Sleep, 5(3):290-299
© 1982 Raven Press, New York
Self-Representation in Dream Experiences
During Sleep Onset and REM Sleep
Marino Bosinelli, Corrado Cavallero, and PierCarla Cicogna
Institute of Psychology, University of Bologna School of Medicine, Bologna, Italy
Summary: The Scoring System for Latent Structure (SSLS) was used to test
three hypotheses regarding the degree of self-participation in dreams reported
during sleep onset (SO) and rapid eye movement (REM) sleep, These hypotheses were that (a) the ratio of interactive to associative sentences would be
significantly greater in REM than in SO; (b) the level of Ego activity would be
significantly greater in REM than in SO; and (c) the ratio between interactive
sentences with Ego not present and interactive sentences with Ego present
would be significantly different in REM and in SO. None of these hypotheses
was confirmed. However, the following significant differences were found: (a)
the ratio between sentences with Ego substituted and the total number of
interactive sentences was greater in REM than in SO (p < 0.05); (b) the ratio
between sentences with Ego substituted and sentences with Ego present was
greater (p < 0.05) in REM than in SO; (c) the ratio between sentences with Ego
inserted and sentences with Ego substituted was greater in SO than in REM (p
< 0.01); and (d) the relative incidence of defective Ego was greater (p < 0.01) in
SO than in REM. These data were interpreted in terms of psychodynamic
models of dream formation. Key Words: Self-representation-Sleep onsetREM sleep-SSLS-Dream analysis.
The demonstration of a dream-like mental activity during sleep onset (SO) and,
to a lesser degree, in stages of NREM sleep different from SO, has permitted
refutation of the hypothesis that dreams occur nearly exclusively during REM
sleep (1-12). During SO, various kinds of rather diverse experiences and mental
activities have been described, ranging from loss of voluntary control of thought,
to presence of bizarre body sensations, to hypnagogic hallucinations, to true
oneiric experiences (3,5,6,8,12). Among these phenomena, the oneiric experiences of SO seem not to be distinguishable from those typical of REM sleep (11).
We can conclude, therefore, that dreamlike mental activity appears most frequently during REM sleep, less frequently in SO, and even less frequently in the
Accepted for pUblication May 1982.
Address correspondence and reprint requests to Dr. Marino Bosinelli, Institute of Psychology,
University of Bologna, Viale Berti Pic hat 5,40127 Bologna, Italy.
290
SELF-REPRESENTATION IN DREAM EXPERIENCES
291
other stages ofNREM sleep. Different psychophysiological models (REMINREM
dichotomy (13); tonic/phasic model (14); P hypothesis (15); activation/synthesis
hypothesis (16); REM sleep sensory control hypothesis (17) ) can justify the presence of dreams in REM sleep rather satisfactorily. But so far, no psychophysiological model has explained the appearance of dreams during SO. We
believe that a psychodynamic model may, at present, be more appropriately
applied to SO dreaming. SO mental experience becomes most dreamlike when
reality testing is no longer efficient (6) and when the feeling of self-participation
appears (8). These two phenomena (the loss of reality testing and the appearance
of self-participation) could represent an attempt of defense against the regressive
ideation and destructuralization that accompany sleep onset.
That REM dreams and SO dreams may currently best be explained by different
models (psychophysiological (REM) vs. psychodynamic (SO) ) does not preclude
the future integration of these models. A more immediate task, however, is the
delineation of common characteristics and differences between the two classes of
dreams. For instance, if personal participation in the dream scene is observed in
both SO and REM dreams, one may reasonably wonder whether the degree of
such participation, or hallucinatory representation of the self, varies significantly
between the two different conditions.
Using an ordinal rating scale, Bosinelli et al. (18) attempted to quantify dream
self-participation in the dream scene. On their 1 to 6 point scale, they observed
significant differences in self-participation between phasic REM and SO dreams,
but not between tonic REM and SO dreams. Their methodology, however, may
have been limited by the fact that the dream was rated as a whole, without taking
into account its narrative or dynamic development.
