Why Do Some People Experience Isolated Sleep Paralysis?


by Robert Chapin

Dr. Christopher Peterson

Mrs. Orít Greenberg

Psychology 111, Section 032

Friday, December 10, 1999


The goal of this paper is to answer two questions: Why does isolated sleep paralysis occur, and why does it matter?  These questions stem from a distinction that exists between the terms, ‘sleep paralysis,’ and ‘isolated sleep paralysis.’  Sleep paralysis (SP) is the state between sleep and wakefulness characterized by the inability to move.  Isolated sleep paralysis (ISP) describes the experience of SP by people who do not suffer from the symptoms of narcolepsy.  This distinction may seem misleadingly subtle.  SP is actually a component of narcolepsy.  ISP occurs without the other components of narcolepsy, which are sleep attacks, cataplexy, and hypnagogic hallucinations (Takeuchi, Miyasita, Sasaki, Inugami, and Fukuda, 1992).

SP can be a very frightening experience.  During an episode, the eyes usually function as they would otherwise (Spanos, McNulty, DuBreuil, Pires, & Burgess, 1995), however psychological symptoms frequently accompany the paralysis.  Commonly reported symptoms include visual, auditory, and tactile hallucinations, as well as the sense of an unseen presence.  These stimuli increase the desire to be able to move.  Although these symptoms do not persist after the paralysis wears off, and their short-term effects appear to be slight, there are indications that the fear and anxiety brought on by ISP cause insomnia, paranoia, and depression (Ohayon, Zulley, Guilleminault, & Smirne, 1999).  This creates the need for an understanding of why ISP can occur in otherwise healthy people.  The results of current research, while encouraging, are far from explaining the causes of ISP, and further from finding a remedy.

This paper will analyze and report the results of five journal articles that discuss research on ISP.  The first three studies focus on sociocultural aspects of ISP research.  Several hundred subjects from China, Japan, and Canada participated in these studies.  The last two studies try to find biological explanations for ISP, including abnormal rapid eye movement (REM) sleep and correlations with other disorders.  All five of the studies attempt to determine the prevalence of ISP.  The conclusion of this paper will summarize the advances made by the presented research, and the evidence they have gathered.

In the first article, Wing, Lee, and Chen (1994) reported on their study of ISP in Hong Kong.  They found that SP was a particularly ancient phenomenon in China.  There was a book written around 403-221 B.C. that referred to ‘dreams of surprise’, which were thought to be similar to SP.  The first Chinese dictionary to contain the common symbol for SP, meaning ‘ghost oppression’, appeared around 30-124 A.D. (Wing, et al.)  With this in mind, the authors created a questionnaire that asked a series of questions about ghost oppression attacks.

Of the 603 Chinese undergraduates who completed and returned a questionnaire, 37% reported having at least one experience of ghost oppression (GO+).  The prevalence decreased to 34.2% after accounting for the subjects who did not report an inability to move.  GO+ subjects reported psychological symptoms including auditory, visual, and tactile hallucinations, the inability to speak, difficulty in breathing, the sense of weight on the chest, and the sense of an unseen presence.  There were no significant differences between sexes (Wing, et al., 1994).

The authors concluded that ghost oppression and SP were indeed the same experience.  Based on the fact that over one sixth of the GO+ subjects remembered certain unusual conditions before having an SP experience, the authors also hypothesized that exhaustion and sleep disruption can cause SP.  The data also showed that one fifth of the GO+ subjects reported having a family member who had also experienced ghost oppression.  The authors finally concluded, “that isolated sleep paralysis is very common among the healthy population.  Both environmental and familial factors seem to contribute to the attacks” (Wing, et al., 1994).

