists upon a literal translation ofthe idea championed by Alexander, that "anagent foreign to the self" is "responsible" forhypnagogic phenomena, then the creative un-conscious theory must be regarded with ampleskepticism. However, the basic idea of thetheory is captured in the argument presentedby Leroy (1926) and Rapaport (1967a,1967b): In the hypnagogic state, there is aqualitatively unique synthesis of mental con-tent. While the mode in which the creativeunconscious theory was originally concep-tualized does appear to be excessively anthro-pomorphic, the observations which led to theformulation of the theory are still in need ofan adequate explanation.

The Ego-Regressive Theory

The notion that the hypnagogic state ischaracterized by regressive tendencies of theego has been espoused by Isokower (1938),Lewin (1949), Rosett (1939), and most sys-tematically by Vogel et al. (1966), whosetheory is examined here.

Vogel et al. (1966) averred that in thehypnagogic transition, a person's relation to

the external world can be characterized bytwo major processes: (a) increasing with-drawal from the external world, marked byloss of volitional control over mentation, and(b) loss of reality testing. Additionally, threeego states were identified based on analysisof mental content: (a) I—intact ego, non-regressive content; (b) D—destructuralizedego, regressive content; and (c) R—restruc-turalized ego, nonregressive content.

It was observed that regressive content ap-pears only after some degree of withdrawalfrom the external world has begun and thata return to nonregressive content occurs onlyafter reality testing has been lost. In otherwords, the D state follows some degree ofwithdrawal; the R state follows a loss of real-ity testing. Vogel et al. explained this seriesof events by first positing that the ego re-quires sensory input to remain intact. Re-duced sensory input results in a withdrawalof ego-cathexes from the external world anda functionally regressed ego. The tendencytoward regression threatens the ego, engen-dering a need for defense. The observationthat a loss of reality testing precedes a returnto nonregressive content leads the authors topostulate that the loss of reality testing is theego's needed defense. It is noted that thisdefense persists into sleep, as the NREMdreams which characterize the interval fol-lowing the hypnagogic period are typicallymore thoughtlike than the more bizarre REMdreams. Vogel et al. concluded that

some hypnagogic dreams are instigated by with-drawal from the external world, and from a psycho-logical point of view, they appear to reflect primar-ily a regressed ego state rather than the rise anddisguised fulfillment of unconscious wishes which areexpressed by night (REM) dreams, (p. 248)

Support for this theory is found in severalsources. Kubie (1943) found that childhoodmemories not accessible in dreams surface inthe hypnagogic state, implying a state ofregression. Similarly, Bertini et al. (1964)and Green et al. (Note 1) have commentedupon the appearance of archaic childhoodmemories in their experimental-hypnagogicsubjects. Powerful support for an ego-involve-ment theory is found in the previously men-tioned correlations between personality and


hypnagogic dreaming reported by Foulkes etal. (1966). Their findings led them to con-clude that hypnagogic experience is a func-tion of waking ego strength and flexibility.However, this does not directly support a re-gression hypothesis. Finally, it should benoted that this theory is the only one of thethree considered here which is based on sys-tematic experimental investigation.

Two criticisms can be leveled at the experi-mental procedure through which Foulkes andVogel obtained the data germane to their the-ory. The first question is the validity of label-ing certain material regressive. Content waslabeled regressive if one or more of six quali-ties was found: (a) single, isolated images;(b) an incomplete scene or bits and pieces ofa scene; (c) bizarre, inappropriate, or dis-torted images; (d) bizarre sequence or super-position of the images; (e) dissociation ofthought and image; (f) magical, omnipotentthinking. Categories a-d (especially a and b)could be just as easily attributed to entopticphenomena or the inability of untrained ob-servers to fully observe their train of imagery.The point is that the lumping together ofmyriad qualities under the imprecise headingof regressive might not be justified undercloser analysis. A second possible flaw in theFoulkes and Vogel work concerns the inter-views conducted after each awakening. Thesame experimenter who does the wakeningalso does the interviewing; the possibility ofan experimenter effect influencing the courseof the interview cannot be excluded (Rosen-thai, 1966).