More recently, Foulkes (19) has proposed a complex technique of identifying
and scoring the discrete elements of dream reports. His system, Scoring System
for Latent Structure (SSLS), is meant to be applicable both to the manifest content of the dream itself and to the free associations. Foulkes's system is in part
psychodynamic and in part psycholinguistic: the text of the dream report is subdivided into linguistic units (sentences). The most important feature of the SSLS
textual analysis is that a distinction is made between interactive sentences and
associative sentences. These sentence classes are discriminated on the basis of
their relational or verb terms: where these terms indicate the presence of relative
movement between the subject and the object of the relationship, the sentence is
interactive; where they indicate the absence of relative movement, the sentence is
associative. According to Foulkes, interactive sentences represent motive structures, wishes, intentions, and/or behavioral programs of the dreamer, while associative sentences correspond to more purely cognitive structures.
Within rules of SSLS, the dreamer, i.e., the Ego, must be scored as present in
each interactive sentence, either in the position of subject or of object, even if
Ego's role is not manifest in the text being scored. For example, where a dreamer
states that X ran toward Y, and both X and Yare different from the Ego, the Ego
nonetheless must be introduced as one or the other of these elements. This may be
debated as much on theoretical grounds as on the practical difficulties implied by
Sleep, Vol. 5, No.3, 1982
292
M. BOSINELLI E1' AL.
an artificial introduction of the Ego. However, it remains a fundamental element
of SSLS, without which the system would lose a considerable amount of its
internal consistency.
In this research, we have specified two principal modalities by which the Ego is
introduced into SSLS interactive sentences: substitution and insertion (see Methods). We have used these and other, more standard, SSLS scores to evaluate
differences in self-representation between SO and REM dreams. Our initial
hypotheses, and their justifications, are presented below, following an elaboration
of the SSLS methods used to evaluate them. Conceptually, the present research
can be viewed as a test of the possibility of explaining certain aspects of dreamlike
activity during SO, as well as of explaining the difference between REM dreamlike
mental activity and SO dreamlike mental activity, by means of psychodynamic
hypotheses. Such hypotheses, which derive from the psychoanalytical model,
more often than not are formulated and tested in a clinical setting; our research
attempts to examine the power of such a model in an experimental setting.
METHODS
Subjects and experimental procedures
Ten male university student volunteers between ages of 18 and 28 were
used as subjects. The electropolygraphic recording of nocturnal sleep was carried
out according to the following scheme: two electroencephalogram channels, two
electrooculogram channels, and one electromyogram channel. Two awakenings
were made per night over the course of the 4 nonconsecutive nights for which each
subject was studied. Dream reports were solicited on each awakening.
SO awakenings were made 3 min after the appearance of the first EEG spindles,
on the condition that the tracing still maintained the characteristics of descending
stage II. The choice of this moment for awakening the subjects at sleep onset was
justified by the fact that, in previous research, we have made awakenings at two
points during descending stage II (SOa, from 1 to 1.5 min after the appearance of
the first spindles; SOb, from 4 to 5 min after the appearance of the first spindles)
and have, in both cases, obtained the recall of clear, dreamlike experiences. REM
awakenings were made 10 min after the appearance of the first clear rapid eye
movements. On each night, the SO awakening always preceded the REM awakening.
Of the 80 experimental awakenings, subjects reported no recall in 5 cases (4 in
SO and 1 in REM sleep) and in 75 cases reported recall of a mental experience. Of
the 75 mental experiences, 7 (2 in SO and 5 in REM) were judged to be not
dreamlike but merely thought-like, while 68 were dreamlike; i.e., dramatized visual experiences with a loss of reality testing. The number of SO dreamlike reports
ranged from 3 to 4 for each subject (x = 3.4); the number of REM dreamlike
reports ranged from two to four for each subject (x = 3.4); and finally, the total
(SO + REM) number of dreamlike reports ranged from five to eight per subject (x
= 6.8). This relatively stable distribution of reports across subjects suggests that
the data evaluated here are adequate to test hypotheses about differences in
self-participation between SO and REM reports.