There appeared to be three faults in this study.  First, the questionnaire contained the word for ghost oppression in several places, but it did not offer a definition of the word.  Of the 603 subjects who responded, only 93.3% had heard of ghost oppression.  It is possible that the other 6.7% of the subjects had experienced ghost oppression.  Second, all of the questions appeared to be open-ended, eliciting freely formatted responses.  Because of this, when the authors eliminated subjects who did not report the inability to move as a symptom, they may have eliminated people who thought that this symptom was a fundamental component of ghost oppression, and that it was unnecessary to report it.  If the survey had contained a list of symptoms to choose from, then this uncertainty would not exist.  Third, while their questionnaire was certainly useful in determining what the subjects thought about ghost oppression, the authors ostensibly misused the collected data to determine the prevalence of ISP.  According to a data table in this article, 13% or more of the GO+ subjects suffered from sleep attacks or cataplexy.  The authors failed to eliminate those subjects after reporting the prevalence of SP.  Therefore, the prevalence of ISP was never actually calculated (Wing, et al., 1994).

In the second article, Arikawa, Templer, Brown, Cannon, and Thomas-Dodson (1999) studied ISP in Japan.  After a brief introduction, the authors explained that they had created a questionnaire to study kanashibari (the Japanese word for ISP) and its correlations with death anxiety and locus of control scales.  The hypothesis was that these two other scales would correlate with the occurrence of ISP because of the sense of helplessness and the sense of dying that have been associated with the ISP experience (Arikawa, et al.).

The results of this questionnaire study indicated that 33.9% of the 720 Japanese subjects had experienced kanashibari at least once.  35.8% of the SP sufferers reported that a family member had also experienced kanashibari.  Psychological symptoms measured by this questionnaire resulted in associations similar to those found in the previous study.  78.8% of the subjects who had experienced kanashibari said that they had been anxious or fearful while paralyzed.  41.6% reported difficulty in breathing, about twice the prevalence as reported in the previous study.  However, this increase of the reported difficulty in breathing may represent a fault in this study (Arikawa, et al., 1999).

Unlike the previous study, this study found a positive correlation between being female and experiencing kanashibari.  Although the authors did not address this problem, it became quite apparent. Only 37.2% of the subjects were men.  The results reported a correlation between being female and experiencing difficulty in breathing while paralyzed.  Also reported were the positive correlations of kanashibari to the External-Other Locus of Control[1] scale and the death anxiety scale.  It came as no surprise then, that women tended to have both a greater external locus of control and a greater death anxiety than men did.  The coefficient of correlation with kanashibari and being female was greater than the coefficients for both of the scales (Arikawa, et al., 1999).

There were three other faults of reasoning in this study.  Since the questionnaire did not contain any questions related to sleep attacks or cataplexy, the authors could not have determined the true prevalence of kanashibari.  By the authors’ definition, “In the present study, we focused on an isolated sleep paralysis called kanashibari in Japanese” (Arikawa, et al., 1999).  However, their data did not support the distinction between SP as a symptom of narcolepsy, and ISP.  Second, the article did not report the age at which the subjects had first experienced kanashibari, even though the questionnaire had collected those data.  Finally, when the authors concluded that, “Kanashibari does not seem to be associated with serious psychopathology,” (Arikawa, et al.) the statement was evidently related neither to the collected data, nor to the rest of the article.

In the third article, Spanos, et al. (1995) analyzed research that they had conducted in Canada.  They did not attempt to differentiate ISP from the SP in narcoleptics, so this study was easily comparable to the two previous studies.  Before discussing their methods, the authors mentioned previous research showing that people often describe experiencing the symptoms of SP when they report encounters with UFO aliens.  They suggested that this warrants future research into the evidence of ghosts, demons, even mythological creatures, and their basis in the hallucinations and false sense of presence experienced in SP (Spanos, et al.).

The purpose of this third study was to estimate the prevalence of SP, look for relationships with other psychological traits, and determine if there was a correlation with sexual abuse.  The authors created a questionnaire about sleep that they distributed to 1798 psychology students.  Subjects who indicated experiencing at least one episode of SP received invitations to complete the second phase of the questionnaire.  A comparable number of the subjects who had not experienced SP participated as controls in the second phase.  Analysis of the data collected from the first phase of the questionnaire assured that the subjects’ decision to attend the second phase did not affect the results (Spanos, et al., 1995).