The interpretive approaches presented arecharacterized by several themes. First, theytend to emphasize one factor to the exclusionof all others. Kliiver (1942) exposed thistendency in relation to theories of hallucina-tion:

The theories that have been evolved for an under-standing of hallucinatory phenomena frequentlystressed one set of factors to the exclusion of oth-ers. They have stressed either psychological or physi-ological factors, "peripheral" or "central" factors,sensory or motor factors, cortical or subcorticalmechanisms. In recent years, however, it has beengenerally recognized that all these factors are of im-portance. Since all of them are involved in any com-plex behavioral reaction, it is to be expected thatthey are involved in hallucinations, (p. 175)

Similar considerations are relevant to hypna-gogic phenomena. The one-sidedness of hyp-nagogic theories reflects how little is knownabout the phenomena.

A second characteristic of the above ap-proaches is that theoretical biases and ante-cedent beliefs heavily influence subsequentinterpretive strategies. Alexander (1909) andLeaning (1926), writing from a parapsycho-logical perspective, postulated a "disem-bodied" agent to account for hypnagogic phe-nomena; Vogel et al. (1966), of a psycho-analytic persuasion, offered an ego-regressionhypothesis; and so on. Clearly this criticismis applicable to many sectors of psychologicalresearch. However, considering the extremelysuggestible and permeable nature of the hyp-nagogic state, it is entirely conceivable thatthe type of phenomena which emerge in thehypnagogic state may be a direct function ofthe matrix of beliefs and expectations whicha person holds concerning the anticipatednature of the experience. This point is bestdisplayed in an old criticism of psychoanalyticpractice: Freud's patients have Freudiandreams, Jung's patients have Jungian dreams.The possibility of inadvertently influencingsubjects to produce the desired phenomena isa problem which future experimental researchmust face. Indeed, experimental investigationof precisely this parameter of the hypnagogicstate would be useful.

CONCLUDING REMARKSIn this review, the literature concerning the

hypnagogic state has been presented in sev-eral perspectives. The theme common to thedifferent approaches is that all are in anearly stage of development. Richardson (1969,pp. 95-100) came to a similar conclusion andurged experimenters to get on with systemati-cally investigating the hypnagogic state.

On the basis of this review, several poten-tially valuable lines of research are suggested.First, further development of experimentaltechniques which permit systematic explora-tion of the hypnagogic state is critical to morerefined understanding; the major methodo-logical problems were outlined earlier. Second,a more precise phenomenology of the imageryand verbal constructions which emerge in the


hypnagogic state is needed. Can one categor-ize by content the phenomena which appearin particular individuals? Are the phenomenaconsistent over time? Can the phenomenologi-cal qualities which differentiate hypnagogicmentation from other sorts of imagery andthought be more clearly specified? These arebut a few of the phenomenological issues yetto be adequately researched. Third, cognitiveprocesses in the hypnagogic state merit moredetailed consideration. What is the organism'scapacity for processing environmental infor-mation while in the hypnagogic state? Howdoes prehypnagogic state sensory stimulationaffect subsequent hypnagogic mentation?Fourth, it is not clear how differences in con-tent and modality of mentation manifestedin the hypnagogic state are related to cogni-tive, perceptual, and personality variablesamong individuals. What type of person ex-periences much visual hypnagogic imageryand few verbal constructions? Do more cre-ative individuals tend to exhibit greateramounts of hypnagogic content? Fifth, thepsychophysiological approach requires fullerdevelopment. Aside from the interest of ques-tions which can be explored psychophysio-logically (Are there cerebral hemisphericasymmetries related to hypnagogic content?What is the relation between eye movementsand the EEC in hypnagogic episodes?), thisapproach also offers potent methodologicaladvantages for exploring the hypnagogicstate. Finally, specification of the conditionsunder which a person enters the hypnagogicstate is critical for future research. As notedearlier, it may be that the "natural" hypna-gogic state which spontaneously occurs justbefore sleep is characterized by patterns ofphenomena that are qualitatively differentfrom those that characterize a biofeedback-induced hypnagogic state; and both of thesemay differ from the hypnagogic interval whichfollows awakening from a sleep stage. Re-search which confronts this problem would beextremely valuable.