Sleep, Vol. 5, No.3, 1982
SELF-REPRESENTATION IN DREAM EXPERIENCES
293
SSLS scoring reliability
The authors, separately, scored all the 68 dreamlike reports according to SSLS.
Individual scorings then were compared (Table 1), with the aim of calculating
reliability indices and, moreover, of obtaining concordant scorings against which
the formulated hypotheses could be tested. The values in Table 1 labeled total
reliability indicate agreement on all of the terms (subject, relational form or verb,
object) of sentences scored in common. Greater than 80% agreement was
achieved in all tabulated comparisons. Thus, SSLS scoring reliabilities were quite
satisfactory.
SSLS scoring conventions
In SSLS scoring, we introduced differentiations to take account of the diverse
possible modalities of Ego presence in interactive sentences. As already mentioned, SSLS demands that the Ego be present in one of the terms of an interactive
relation, but not in both, inasmuch as reflexive relations are proscribed. (a) The
least problematic case, obviously, is that in which the presence of Ego is textually
explicit: the dreamer narrates a dynamic action in which he is personally involved.
We have indicated this type of interactive sentence as E. (b) In any interactive
sentence where the Ego is personally involved as subject, but where the object
that receives the action is not recuperable from the context, we have used the
symbol E . .. to indicate the sentence as defective. (c) For the cases in which the
Ego is not present in the text as a character involved in the action, we have
indicated the corresponding interactive sentence with the symbol Enp (Ego not
present). The introduction of the Ego in the scoring foresees two possibilities. If
the Ego is put in the place of one of the characters of the interaction [That person
looked at the women = (E) ~ Pf1], the symbol (E) indicates the process of
substitution: Ego substituted. If, on the other hand, the Ego is introduced as the
subject of an action that, from a textual point of view, does not contain an interactive process between two characters (That guy ran away = +-Pm 1 , E), the symbolE
indicates the process of insertion: Ego inserted. (d) Finally, for any interactive
sentence, the role of the Ego can be characterized dynamically either as active
(subject role) or as passive (object role).
Examples of all the symbols utilized, those originally conceived by Foulkes (19)
TABLE 1. Reliability (proportions of agreement)
Sentences scored in common
Reliability on the identification of interactive
and associative sentences
Reliability on the relational form
Reliability on subject and object
Total reliability
A-B
A-C
B-C
0.81
0.84
0.81
0.97
0.89
0.87
0.81
0.99
0.94
0.89
0.86
0.97
0.90
0.84
0.80
Sleep, Vol. 5, No.3, 1982
M. BOSINELLI ET AL.
294
as we!! as those elaborated for purposes of the present research, 1 are as follows:
Interactive sentences.
~ Approach relationship: I was talking to a friend:
E ~ Pm
~Avoidance relationship: I went away from the house: ~ E, SylA
--< Attack relationship: I was yelling at my child:
E --< C
@+Creating relationship: I wrote a book:
E ~ SylA
Associative sentences.
(E = F]
= Equivalence relationship: I became a teacher:
Means relationship:
~With
1I sent ~e~ .s~~e flowers:
relationship: I am with my father:
Active interactive sentence.
Act: I looked at a woman:
Passive interactive sentence.
Pass: The friend came near me:
Levels of the presence of the Ego.
E
I was talking to a person:
E ... I was running:
(E) She was listening to a concert:
E My father died:
E~ Pf
(SylA = EBPfJ
(E ~F]
Pm~E
E~Pm
E ...
(E) (pt)
•
E--<F
~
SylA
For the differential analysis of dreamlike reports from SO and REM, the eight
indices presented in Table 2 were calculated for each condition.
