[From NIGHTLIGHT 5(2), Summer 1993, Copyright, The Lucidity Institute.]
TESTING THE LIMITS OF DREAM CONTROL:
THE LIGHT AND MIRROR EXPERIMENT
by Lynne Levitan and Stephen LaBerge
Lucid dreaming offers the promise of enhanced control over dreams. Yet
the question remains quite unanswered of how much dream control is
possible. The ability to have lucid dreams also makes possible a way to
study this issue. After having attained lucidity in a dream, dreamers
can choose to carry out predetermined experiments testing their ability
to achieve certain goals. In the "Free Fall" issue of NightLight (Vol.
4, No. 4) we asked lucid dreamers to attempt certain tasks in lucid
dreams and to report on the outcome. An introduction to the many
viewpoints on dream control will put the findings from this study in
WAYS OF CONTROLLING
The definition of "control" used here is "the ability to determine or
influence the course of events." This means that an individual's actions
are causes with subsequent effects. One way to refine this definition is
to distinguish between voluntary and involuntary control. Voluntary
control means that you decide you want to produce a certain effect and
take steps to cause it. For example, you want the house to be cleaner so
you throw away stray papers. Involuntary control refers to unintended
consequences of your actions. For example, one of those stray papers you
threw away was the outline of the presentation you are to give tomorrow.
The result is you have to write a new one. In a dream, an example of an
involuntary effect would be causing a dream monster to pursue you by
There are different ways to approach controlling dreams. A method that
does not require lucidity is predetermination: selecting the setting or
topic of the dream prior to sleep. This is akin to the idea of "dream
incubation" in which a person works to induce a dream about an important
topic in order to answer a question or resolve a conflict. In her book,
Creative Dreaming, Patricia Garfield presents some evidence that
motivated people can choose to dream about desired topics. Post-hypnotic
suggestions have also been employed in attempts to elicit particular
dream topics, again with some success, as described by Charles Tart in
his essay in Conscious Mind, Sleeping Brain, edited by Gackenbach and
LaBerge (1988). Success with creating a particular dream setting,
however, does not imply the ability to control the sequence of events in
Concurrent control is ability to determine or alter the course of a
dream in "real time," as it happens. This type of control is not limited
to lucid dreams, anymore than our effect on the waking world is limited
to times when we are thinking about what we are doing. Anytime we make a
choice or act in a dream, we are controlling it. We may be unconscious
of the reason for our choice, but the decision nonetheless originates
within the self. However, people possessing lucid consciousness in their
dreams are able to make deliberate choices and actions with full
knowledge that they are experiencing a dream, and observe their effects
on the course of the dream.
The question addressed here is how well can we influence dreams in the
directions we desire? Do actions produce the aimed-for effects? Do we
have more or less control over our experiences in dreaming than in
THE RANGE OF OPINIONS REGARDING DREAM CONTROL
In the modern world, a wide variety of theories and opinions about the
possibility or impossibility of dream control coexist. At one extreme,
stand (or, perhaps, stood, as this viewpoint may have faded in the face
of irrefutable evidence) some sleep researchers whose reluctance to
believe in the possibility of deliberate dream control came hand in hand
with their disbelief in the verity of lucid dreaming. Their opinions
rose out of a faulty philosophy defining sleep as "unconsciousness,"
meaning lack of cognition. A better operational definition of sleep
would be, "lack of perceptual awareness of the sleeper's environment."
Without consciousness, clearly one could not consciously will anything.
So, until an awareness arose among those studying dreams that dreaming
was a state of consciousness, not unconsciousness, progress was not
possible in this area.
Another kind of disbelief has arisen from dreamworkers, who employ
dreams to help people achieve better psychological balance. Much of
dream-based therapy (although not all) has operated on the premise that
dreams are things that happen to people, rather than events that people
create. The creator of dreams has been named the "unconscious." Because
of this and the prevailing notion in the scientific world that sleep is
unconsciousness, it has become common for people to believe that dreams
occur in the unconscious mind, independent of the conscious "ego."
This cannot be true, however, because if it were, we would not be able
to recall the experiences we have in dreams. Events that do not reach
consciousness are not accessible to memory. The "I" of the dreamer, the
one who sees, hears, feels, and reflects on the events happening in a
dream is the self-awareness, the "ego," and it is conscious, although it
may not be aware that its present circumstance is an entirely mentally-
constructed world not guided by sensory information from physical
As an illustration of the point of view that dreams are both from and in
the unconscious, here is an excerpt from Working with Dreams by Ullman
Q. Can we program or control our dreams?
A. No, not consciously. If we look upon a dream as a kind of natural
resource flowing within us, if we liken it to a river, a river shaped
by our life experience, then its flow will not be changed simply by
having someone on the shore urge a new direction on it. But if the
person on the shore does the work necessary to make a change in
direction possible, the flow will alter as desired. The point of the
analogy is that there has to be more than conscious intent to
influence the flow. There has to be a genuine emotional investment.