These possible directions for future re-search are certainly not exhaustive; but eluci-dation of these questions may prove useful inunderstanding the patterns of phenomenawhich characterize the hypnagogic state.

REFERENCE NOTES1. Green, A., Green, E., & Walters, D. Brainwave

training, imagery, creativity, and integrative ex-periences. Unpublished manuscript, The MenningerFoundation, 1973.

2. Owens, C. A study oj imagination. Unpublisheddoctoral dissertation, University of Liverpool, 1963.

3. Braud, L. W., & Braud, W. G. The psi conducivesyndrome: Free response GESP performance dur-ing an experimental hypnagogic state induced byvisual and acoustic ganzfeld technique. Paperpresented at the annual meeting of the Para-psychological Association, Queens, New York,August 1974.

4. Schacter, D. L. Unpublished data, August 1974.

REFERENCESAlexander, H. E. The subconscious in light of dream

imagery. Proceedings of the Society for PsychicalResearch, 1909, 8, 623-634.

Archer, W. On dreams. London: Metheun, 1935.Ardis, J., & McKellar, P. Hypnagogic imagery and

mescaline. Journal oj Menial Science, 1956, 102,22-29.

Aserinsky, E., & Kleitman, N. Two types of ocularmotility occurring in sleep. Journal of AppliedPhysiology, 1955, 8, 1-14.

Barber, T. X. Experiments in hypnosis. ScientificAmerican, 1957, 196W, 54-61.

Bertini, M., Lewis, H. B., & Witkin, H. A. Somepreliminary observations with an experimentalprocedure for the study of hypnagogic and relatedphenomena. Archivio di Psicologia Neurologia ePsychiatria, 1964, 6, 493-534. (Also printed, incondensed form, as a chapter in C. T. Tart (Ed.),Altered states of consciousness. New York: Wiley,1969.)

Bexton, W. H., Heron, W., & Scott, T. H. Effects ofdecreased variation in the sensory environment.Canadian Journal of Psychology, 1954, 8, 70-76.

Bliss, E. L., & Clark, L. D. Visual hallucinations.In L. J. West (Ed.), Hallucinations. New York:Grune & Stratton, 1962.

Buck, L. A., & Geers, M. B. Varieties of conscious-ness: I. Intercorrelations. Journal of Clinical Psy-chology, 1967, 23, 151-152.

Budzynski, T. H. Some applications of biofeedbackinduced twilight states. Fields Within Fields WithinFields, 1972, 5, 105-114.

Budzynski, T. H., & Stoyva, J. M. An instrumentfor producing deep muscle relaxation by means ofanalog information feedback. Journal oj AppliedBehavior Analysis, 1969, 2, 231-237.

Budzynski, T. H., Stoyva, J. M., & Adler, C. S.Feedback-induced muscle relaxation: Application totension headache. Behavior Therapy and, Experi-mental Psychiatry, 1970, 1, 205-211.

Caldwell, W. V. LSD psychotherapy. New York:Grove Press, 1968.

Critchley, M. The pre-dormitum. Revue Neuro-logique, 1955, 93, 101-106.

HVPNAGOGIC STATE 479

Davis, H., Davis, P. A., Loomis, A. I,., Harvey, E.N., & Hobart, G. Human brain potentials at theonset of sleep. Journal of Neurophysiology, 1937,I, 24-37.

Dement, W. C. Perception during sleep. In P. M.Hoch & J. Zubin (Eds.), Psycho pathology ofperception. New York: Grune & Stratton, 1965.

Dement, W. C., & Kleitman, N. Cyclic variations inEEG during sleep and their relation to eye move-ments, body motility, and dreaming. Electroen-cephalography and Clinical Neurophysiology, 1957,9, 673-690.

DeStrooper, J., & Broughton, R. REM awakeninglatencies and a possible REM breakthrough phe-nomenon. Psychophysiology, 1969, 6, 216.

Ellis, H. A note on hypnagogic paramnesia. Mind,1897, 6, 283-287.

Ellis, H. The world of dreams. New York: Consta-ble, 1911.