Hypotheses about SSLS indices
With respect to the principal indices, the following hypotheses were formulated.
(a) Given (i) that interactive sentences are supposed to express motivational
drives, (ii) that in those sentences, even independently of the processes of substitution and insertion, the presence of the Ego weighs more heavily than in
associative sentences, and (iii) that the dreamlike structure of REM experiences,
compared to SO experiences, seems to be richer and more complete as far as
perceptual vividness and self-participation level are concerned (18), the hypothesis was formulated that the index II A (ratio between interactive and associative
sentences) would be significantly greater in REM sleep than in SO.
(b) On the basis of these same considerations, the hypothesis was formulated
that the level of activity of the Ego might be greater in REM sleep than in SO, and
that, therefore, the ratio Pass! Act would be significantly less in REM sleep than in SO.
(c) The last hypothesis concerns the index Enp/E, or the ratio between interactive sentences with the Ego not present (substituted or inserted) and interactive
sentences with the Ego textually present. Taking account of the finding of greater
intensity of the feeling of personal participation in REM sleep (18), one of us
I
1 The scoring was carried out using Foulkes's symbols, except for Act, Pass, (E), and E, which were
introduced by the authors.
Sleep, Vol. 5, No.3, 1982
SELF-REPRESENTATION IN DREAM EXPERIENCES
295
TABLE 2. SSLS indices
IIA
Pass/Act
Enp/E
(E)/I
E .. .IE
(E)/E
E .. .11
E/(E)
,j,
I, Interactive sentences; A, associative sentences; Pass, Passive
interactive sentences; Act, Active interactive sentences; Enp, Interactive sentences with the Ego not present; E, Interactive sentences with the Ego present; E ... , Defective interactive sentences;
"
(E), Interactive sentences with the Ego substituted; E, Interactive
sentences with the Ego inserted.
(M.B.) predicted that there would be a relatively lower number of interactive
sentences in which the Ego would have to be either substituted or inserted; hence,
that the ratio Enp/E should be significantly less in REM sleep than in SO. However, another of us (P.C.) formulated an opposing hypothesis, in which the ratio
Enp/E should be significantly greater in REM sleep than in SO. This second
hypothesis was proposed on the following two grounds.
(i) In SO, the physiological condition usually allows for a brief, dreamlike experience very close in time to waking feelings and experiences. It does not seem
probable, then, that in this condition one can easily or frequently activate the more
complex mechanisms of dream production (condensation, identification, projection, displacement, and so on) that create, according to a psychodynamic model,
the disguised representation of the Ego's motivations in different characters.
(ii) Following the theory that the dream-like experience of SO is interpretable as
a hallucinatory defense against the destructuralization that is taking place, a direct
self-hallucination would seem necessary to achieve this end. This might be expressed in the number of interactive sentences with the Ego actually present,
without substitution, dispersion, or condensation in other characters.
Considering, finally, all of the other indices analyzed, we did not (even though
holding that some significant differences between the REM and SO reports might
exist) believe it possible, within the limits of current knowledge, to formulate
precise hypotheses about the direction of probable differences.
Statistical analysis
The t test was used for statistical analysis of related measures. For each index,
data were averaged so that a single value for each sleep condition for each of the
ten subjects could be obtained.
RESULTS
None of the three comparisons for which predictions were made (IIA, Pass/Act,
Enp/E) produced significant results (Table 3). The observed difference in Enp/E
Sleep. Vol. 5, No.3, 1982
296
M. BOSINELLI ET AL.
between SO and REM was, however, marginally significant (p < 0.10), indicating
more Enp sentences relative to E sentences in REM than in SO.