In the view expressed, dreams are predetermined "plays" somehow
programmed out of the individual's current psychological processes. They
are nothing like waking life. These same authors make an interesting
comment about lucid dreaming. They state: "Although the dreamer can
influence the subsequent course of dream once it becomes a lucid dream,
the element of control occurs only within certain limits. An analogy
might be Living Theater where, after the actors have created a certain
framework, the audience is invited to influence the subsequent course of
This statement implies that dream control is limited to actions
appropriate to the original setting of the dream, which has its own
defined boundaries and rules. This seems to imply that whatever part of
the mind determines the original dream setting has primacy over other
parts of the mind. Certainly, one of the great mysteries of dreams is
what determines the original setting and situation one finds oneself in
a dream. Despite reports that some people are able to decide what they
will dream about on occasion, for the most part, dream topics seem to
arise out of some source that is definitely not in consciousness.
However, there is no evidence in support of Ullman and Zimmerman's
contention that dream control is limited by the framework of the
original dream setting, and many would refute it based on their own
The Tibetan Buddhists, creators of the Dream Yoga, teach that it is
possible to control every aspect of dream imagery. They use dream
control as a method of comprehending the illusory nature of all
experience, with the ultimate goal of transcending the relative and
embracing the Absolute. In the "Doctrine of the Dream State" from
Tibetan Yoga and Secret Doctrines translated by Evans-Wentz, we find the
At the outset, in the process of realizing [the dream] to be maya,
abandon all feeling of fear;
And, if the dream be of fire, transform the fire into water,
the antidote of fire.
And if the dream be of minute objects, transform them
into large objects;
Or if the dream be of large objects, transform them
into small objects:
Thereby one comprehendeth the nature of dimensions.
And if the dream be of a single thing, transform it
into many things;
Or if the dream be of many things, transform them
into a single thing...
Then, the editors comment:
By such practices, the yogin is taught to realize that matter, or
form in its dimensional aspects, large or small, and in its numerical
aspects, of plurality and unity, is entirely subject to one's will
when the mental powers have been sufficiently developed by the yoga.
In other words, the yogin learns by actual experience, resulting from
psychic experimentation, that the character of any dream can be
changed by transforming or willing that it shall be. A step further
and he learns that form, in the dream-state, and all the
multitudinous content of dreams, are mere playthings of the mind, and
therefore, as unstable as mirage. A further step leads him to the
knowledge that the essential nature of form and of all things
perceived by the senses in the waking state are equally as unreal as
their reflexes in the dream state..."
The Tibetan Buddhists, however, were and are not given to sharing their
personal dream experiences, so we cannot examine the nature of their
dreams and their efforts to control them. Some notable Western expert
lucid dreams have given us a look into what they have been able to
accomplish. The Marquis d'Hervey de Saint-Denys was an extraordinarily
accomplished lucid dream and wrote instructively about his experiences.
He exhorted his readers to strive to control their dreams in his 1867
book Dreams and How to Guide Them:
Those who would see in the incidents of our dreams merely a succession
of mechanically produced impressions over which one has no more control
that a simple spectator has over some pictures will naturally declare
any effort and any exercise of attention or will to be incompatible with
the very nature of dreaming. Since the most valuable observations I have
been able to make seem to me to be due to my ability to maintain the
faculties of attention and will during sleep, I shall naturally place
great emphasis on convincing the reader that he can and should exercise
the same control over himself. Here I come to what is perhaps the most
interesting of my new propositions, and one that is open to
experimentation on any reader's part. For it is through the combined
action of attention and will during dreams that one can take the first
steps in directing and modifying the course of dreams as one wishes.
Perhaps no one has experimented personally with dream control as much as
Alan Worsley, the inveterate lucid dreamer who can claim to be the first
to signal lucidity with eye movements in a sleep laboratory. First, a
comment from Worsley regarding voluntary and involuntary control in
dreams: "Non-lucid dreams use many principles that can be used in lucid
drams. For instance, it is likely, in a non-lucid dram, that if one
believes one looks into a book about a certain subject, one will find
relevant pictures in it. In lucid dreaming, one can use this principle
by deliberately selecting a book about a subject one wishes to study."