Fiss, H., Klein, G. S., & Bokert, E. Waking fantasiesfollowing interruption of two types of sleep. Ar-chives of General Psychiatry, 1966, 14, 543-551.

Foulkes, D. Dream reports from different stages ofsleep. Unpublished doctoral dissertation, Universityof Chicago, 1960.

Foulkes, D. Dream reports from different stages ofsleep. Journal of Abnormal and Social Psychology,1962, 65, 14-25.

Foulkes, D. The psychology of sleep. New York:Scribner's, 1966.

Foulkes, D. Longitudinal studies of dreams in chil-dren. In J. H. Masserman (Ed.), Science andpsychoanalysis, 1971, 19, 48-71.

Foulkes, D., Spear, P. S., & Symonds, J. D. Indi-vidual differences in mental activity at sleep onset.Journal of Abnormal Psychology, 1966, 71, 280-286.

Foulkes, D., & Vogel, G. Mental activity at sleep on-set. Journal of Abnormal Psychology, 1965, 70,231-243.

Freedman, S. J., & Greenblatt, M. Studies in humanisolation: II. Hallucinations and other cognitivefindings. U.S. Armed Forces Medical Journal, I960,II, 1479-1497.

Freedman, S. J., Grunebaum, S. U., Stare, F. A., &Greenblatt, M. Imagery in sensory deprivation. InL. J. West (Ed.), Hallucinations. New York:Grune & Stratton, 1962.

Freedman, S. J., & Marks, P. A. Visual imageryproduced by rhythmic photic stimulation: Person-ality correlates and phenomenology. British Jour-nal of Psychology, 1965, 56, 9S-112.

Froeschels, E. A peculiar intermediary state betweenwaking and sleeping. American Journal of Psycho-therapy, 1949, 3, 19-25.

Fruhstorfer, H., & Bergstrom, R. M. Human vigi-lance and auditory evoked responses. Electroen-cephalography and Clinical Neurophysiology, 1969,27, 346-355.

Fruhstorfer, H., Partanen, J., & Lumio, J. Vertexsharp waves and heart action during the onset ofsleep. Electroencephalography and Clinical Neuro-physiology, 1971, 31, 614-617.

Gallon, F. Inquiries into the human faculty and itsdevelopment. London: Macmillan, 1883.

Giora, Z., & Elam, Z. What a dream is. BritishJournal of Medical Psychology, 1974, 47, 283-289.

Goldberger, L., & Holt, R. R. Experimental inter-ference with reality contact (perceptual isolation):Method and group results. Journal of Nervous andMental Disease, 1958, 127, 99-112.

Goodenough, D. R., Lewis, H. B., Shapiro, A.,Jaret, L., & Sleser, I. Dream reporting followingabrupt and gradual awakenings from differenttypes of sleep. Journal of Personality and SocialPsychology, 1965, 2, 170-179.

Gray, C. R., & Gummerman, K. The enigmaticeidetic image: A critical examination of methods,data, and theories. Psychological Bulletin, 1975,82, 383-407.

Green, E., Green, A., & Walters, D. Voluntary con-trol of internal states: Psychological and physio-logical. Journal of Transpersonal Psychology,1970, 1, 1-26.

Gurney, E., Myers, F. W. H., & Podmore, F.Phantasms of the living (Vol. 1). London: Trub-ner, 1886.

Hanawalt, N. G. Recurrent images. American Jour-nal of Psychology, 1954, 67, 170-174.

Hebb, D. O. Concerning imagery. PsychologicalReview, 1968, 75, 466-477.

Hicks, G. D. On the nature of images. BritishJournal of Psychology, 1924, IS, 121-148.

Hilgard, E. R. Altered states of awareness. Journalof Nervous and Mental Disease, 1969, 149, 68-79.

Hobbes, T. Leviathan, London: A. Crooke, 1651.Hollingworth, H. L. The psychology of drowsiness.

American Journal of Psychology, 1911, 22, 99-111.Holt, R. R. Imagery: The return of the ostracized.

American Psychologist, 1964, 19, 254-264.Holt, R. R. On the nature and generality of mental

imagery. In P. Sheehan (Ed.), The junction andnature of imagery. New York: Academic Press,1972.

