Consciousness and Cognition 17 (2008) 12281237

Contents lists available at ScienceDirect

Consciousness and Cognition

journal homepage: www.elsevier.com/locate/concog

Do we only dream in colour? A comparison of reported dream colour

in younger and older adults with different experiences of black
and white media
Eva Murzyn
School of Psychology, University of Dundee, Dundee DD1 4HN, UK

a r t i c l e

i n f o

Article history:
Received 14 December 2007
Available online 8 October 2008

Keywords:
Dreams
Colour
Methodology
Age differences

a b s t r a c t
This study aimed to nd out whether differences in the reported colour of dreams can be
attributed to the inuence of black and white media or to methodological issues. Two age
groups, with different media experience, were compared on questionnaire and diary measures of dream colour. Analysis revealed that people who had access to black and white
media before colour media experienced more greyscale dreams than people with no such
exposure, and there were no differences between diary and questionnaire measures of
dream colour. Moreover, there were inter-group differences in the recall quality of colour
and black and white dreams that point to the possibility that true greyscale dreams occur
only in people with black and white media experience.
2008 Elsevier Inc. All rights reserved.

1. Introduction
The question of whether people dream in colour or black and white, while not central to dream research, is an interesting
one to study. This is because it touches upon the issues of how experience and beliefs can change what we feel to be a fundamental aspect of our life, namely visual imagery and dreaming. Interest in this question has been recently revived when a
surprising inconsistency in the results of the early and later studies was discovered by Schwitzgebel (2002). The research
conducted in the early 20th century unanimously concluded that the vast majority of people dream in black and white.
For example, Bentley (1915) reported that 20% of dreams contain colour; in 1942 only about 29% of college students reported
having at least occasional coloured dreams (Middleton, 1942; see also de Martino, 1953, and Middleton, 1933). The proportion of people reporting coloured dreams even decreased in the 1950s: Knapp (1956) claimed that as little as 15% of dreams
contain colour, while Tapia, Werboff, and Winokur (1958) found that only 9% of people who reported to a hospital in St. Louis
for non-psychiatric medical problems remembered having coloured dreams. Moreover, this gure was contrasted with a 12%
rate of reporting coloured dreams among psychiatric inpatients in the same hospital and the researchers concluded that vivid and coloured dreams may be a sign of psychological problems. Overall, researchers and study participants agreed that
black and white dreams were the norm, and rare cases of coloured dreams were dubbed Technicolor dreams (Calef,
1954; Hall, 1951), highlighting their perceived articiality.
This tendency to report black and white dreams suddenly disappeared in the 1960s. Kahn, Dement, Fisher, and Barmack
(1962) wrote that with careful interrogation close to the time of dreaming, color was found to be present in 82.7% of the
dreams and Herman, Roffwarg, and Tauber (1968) discovered that coloured dreaming was reported after 69% of REM awakenings of their subjects. Similar results were reached in studies carried out by Berger (1963), Jankowski, Dee, and Cartwright
(1977) and Snyder, Karacan, Tharp, and Scott (1968). Most recently, Schwitzgebel (2003) replicated Middletons (1942) study
and found that only 17.7% of US college students say they rarely or never experience coloured dreaming. Interestingly, early
E-mail address: e.murzyn@dundee.ac.uk
1053-8100/$ - see front matter 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.concog.2008.09.002