Worsley has tabulated his attempts to influence dreams. A complete table
of his results appears in Conscious Mind, Sleeping Brain, edited by
Gackenbach and LaBerge. He rates the difficulty of various tasks as
Hard, Medium, and Difficult, based on the percentage of times he was
able to succeed at them. For example, he finds all attempts to penetrate
dream matter with his dream body to be easy. Making sounds by striking
things or speaking is easy. Reading single words or short phrases is
easy, but reading long sentences is hard. He was never able to suddenly
turn on a light in a dark room, although he was able to do so easily in
a light room. Flying close to the ground was easy, and got progressively
more difficult the higher he would try to go.
EXPERIMENTAL STUDY OF DREAM CONTROL
The wide range of opinions on the topic of dream control, and the
reports we received of people's attempts to control dreams, piqued our
curiosity about why it is sometimes it possible to achieve a desired
outcome, and sometimes it is not. Because dreams are entirely illusory,
it should be possible to experience anything imaginable. Thus, perhaps
failures arise from not imagining strongly enough, or not believing a
certain experience is possible. On the other hand, perhaps there are
physiological limitations on the ability to control dream imagery.
Our theoretical approach to dreams is based on the idea that the
perceptual experiences in dreams arise out of activity in the same brain
areas that produce perceptual experience in waking. This is why people
have difficulty distinguishing dreaming from waking experience, and have
to employ special techniques to recognize when they are dreaming.
Physiological constraints on dream perception might occur if a certain
brain area is not in a state conducive to the desired experience. For
example, it might be hard to make a dark dream light, because the visual
cortex is not active enough. This is one of the topics of research we
would very much like to see explored: the relationship between dream
perception and brain activity.
The NightLight study was designed to assess how successful people would
be at accomplishing certain well-defined tasks in lucid dreams.
1. LIGHT SWITCH TASKS
A. Find a light switch (indoors).
B. Turn it on, and see what happens, then turn it off, and see what
C. Turn the lights on and off by willing it to happen and observe the
results. (These two tasks were counterbalanced so that some tried the
"magic" first and some second.)
2. MIRROR TASKS
A. Find a mirror.
B. Observe your reflection in the mirror.
C. Move your hand to your face, watching it in the mirror and observe
how the reflection behaves.
D. Pass through the mirror and see what is on the other side. (The
instructions gave an example in which the dreamer passes through the
mirror and ends up in a different scene.)
These tasks represented a variety of types of influence, ranging from
things that are easy to accomplish in waking (turning on a light,
looking in a mirror), to impossible in waking (passing through a
mirror). In addition, some tasks we thought might be impeded by brain
state were included (the changing of light level). The purpose of asking
people to both will a light on and off and switch it was to see what
effect belief might have on the outcome. It is easier to believe a light
will turn on when a switch is flipped than that will alone will turn on
The instructions asked the participants to try each task in waking prior
to attempting the tasks in lucid dreams, so that they would have the
procedure well-rehearsed. Then, they were to try each task at least once
in a lucid dream. They did not need to complete all the tasks in one
dream, but could use as many lucid dreams as they needed. So that they
would not forget details, the participants were asked to awakening
immediately after the experimental lucid dreams and make complete
reports of their experiences.
Twenty-seven people submitted reports of their attempts to carry out the
assigned tasks, fourteen women and thirteen men. Altogether they
provided 65 lucid dream reports, an average of 1.4 per person. The
maximum number of reports from one individual was four.
A judge reviewed the reports to determine which tasks were attempted in
each dream, and scored the result. The scoring for the tasks of finding
a light switch or a mirror was either as "success" or "failure." The
results of the actions of turning a light switch on and off, willing a
light on and off, looking at a reflection, touching hand to face, and
passing through a mirror received scores of "expected" if the result
achieved the goal, "no result" if the action produced no response, and
"unexpected" if something unpredictable happened.
"Expected" for turning and willing a light on and off meant that the
light went on and off as it would in waking. "No result" meant the light
did not change. "Unexpected" meant something other than the chosen light
turning on and off. Examples of unexpected light results were: "the bulb
slowly filled with what appeared to be thick, black tar," and "When I
threw the switch, the outside porch light came on instead of the room
light...didn't really increase the overall illumination."