Holt, R. R., & Goldberger, L. Assessment of indi-vidual resistance to sensory alteration. In B. E.Flaherty (Ed.), Psychophysiological aspects ofspace flight. New York: Columbia UniversityPress, 1961.

Honorton, C., Drucker, S., & Hermon, H. Shifts insubjective state and ESP under conditions of par-tial sensory deprivation. Journal of the AmericanSociety for Psychical Research, 1973, 67, 191-196.

Honorton, C., & Harper, S. Psi-mediated imageryand ideation in an experimental procedure forregulating perceptual input. Journal of the Ameri-can Society for Psychical Research, 1974, 68, 156-158.

Horowitz, M. J. Image formation and cognition.New York: Appleton-Century-Crofts, 1970.

Hull, C. L. Psychology of the scientist: IV. Pas-sages from the "idea books" of Clark L. Hull.Perceptual and Motor Skills, 1962, IS, 807-822.

Isokower, 0. A contribution to the patho-psychologyof phenomena associated with falling asleep. Inter-


national Journal of Psychoanalysis, 1938, 19, 331-345.

James, W. Principles of psychology. London: Mac-millan, 1890.

Johnson, L. C. A psychophysiology for all states.Psychophysiology, 1970, 6, 501-516.

Kamiya, J. Behavioral, subjective, and physiologicalaspects of drowsiness and sleep. In D. W. Fiske &S. R. Maddi (Eds.), Functions of varied experi-ence. Homewood, 111.: Dorsey Press, 1961.

Kleitman, N. Sleep and wakefulness. Chicago: Uni-versity of Chicago Press, 1963.

Kliiver, H. Mechanisms of hallucinations. In Q.McNemar & M. A. Merrill (Eds.), Studies inpersonality. New York: McGraw-Hill, 1942.

Knoll, M., Kugler, J., Eichmeier, J., & Hofer, O.Note on the spectroscopy of subjective light pat-terns. Journal of Analytic Psychology, 1962, 7, 55-70.

Koestler, A. The act of creation. New York: Mac-millan, 1964.

Kubie, L. S. The use of induced hypnagogic reveriesin the recovery of repressed amnesiac data. Bulle-tin of the Menninger Clinic, 1943, 7, 172-182.

Kubie, L. S., & Margolin, S. An apparatus for theuse of breath sounds as a hypnagogic stimulus.American Journal of Psychiatry, 1944, 100, 610.

Ladd, G. T. Contribution to the psychology ofvisual dreams. Mind (New Series), 1892, 1, 299-304.

Lavie, P. Differential effects of REM and non-REMawakenings on the spiral aftereffect. PhysiologicalPsychology, 1974, 2, 107-108.

Lavie, P., & Giora, Z. Spiral aftereffect durationsfollowing awakening from REM sleep and non-REM sleep. Perception and Psychophysics, 1973,14, 19-20.

Leaning, F. E. An introductory study of hypnagogicphenomena. Proceedings of the Society for Psychi-cal Research, 1926, 35, 289-409.

Leroy, E. B. Les visions du demi-sommeil. Paris:Librairie Felix Alcan, 1926.

Lewin, B. D. Sleep, the mouth, and the dreamscreen. Psychoanalytic Quarterly, 1946, 15, 419-434.

Lewin, B. D. Inferences from the dream screen.International Journal of Psychoanalysis, 1948, 29,224-231.

Liberson, W. T., & Liberson, C. W. EEC records,reaction times, eye movements, respiration, andmental content during drowsiness. Recent Ad-vances in Biological Psychiatry, 1965, 8, 295-302.

Liddon, S. C. Sleep paralysis and hypnagogic hal-lucinations. Archives of General Psychiatry, 1967,17, 88-96.

Lowy, S. Psychological and biological foundations ofdream interpretation. London: Paul, Trench, &Trubner, 1946.

Lukianowicz, 'N. Hallucinations a troix. Archives ofGeneral Psychiatry, 1959, 1, 322-331.

Manaceine, M. de. Sleep: Its physiology, pathology,hygiene, and psychology. New York: Scribner's,1897.