21% of all the subjects reported experiencing SP at least once.  70% said that they were lying on their back when the most recent episode occurred.  Estimations of the episode’s duration ranged from ten seconds to over an hour, with a median of three minutes.  The average age at which the subjects had first experienced SP was 14.  Psychological symptoms were similar to those found in the previous studies.  Symptoms unique to this study included floating in the air, and hearing footsteps.  Compared to the first study, this study found a much greater prevalence of tactile hallucinations and the sense of an unseen presence.  This study did not find any significant sex differences in SP (Spanos, et al., 1995).

The second phase of the study found several statistics contrasting the SP group and the control group.  Subjects in the SP group scored higher than controls for the following variables: night vividness,[2] nightmare/terror, out-of-body experiences, imaginativeness, objective and subjective hypnotizability, psychopathology, and headaches.  Subjects in the two groups did not differ significantly in the variables for sleepwalking/talking and sexual or physical abuse (Spanos, et al., 1995).

Based on these data, Spanos, et al. (1995) developed the core of their conclusion, “the salience hypothesis of sleep paralysis.”  This hypothesis stated that SP was probably more prevalent than the study indicated.  Because imaginativeness correlated not only with the occurrence of SP, but also with the frequency and intensity of SP, the occurrence of SP could be related directly to the intensity of the psychological symptoms.  The hypothesis explained that without the hallucinations associated with SP, the desire to move would be absent.  If a person had no desire to move their body, then they would probably not discover the paralysis.  In fact, the authors found that of all of the subjects who reported experiencing SP, 98.4% also reported experiencing at least one psychological symptom (Spanos, et al.).

The authors also concluded, based on their research of other studies, that SP was more frequently recognized and reported in cultures that had public knowledge and understanding of SP.  To extend their salience hypothesis a step further, they suggested that, “individuals whose culture has primed them to expect sleep paralysis are probably more likely … to recognize subtle or ambiguous symptoms as the signs of paralysis and then confirm their suspicions by attempting to move” (Spanos, et al., 1995).  While these hypotheses were confounding for determining the actual prevalence of SP, they helped explain the difference in prevalence found by the two previously mentioned studies.  In China and Japan, there was a widely understood term for SP, which suggested that these cultures were indeed ‘primed’ to recognize the phenomenon.

In addition to those differences, similarities in SP were also abundant among the studied cultures.  Despite some variation, the first three studies all found the prevalence of SP to be between 20% and 35%.  Two of the three studies found no sex differences in SP.  The first two studies both found that a significant number of their SP subjects had family members who also suffered from SP.  Both the first and third studies concluded that physiological conditions including exhaustion and stress “increase the likelihood of a sleep paralysis episode” (Spanos, et al.).

The three studies presented so far seemed to lack a clear definition and separation of SP and ISP.  The first two claimed to have estimated the prevalence of ISP, but clearly did not show an analysis of subjects’ narcoleptic symptoms.  The third study intended to focus on SP, and entirely failed to mention the existence or distinction of ISP.  Unfortunately, the organization of the next two studies in this regard did not make up for this void in the research.  The amount of information focused specifically on ISP was insufficient to adequately determine prevalence, leaving a need for additional research in this area.

The next two articles, specifically the last one, explored the physiological nature of SP.  Determining the prevalence of ISP took a lower priority because doing so only facilitated certain demographical components of the research.  These articles contributed a basic understanding of SP and how it affected the mind and body, during and after an episode.

In the fourth study, Ohayon, et al. (1999) aimed to estimate the prevalence of SP and its correlates.  They used a computer program called Sleep-EVAL to interview 8085 subjects in Germany and Italy.  Subjects fell into the severity categories of severe, moderate, mild, or no SP.  The authors investigated the relationship of these categories with over twenty different disorders.

The results of this study were displeasing.  6.2% of the subjects experienced SP at least once.  The prevalence of ISP was 1.4%.  Ohayon, et al. (1999) explained that the lower prevalence found in this study may be due to the way questions in other studies were worded.  It seemed doubtful that this was the case.  In the second study, Arikawa et al. (1999) asked their subjects, “When you woke up from sleep or in falling asleep, have you ever not been able to move?”  In contrast, the authors of this fourth study asked their subjects, “Have you ever awakened in the morning and found yourself unable to move?”  The question was repeated twice more to ask about falling asleep for the night and napping.