Significant differences were found, however, for indices for which preliminary
hypotheses had not been formulated. First, significant differences (p < 0.01) were
obtained in both E .. .II and E .. .IE, revealing a greater incidence of defective
Ego in SO than in REM. Second, by dividing the category Enp (Ego not present)
• we observed (a) a
into the two modalities of substitution (E) and insertion E,
significant difference (p < 0.01) in (E)/I indicating a greater incidence of substituted Ego sentences, compared with the total number of interactive sentences, in
REM than in SO; (b) a significant difference (p < 0.05) in (E)/E indicating a greater
incidence of substituted Ego sentences, compared with sentences with the Ego
actually present, in REM than in SO; and (c) a significant difference (p < 0.05) in
•
E/(E)
indicating a greater incidence of inserted Ego sentences, compared with
substituted Ego sentences, in SO than in REM.
DISCUSSION
Interpretation of the results must take into account that the formulated hypotheses were not confirmed, but that significant differences were found for the indices for which hypotheses had not been formulated.
Explicitly formulated hypotheses
Hypotheses a and h. The data did not confirm the hypothesis that there would
be a significantly greater incidence of interactive sentences (as compared with
associative sentences) in REM than in SO. Within the limits of the methodology
used, this means that there was no difference in the two conditions with respect to
the quantitative relationship between the motivational and the cognitive aspects of
dreams (the terms motivational and cognitive are used here in the sense suggested
by Foulkes (19». Analogously, the hypothesis that there would be a greater level
of Ego activity in REM as compared to SO was not confirmed. Taken as a whole,
TABLE 3. Results and significant differences
SO
-->VA
--> Pass/Act
-->Enp/E
E .. .11
(E)/I
E . . .lE
(E)/E
'"
E/(E)
REM
x
SO
x
SO
0.559
0.674
0.198
0.194
0.060
0.235
0.089
0.305
0.489
0.220
0.099
0.101
0.111
0.146
0.530
0.555
0.446
0.040
0.179
0.059
0.336
0.219
0.364
0.513
0.057
0.191
0.090
0.492
0.994
0.930
0.322
0.614
-->, Indices for which hypotheses were formulated.
* p < 0.10; +, p < 0.05; ++, p < 0.01.
Sleep, Vol. 5, No.3, 1982
ns
ns
ns*
++
++
++
+
+
SELF-REPRESENTATION IN DREAM EXPERIENCES
297
these two results seem to falsify the hypothesis of a greater prominence, in REM
sleep, of oneiric events tied to active self-representation.
These data contradict, in part, some of the results of a previous research (18),
which revealed a significant difference in self-participation between phasic REM
and SO. This discrepancy might be explained in two ways. First, the measuring
instruments used in the two experiments were different. In the earlier experiment,
a single score was assigned to each dream on the basis of the last discrete experience in that dream. This score was achieved by transforming various characteristics of the dreamer's activity, both quantitative and qualitative, into an ordinal
scaling system. The system used in the present experiment, on the other hand,
measured the frequency with which specified sentences appeared in the report.
Second, the differences revealed by the ordinal scale were found in a comparison between phasic REM and SO dream reports. No significant differences were
found between tonic REM reports and those from SO. In the present experiment,
REM sleep deliberately was considered as a single unified category, i.e., no distinction was made between awakenings in tonic vs. phasic REM sleep. This
choice was not made because of a rejection either of the significance and heuristic
value afforded by the tonic-phasic model (14,20) or of "microscopic" research
methods (21). Rather, it was based on the consideration that a distinction between
phasic REM and tonic REM mentation depends on the isolation of the last discrete
mental experience before awakening. That methodology is incompatible with the
attempt to analyze, by means of SSLS, the dream experience as a psychic event
whose development may involve a considerable temporal duration. With respect
to possible tonic vs. phasic REM differences, we have not classified dream reports
according to whether or not they included thought-like activity, but we have
excluded from analysis those reports that consisted only of such activity. That is,
reports in which clear hallucinatory characteristics (perceptual, dramatized experiences with the loss of reality testing) were not present, were eliminated from our
analyses.