Marsh, H. D. Diurnal course of efficiency. Archivesof Psychology, 1906, 7, 59-69.

Marshall, C. R. Entoptic phenomena associated withthe retina. British Journal of Opthalmology, 1935,19, 177-201.

Maury, A. Des hallucinations hypnagogiques ou deserreurs des sens dans 1'etat interm^diare entre laveille et le sommeil. Annales Medico-Psycho-logiques du Systeme Nerveux, 1848, 11, 26-40.

McKellar, P. Imagination and thinking. New York:Basic Books, 1957.

McKellar, P. Thinking, remembering, and imagining.In J. G. Howells (Ed.), Modern perspectives inchild psychiatry. Edinburgh: Oliver & Boyd, 1965.

McKellar, P. Imagery from the standpoint of intro-spection. In P. Sheehan (Ed.), The function andnature of imagery. New York: Academic Press,1972.

McKellar, P., & Simpson, L. Between wakefulnessand sleep: Hypnagogic imagery. British Journalof Psychology, 1954, 45, 266-276.

Miles, W. R. Horizontal eye movements at the onsetof sleep. Psychological Review, 1929, 36, 122-141.

Mintz, A. Schizophrenic speech and sleepy speech.Journal of Abnormal and Social Psychology, 1948,43, 548-549.

Mitchell, S. W. The effects of Anhalonium Lewinii(The mescal button). British Medical Journal,1896, 2, 1625-1628.

Miiller, J. The physiology of the senses. London:Taylor, Walton, & Mayberry, 1848.

Myers, F. W. H. Human personality and its sur-vival of bodily death, London: Longmans, Green,1904.

Myers, O. H. Images in the mind. Journal of theAmerican Society for Psychical Research, 1957, 51,62-73.

Neisser, U. Cognitive psychology. New York: Ap-pleton-Century-Crofts, 1967.

Ornitz, E. M., Ritvo, E. R., Carr, E. M., LaFranchi,S., & Walter, R. D. The effect of sleep onset onthe auditory averaged evoked response. Electro-encephalography and Clinical N euro physiology,1967, 23, 335-341.

Oswald, I. Sleeping and waking—Physiology andpsychology. New York: Elsevier, 1962.

Partridge, G. E. Reverie. Pedagogical Seminary, 1898,5, 445-474.

Rapaport, D. Cognitive structures. In M. M. Gill(Ed.), The collected papers of David Rapaport.New York: Basic Books, 1967. (a)

Rapaport, D. States of consciousness: A psycho-pathological and psychodynamic view. In M. M.Gill (Ed.), The collected papers of David Rapa-port. New York: Basic Books, 1967. (b)

Rawcliffe, D. H. The psychology of the occult. Lon-don: Ridgeway, 1952.

Rechtschaffen, A., Wolpert, E. A., Dement, W. C.,Mitchell, S. A., & Fisher, C. Nocturnal sleep ofnarcoleptics. Electroencephalography and ClinicalNemo physiology, 1963, 15, 599-609.

Richardson, A. Mental imagery. New York: SpringerPublishing, 1969.


Roheim, G. The gates of the dream. New York:International Universities Press, 1952.

Rosenthal, R. Experimenter effects in behavioral re-search. New York: Appleton-Century-Crofts, 1966.

Rosett, J. The mechanism of thought, imagery, andhallucination. New York: Columbia UniversityPress, 1939.

Rubin, F. (Ed.) Current research in hypnopaedia.London: Macdonald, 1968.

Salzarulo, P., & Cipolli, C. Spontaneously recalledverbal material and its linguistic organization inrelation to different stages of sleep. BiologicalPsychology, 1974, 2, 47-57.

Sartre, J. P. The psychology of imagination. NewJersey: Citadel Press, 1972.

Schacter, D. L., & Kelly, E. F. ESP in the twilightzone. Journal of Parapsychology, 1975, 39, 27-28.

Segal, S. J. Assimilation of a stimulus in the con-struction of an image: The Perky effect revisited.In P. Sheehan (Ed.), The function and nature ofimagery. New York: Academic Press, 1972.