These and other results (as well as the incorrectly numbered references) looked out of place.  A moment of calculations spawned a great amount of suspicion.  Punching at a calculator for a few minutes confirmed that the numbers presented in this article were mysteriously incorrect.  To begin with, the overall prevalence of SP, derived from 494 out of the 8085 subjects, was 6.1%, not 6.2%. According to their results, “SP was severe in 0.8% of the sample (n = 66), moderate in 1.4% (n = 115), and mild in 4.0% (n = 315)” (Ohayon, et al., 1999).  According to the calculator, 4.0% of the sample was 323, not 315.  The addition of 66, 115, and 315 equaled 496.  However, not that many subjects (only 494) suffered from SP in the first place!  “Finally, isolated SP was observed in 40.6% of the severe SP group, 65.8% of the moderate SP group, and 66.5% of the mild SP group.  This represents 1.7% of the sample” (Ohayon, et al., 1999).  This statement was the one that prompted the calculations. Ignoring the fact that these percentages unavoidably represented fractions of a person, the calculator showed that the prevalence of ISP, based on the given number of people in each group, was 3.9%.[3]

This mathematical preposterousness made the authors’ credibility questionable on all but one point.  Ohayon, et al. (1999) “found that SP is just as likely to begin at any age.”  There was a possibility that the older subjects had forgotten about earlier experiences of SP.  However, they argued, these results were more reasonable than those found in studies such as the previous one by Spanos, et al. (1995), where all of the subjects were college undergraduates.  Such studies have lead to the common belief that experiences of SP begin in adolescence.

In the fifth study, Takeuchi, et al. (1992) sought to induce subjects into an episode of ISP while they recorded vital signs on a polysomonograph (PSG).  Prior research had shown that SP in narcoleptics only occurred during a period of sleep called sleep onset rapid eye movement (SOREM).  The authors of this study defined SOREM as a cycle of sleep in which REM began less than 25 minutes after the subject had entered level one sleep (Takeuchi, et al.).

The results of a questionnaire distributed to 1314 Japanese college students showed that 43% had experienced SP.  Using these data, the authors selected 16 subjects who had experienced SP at least twice, but who had never experienced cataplexy or sleep attacks.  These subjects slept in a laboratory on seven consecutive nights for exactly 7.5 hours per night.  On the third night experimenters recorded a baseline PSG.  During the final four nights, the experimenters interrupted the subjects’ sleep when exactly 40 minutes had passed from the end of the subjects’ first or third cycle of REM sleep.  The interruption consisted of a subject participating in “an auditory vigilance task,” and then returning to sleep after 60 minutes.  Experimenters woke the subjects again after 5 minutes of REM sleep to find out if they had experienced ISP (Takeuchi, et al., 1992).

SOREM occurred after 46 out of the 64 sleep interruptions.  Eighteen of those SOREM cycles took place after a first-cycle interruption, whereas 28 took place after a third-cycle interruption.  These interruptions elicited ISP in five subjects.  ISP always occurred during a SOREM cycle, except for one instance.  SOREM did not occur when one subject reported experiencing ISP a second time (Takeuchi, et al., 1992).

Previous research found that SOREM often occurred as the first cycle of sleep in narcoleptics.  The authors explained that none of their subjects showed SOREM in the first cycle of any night’s sleep. However, because five of the six instances of ISP occurred during SOREM, they hypothesized that most ISP occurs when conditions have lead to an occurrence of SOREM.

Interestingly, the authors found that the intensities of the ISP experiences closely correlated with the amount of time that passed between falling asleep and the beginning of SOREM (called REM latency).  One subject, who reported having the worst symptoms of ISP that she had ever experienced, entered REM sleep without any latency.  Two other subjects, whose experiences were difficult to distinguish from dreams, had REM latencies of 8.0 and 11.5 minutes, respectively (Takeuchi, et al., 1992).

The most significant conclusion of the fifth study was that ISP was a “state dissociation, which means the simultaneous appearance of the elements of two states” (Takeuchi, et al., 1992).  The PSG showed that both the alpha-wave activity characteristic of the waking state, and the muscle atony characteristic of the REM state occurred simultaneously during ISP.  This was different from, but similar in some ways, to the types of state dissociation represented by lucid dreaming, REM behavior disorders, and panic attacks (Takeuchi, et al.).