Hypothesis c. We hypothesized that there would be a significant difference
between REM and SO reports in the ratio between Ego-present and Ego-notpresent sentences. We did not agree, however, on the expected direction of this
difference. The results did not permit definitive rejection of the null hypothesis,
although they were, at p < 0.10, consistent with the hypothesis that the ratio
Enp/E would be higher in REM than in SO. Moreover, the arguments supporting
this specific hypothesis found some additional justification in results for other
dream variables for which we had not made specific predictions (see below).
Other dream variables
The following significant differences were observed: the ratios (E)/I and (E)/E
were greater (i.e., substitution of the Ego was relatively more frequent) in REM
sleep than in SO; the ratios E .. .II and E .. .IE (i.e., interactions with the Ego as
subject but without a textually identifiable object) were relatively more frequent in
• (the insertion of the Ego relative to its
SO than in REM sleep; and the ratio E/(E)
substitution) was significantly greater in SO than in REM.
The first two of these results can be considered as a single datum, and as
Sleep, Vol. 5, No.3, 1982
M. BOSINELLI ET AL.
298
constituting an indirect confirmation of one of the formulated hypotheses
(hypothesis c, author P.C.). The significant differences were substantiaHy due io
the extremely low number of Ego substitutions scored in the SO condition. For
example, for 6 of 10 subjects, no substitution of the Ego ever was necessary in SO
reports.
One needs to bear in mind that in this experiment, only dreamlike SO experiences were analyzed. It is well known that this type of SO experience requires the
complete loss of reality testing, rather well-organized and not very regressive
dramatization, and the feeling of personal participation. A possible model ofthese
phenomena is provided by the description, by Vogel et al., (6) of a sequence of
Ego states at SO, tempered by the criticisms formulated by Schacter (22). In the
I-D-R sequence (Intact Ego-Destructured Ego-Restructured Ego) proposed by
Vogel et al. the dreamlike SO experience would fall into the last category, in
which the restructuring force seems to be a result of the self-participation that
permits the subject to hallucinate his own presence in the oneiric scene (8). If the
restructuring task is functionally entrusted to personal participation, one readily
can understand why, in this specific type of SO experience, the dream mechanisms which underlie Ego substitution and insertion are scarcely active.
Our results do not contradict earlier interpretations of the function of selfhallucination in the oneiric scene during either phasic REM or SO (18). But the
data are not sufficient to confirm that production mechanisms operate more directly on the available mnemonic material during SO than during REM sleep. To
test this hypothesis, one needs to apply SSLS not only to the manifest textual
structure of the dream itself, but also to the material supplied in the dreamer's free
associations. We intend to carry out a research project using this approach.
As already noted, the significantly different values for the indices (E)/I and
(E)/E in REM vs. SO partially confirm hypothesis c in the direction indicated by
one of the authors. In fact, the incidence of Ego-inserted sentences (the second
type of Ego introduction), with respect either to interactive sentences in general or
to sentences with Ego present, was not significantly greater in REM than in SO.
The remaining significant differences we observed were for the indices E .. .II,
•
E. . .IE and E/(E),
whose values were higher in SO than in REM. These results
could be attributed to the more elaborated and articulated structure of dream
experience in REM sleep. In fact, textual sentences which require an SSLS
translation with a defective Ego or a process of Ego insertion represent narrative
elements formulated on the basis of imprecise, generic descriptions that are
scarcely organized or focused, as can be readily observed from the following
examples.
I ran
I was walking
E. .. I (without defining toward what
E. .. \ object, place, or person)
+
•
People that were walking
Pm -J> E
There were some words but
I don't know who spoke them Pm -J> E•
Sleep, Vol, 5, No.3, 1982
SELF-REPRESENTATION IN DREAM EXPERIENCES
299
Maybe there was a conversationE• -.;> Pm
Pm-.;> E•
Any exact psychodynamic interpretation of structural differences between
dream experiences in SO and those in REM awaits the kind of integrative investigation mentioned above, in which there would be a parallel analysis of dream
structure and of free associations. According to our hypothesis, the free associative structure connected with SO dream experiences should be less complex and
less articulated than the free associative structure connected with REM dream
experiences. In other words, the gap between latent content (the dream deep
structure) and manifest content (the dream experience) should be narrower in SO
dreams than in REM dreams.