Sheehan, P. W. A shortened form of Belts' question-naire upon mental imagery. Journal of ClinicalPsychology, 1967, 23, 386-389.

Short, R. R., & Oskamp, S. Lack of suggestion ef-fects on perceptual isolation phenomena. Journalof Nervous and Mental Disease, 1965,141, 190-194.

Silberer, H. Report on a method of eliciting andobserving certain symbolic hallucination phenom-ena. In D. Rapaport (Ed.), Organization andpathology of thought. New York: Columbia Uni-versity Press, 1951.

Simon, C., & Emmons, W. Responses to materialpresented during various levels of sleep. Journalof Experimental Psychology, 1956, SI, 89-97.

Singer, J. L. Daydreaming. New York: RandomHouse, 1966.

Slight, D. Hypnagogic phenomena. Journal of Ab-normal Psychology, 1924, 19, 274-282.

Starker, S. Two modes of visual imagery. Percep-tual and Motor Skills, 1974, 38, 649-650.

Stoyva, J. Biofeedback techniques and the conditionsfor hallucinatory activity. In F. J. McGuigan &R. A. Schoonover (Eds.), The psychophysiologyof thinking. New York: Academic Press, 1973.

Stoyva, J., & Kamiya, J. Electrophysiological studiesof dreaming as the prototype of a new strategy inthe study of consciousness. Psychological Review,1968, 75, 192-205.

Taine, H. A. De L'intelligence. Paris: Hachette, 1883.Tart, C. T. Altered states of consciousness. New

York: Wiley, 1969.

Timmons, B., Salamy, J., Kamiya, J., & Girton, D.Abdominal-thoracic respiratory movements andlevels of arousal. Psychonomic Science, 1972, 27,173-175.

Titchener, E. B. Lectures on the experimental psy-chology of the thought processes. New York: Mac-millan, 1909.

Van Dusen, W. The natural depth in man. NewYork: Harper & Row, 1972.

Varendonck, J. The psychology of day-dreams. NewYork: Macmillan, 1921.

Vihvelin, H. On the differentiation of some typicalforms of hypnagogic hallucinations. Ada Psychi-atrica et Neurologica, 1948, 23, 359-389.

Vogel, G., Barrowclough, B., & Giesler, D. Limiteddiscriminability of REM and sleep onset reportsand its psychiatric implications. Archives of Gen-eral Psychiatry, 1972, 26, 449-455.

Vogel, G., Foulkes, D., & Trosman, H. Ego func-tions and dreaming during sleep onset. Archivesof General Psychiatry, 1966, 14, 238-248.

Warren, H. C. Some unusual visual after-effects.Psychological Review, 1921, 28, 453-463.

West, L. J. A general theory of hallucinations anddreams. In L. J. West (Ed.), Hallucinations. NewYork: Grune & Stratton, 1962.

White, R. A. A comparison of old and new methodsof response to targets in ESP experiments. Jour-nal of the American Society for Psychical Re-search, 1964, 58, 21-56.

Williams, H. L., Hammack, J. T., Daly, R. L., De-ment, W. C., & Lubin, A. Responses to auditorystimulation, sleep loss and the EEG stages ofsleep. Electroencephalography and Clinical Nemo-physiology, 1964, 16, 269-279.

Zuckerman, M. Hallucinations, reported sensations,and images. In J. P. Zubek (Ed.), Sensory depri-vation: Fifteen years of research. New York:Appleton-Century-Crofts, 1969.

Zuckerman, M., Albright, R. J., Marks, C. S., &Miller, G. L. Stress and hallucinatory effects ofperceptual isolation and confinement. Psychologi-cal Monographs, 1962, 76(30, Whole No. 549).

Zuckerman, M., & Cohen, N. Sources of reports ofvisual and auditory sensations in perceptual iso-lation experiments. Psychological Bulletin, 1964,62, 1-20.

Zuckerman, M., & Hopkins, T. R. Hallucinations ordreams? A study of arousal levels and reportedvisual sensations during sensory deprivation.Perceptual and Motor Skills, 1966, 22, 447-459.

(Received January 20, 1975)

the hypnagogic state: a critical review of the literature

Documents