For the task of looking at a reflection, a score of "expected" was given
when people reported that their reflections in the dream looked like
their reflections in waking. "No result" indicated that the person saw
no reflection. This happened once; the person instead saw gray, swirling
mist. If the reflection looked unlike the waking image, the result was
The same criteria as for the reflection applied to moving the hand to
the face while watching in the mirror. An example of "unexpected" for
this task was, "As I raise my hand to my face I see the reflected image
of my hand go up but from then on I notice an increasingly
'hallucinogenic' breakdown of the image--such things as my finger,
detached from my hand, disappearing into my mouth and holes appearing in
The result of trying to pass through a mirror was scored as "expected"
if the dreamer was able to move through the mirror and found another
setting on the other side. "No result" meant that the dreamer found the
mirror hard and unyielding, as in waking. "Unexpected" applied to cases
in which the dreamer got through the mirror, but was then somewhere
unlike what was described in the task instructions, for instance, in the
same room, or "in a world of cartoon-like images."
The table [below] shows the results of the participants' tries at the
tasks. The left-hand columns list the number of people who attempted
each action and the number who achieved each kind of result. The right
hand columns display the total number of tries at each task, and the
number leading to each result.
Looking at the number of "expected" results, that is, cases in which the
action produced the desired result, it appears that "willing" a light to
turn on or off and using a dream switch are about equally easy. There
seemed to be more cases of "no result" with willing the light on, but
the difference did not pass a statistical test. The data hint that it
may be easier to get a dream light to turn off than on. However, this
conclusion may be premature, given that in the majority of cases, before
trying to turn off a light, the person had already succeeded in turning
it on. There may be a condition in which if you can turn on a light in a
dream, you can also turn it off.
Clearly, it was very easy to find things in dreams that are usually
around in waking, like a light switch and a mirror. People also had no
difficulty performing the normal action of looking in a mirror and
seeing a reflection, although it was more likely than not somehow
different than the usual waking reflection, and in 12 cases (28%) the
image transformed as the dreamer watched. This happened for 41 percent
of the participants.
It would be reasonable to predict that passing through a mirror to
another scene would be the most difficult task, given that it is
impossible in waking life. However, almost half of those who tried
succeeded, and 86 percent of the people were able to get their dream
bodies through the dream mirror at least once, even if they did not end
up in a new scene. An example of an "expected" mirror result was, "I
then went through the mirror and tried to imagine that the mirror was
like water so I could easily slide through it. When I was fully through
the mirror, I came up to the surface of the water I was in and noticed I
was in a bright, sunlit backyard swimming pool with a roof shelter over
DREAMS CAN BE DIRECTED, BUT STILL DO THEIR OWN THING
The lack of large differences in the ease of accomplishment of the
various tasks is in itself quite interesting. Lucid dreamers are able to
exert a large amount of control over their dream experiences. But, it is
far from perfect. Most notable is the reluctance of the mirror
reflections to show normal images, and their fascinating instability.
Self-image is of course a very psychologically loaded thing, probably
with very complex internal representations. This may account for the
strange images. The instability points up the most prominent difference
between waking and dreaming perception. Dreams change. We exploit this
in lucid dream induction training by instructing people to examine
written phrases repeatedly, watching for them to change. An interesting
question is whether the perceptual instability results from the lack of
anchoring sensory input from the physical senses or from a state of the
brain peculiar to REM sleep.
In their studies, the Marquis d'Hervey de Saint-Denys and Alan Worsley
observed something they called the "light switch" phenomenon. This was
an inability to change the illumination of a room on demand. From this
study, it seems that this phenomenon is sometimes present and sometimes
not. Some people who were able to turn on lights reported no concurrent
change in general illumination, but others reported that there was an
increase in brightness—about half and half. So, "the light switch
phenomenon" is not dead, but merely seems to be sleeping some of the
time. A prime target for research would be to discover what the brain is
doing under both circumstances.
It was rather remarkably easy for people to pass through a mirror and
find something else on the other side. One might think, along the lines
of the quote earlier from Ullman and Zimmerman, that a complete scene
change would be a difficult thing to accomplish. In fact, we have
already seen in another study that it is more than possible. In the
results from an experiment published in the April 1987 issue of Omni
magazine, 51 people reported trying to arrive a particular pre-selected
target by spinning in a dream. Eighteen of them (35%) succeeded in
arriving at their target. Thus, not only is it possible to create a new
scene, but also to create one that is specifically desired.
In the final analysis, the Tibetan Buddhist view that all dream images
are transmutable may be exactly right. If so, we wonder if it may be
possible also to learn to control the stability of these images,
creating lasting dream scenes and objects, achieving a state of virtual
reality far beyond the wildest dreams of the computer programmers.
TABLE 1. Results of attempts to control dream content