Acknowledgment: The authors are grateful to Dr. David Foulkes for his excellent
revision of the manuscript.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
Foulkes D. Dream reports from different stages of sleep. J Abnorm Soc Psychol1962: 65:14-25.
Foulkes D. Non-rapid eye movement mentation. Exp Neurol 1967; SuppI4:28-38.
Foulkes D, Vogel G. Mental activity at sleep onset. J Abnorm Psychol 1965; 70:231-43.
Monroe L, Rechtschaffen A, Foulkes D, Jensen J. Discriminability of REM and NREM reports. J
Pers Soc Psychol 1965; 2:456-60.
Foulkes D, Spear P, Symonds J. Individual differences in mental activity at sleep onset.J Abnorm
Psychol 1966; 71:280-286.
Vogel G, Foulkes D, Trosman H. Ego functions and dreaming during sleep onset. Arch Gen
Psychiat 1966; 4:238-48.
Bosinelli M, Bagnaresi G, Molinari S, Salzarulo P. Caratteristiche dell'attivita psicofisiologica
durante il sonno: un contributo aile tecniche di valutazione. Riv Sper Fren 1968; 92:129-50.
Bosinelli M, Molinari S. Contributo aIle interpretazioni psicodinamiche del processo di addormentamento. Riv Psicol 1968; 62:369-94.
Fischgold H, Safar S. Etats de demi-sommeil et images hypnagogiques. Triangle 1967; 7: 135-49.
Pivik T, Foulkes D. NREM mentation: relation to personality, orientation time, and time of night.
J Consult Clin Psychol 1968; 32: 144-51.
Vogel G, Barrowclough B, Giesler D. Limited discriminability of REM and sleep onset reports and
its psychiatric implications. Arch Gen Psychiatry 1972; 26:449-55.
Slap J. On dreaming at sleep onset. Psychoanal Q 1977; 46:71-81.
Dement W, Kleitman N. The relation of eye movements during sleep to dream activity: an
objective method for the study of dreaming. J Exp Psychol 1957; 53:339-46.
Grosser G, Siegal A. Emergence of a tonic-phasic model for sleep and dreaming: Behavioral and
physiological observations. Psyc hoi Bull 1971; 75 :60 - 72.
Dewan E. The programming (P) hypothesis for REM sleep. In: Hartmann E, ed, Sleep and
dreaming. Boston: Little, Brown & Co., 1970:295-307.
Hobson A, McCarley R. The brain as a dream state generator: an activation-synthesis hypothesis
of the dream process. Am J Psychiatry 1977; 134:1335-48.
De la Pena A. The psycho-biological role of the rapid eye movement dream state. Unpublished
doctoral dissertation, Stanford, University, 1971.
Bosinelli M, Cicogna P, Molinari S. The tonic-phasic model and the feeling of self-participation in
different stages of sleep. Ital J Psychol 1974; 1:35-65.
Foulkes D. A grammar of dreams. New York: Basic Books, 1978.
Molinari S, Foulkes D. Tonic and phasic events during sleep: psychological correlates and implications. Percept Mot Skills 1969; 29:343-68.
Salzarulo P, Cipolli C, Lairy G, Pecheux M. L'etude psychophysiologique de l'activite mentale du
sommeil: analyse critique des methodes et theories. Evol Psychiatr 1973; 1:33-70.
Schacter DL. The hypnagogic state: a critical review of the literature. Psychol Bull 1976;
83:452-81.
Sleep, Vol. 5, No.3, 1982