Do state dissociations and correlations with SOREM and sleep disruption answer the question of why people experience ISP?  At the very least, they indicate progress.  While the fifth study contributes quite a bit of information about ISP, there remains a need for more thorough research into the causes and prevalence of ISP.  Takeuchi, et al. (1992) mention the need for a longitudinal study to determine if their subjects will develop narcolepsy or other disorders because of their repetitive experiences of ISP (Takeuchi, et al., 1992).  Wing, et al. (1994) point out that future studies must include a secondary interview to compliment and confirm questionnaire results.  The article by Arikawa, et al. (1999) ends by stating that, “The etiology of this disorder, however, remains enigmatic.”

Analysis of the five articles, although not satisfying this paper’s more intriguing question about the occurrence of ISP, shows a wealth of information that makes ISP a significant topic, worthy of further research and consideration.  Episodes of ISP can start at any age, usually causing fear and anxiety.  Some studies indicate that ISP may lead to insomnia, paranoia, and depression.  Other indications show that ISP might be the true basis of testimony about alien abductions.  The prevalence of ISP is high enough for it to have a great social impact.  As shown, ISP is so common in many cultures as to have its own word.  People in Newfoundland, for example, call ISP “the ‘Old Hag’ experience” (Spanos, et al., 1995).  Studies of SP prevalence in a variety of cultures indicate that some people might be predisposed to recognizing SP, but also that the prevalence is considerably high in all locations.  These are all facts that will motivate the continued research of ISP.

Coupled with the knowledge that ISP correlates with sleep disruption and stress, future research could prove to be invaluable.  If researchers discover a universal cure to alleviate the symptoms of ISP, then people around the world will enjoy anxiety-free sleep and a reduced risk of developing related disorders.  Before researchers find that cure, however, the research must continue, and researchers must strive to answer this question: Why do some people experience isolated sleep paralysis?  That is the ultimate goal of all research on the phenomenon of ISP.


  1. Arikawa, H., Templer, D.I., Brown, R., Cannon, W.G., & Thomas-Dodson, S.  (1999).  The Structure and Correlates of Kanashibari.  Journal of Psychology, 133(4), 369-375.
  2. Ohayon, M.M., Zulley, J., Guilleminault, C., & Smirne, S.  (1999).  Prevalence and pathologic associations of sleep paralysis in the general population.  Neurology, 52(6), 1194-1200.
  3. Spanos, N.P., McNulty, S.A., DuBreuil, S.C., Pires, M., & Burgess, M.F.  (1995).  The Frequency and Correlates of Sleep Paralysis in a University Sample.  Journal of Research in Personality, 29(3), 285-305.
  4. Takeuchi, T., Miyasita, A., Sasaki, Y., Inugami, M., & Fukuda, K.  (1992).  Isolated Sleep Paralysis Elicited by Sleep Interruption.  Sleep, 15(3), 217-225.
  5. Wing, Y., Lee, S., & Chen, C.  (1994).  Sleep Paralysis in Chinese: Ghost Oppression Phenomenon in Hong Kong.  Sleep, 17(7), 609-613.


[1] “External-Other Locus of Control scores assess the extent to which participants perceive life as determined by luck, fate, chance, or God” (Arikawa, et al., 1999).  The correlation coefficient of this subscale reached significance, whereas those of the Internal Locus of Control and External-Social Locus of Control scales did not.

[2] The night vividness factor “included four items that assessed ability to remember dreams, vividness of dreams and hypnogogic and hypnopompic imagery” (Spanos, et al., 1995).

[3] 40.6% of the 66 in the severe group, plus 65.8% of the 115 in the moderate group, plus 66.5% of the 315 in the mild group, equaled 312 subjects out of 8085.  That sum encompassed more than 3.85% of all subjects in the study.

Comment Feed

One Comment

  • lena says:

    experienced 3 times, last one can describe as a jolt with immediate feeling of a presence and overwhelming tingling all over the body,hope this will add to the description

Write a Comment