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

1229

descriptions of dreams and treatises on the nature of dreaming suggest that colour was commonly present in dreams before
the 20th century (Schwitzgebel, 2003).
There are two major differences between the two sets of studies that can possibly explain the changes in reporting of coloured dreaming. The rst one is related to the cultural background. When the rst studies were conducted, black and white
cinema (and later TV) was already quite widespread. At the time Bentley had carried out his rst study in 1915, over 20 black
and white feature lms were produced every year. It was very likely that the average college student (the typical participant
in these studies) had regular contact with black and white media. The rise in coloured dreaming, on the other hand, coincided with the rise of coloured media. In late 1940s colour movies began to be more common and by late 1960, nearly
all movies were produced in colour. The rst colour TV shows were broadcast in 1950 (to be viewed in public places) and
the rst consumer colour TV sets appeared in 1954 and by 1972 the majority of USA households had a colour TV. Thus, it
is possible to stipulate that the period of reporting greyscale dreams was caused by intense black and white media exposure.
Contemporary research that supports that theory has been already carried out by Schwitzgebel, Huang, and Zhou (2006).
Their replication of the Middleton (1942) questionnaire, carried out in China, revealed that groups with more exposure to
black and white media report less coloured dreaming, at levels comparable to the original 1942 study. Contrastingly, exposure to coloured media before the age of 11 was strongly correlated with reporting of coloured dreaming.
There are two possible ways in which black and white media could have created the divide between early and later studies. The rst one is through actually modifying the form of dreams. Although this line of reasoning might seem improbable at
a rst glance, it is nonetheless worth examining. While the content of dreams is not immediately tied to the events of the
previous day (Roussy et al, 2000; Schwartz, 2003) it is strongly inuenced by current concerns and activities (Cartwright,
Agargun, Kirkby, & Friedman, in press; Epstein, 1985; Erlacher & Schredl, 2004; Schredl & Erlacher, 2008; Schredl & Hofmann,
2003), and media (Den & Bulck, 2004). It is conceivable that repeated exposure to black and white media could have an impact on the form of dreams, especially that movies are emotionally intense and involving, which could promote the incorporation of colour schemes into dream production (and would explain why static media such as paintings and photography
did not apparently impact dreaming).
The second explanation requires that black and white media only inuence peoples beliefs about their dreams, without
changing the dream form, so that people would not report the true colouration of their dreams. There are two variants of this
explanation. The weak proposition places the distortion source in the poor long-term memory for dreams. Thus, under a casual examination, people would indeed claim to experience greyscale dreams, and only when questioned closer to the time
of dreaming they could realise their dreams are, in fact, coloured. The strong proposition states that the distortion imposed
by beliefs in the nature of dreaming is implemented much earlier and reconstructs the memories to match the beliefs. In
such a case, there is no way to say anything about the true form of dreaming, except through methods that would tap into
the dream content directly. One possible method would be signalling from within a lucid dream to inform external observers
about the colour nature of the dream (see LaBerge, 1985, for examples of similar studies). However, this assumes that dream
lucidity does not interfere with normal dream form and content, which has not yet been established. While this second possibility seems unlikely, it needs to be considered nonetheless.
The second explanation for the difference in reporting of coloured dreaming is the methodology employed by the
researchers in the two sets o studies. Six out of nine studies in the early 20th century employed questionnaires to assess
how people dream. This method is fraught with problems: the questions are typically asked a long time after a person
has had a dream, which can lead to omissions and misremembering of various aspects of the dream. Having no reliable memories of dreams to inform their answers, people might draw on their beliefs of what dreams should be like. Consequently
they might give responses that are not related to how they actually dream, but to how dreams are perceived in that cultural
milieuwhich revolved around black and white media for the rst half of the 20th century. Also, people were usually asked
to summarise their dream experiences from the last week or month, and these would often be heavily inuenced by the few
most recent or well-recalled dreams. Thus, peoples accounts of how they dream might not be representative of their actual
dreams.
On the other hand, the studies conducted after 1960 took advantage of the discovery of REM sleep and its relationship
with dreams, and were thus methodologically more reliable. Three of the studies (Herman et al. 1968; Jankowski et al.
1977; Kahn et al. 1962) used REM-awakenings, where people were monitored in a sleep lab, woken up during an REM episode and immediately asked to describe their dreams. The remaining four studies used dream diaries, where people were
asked to record their dreams just after spontaneous awakening. While this method is not without its shortcomings (for
example, only some of the dreams are remembered and dream saliency and intensity play a role in what is recalled in
the morning), it yields results comparable to REM awakenings and is much easier to administer to a wider population
(see Domhoff & Schneider, 1999 for a discussion). Importantly, the rates of coloured dreaming arrived at with the two methods were similar, and markedly higher than for previous questionnaire studies. Overall, the use of these techniques made the
new nding more resistant to memory distortions caused by attitude or beliefs. Thus, nding out about the real form of
dreams became much more likely. This line of investigation meshes with the shallow media inuence hypothesisif black
and white media impacted only peoples beliefs about what dreams should look like, a change in methodology would uncover the true form of dreams.
Nonetheless, there is no perfect mapping between research method used and the rate of coloured dreaming achieved. Late
20th century questionnaire studies have shown majority of people claim to have coloured dreams. Early 20th century psychoanalysts, who trained their patients in careful attention to dream detail (and thus presumably overcame many of the lim-

1230

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

itations of the non-direct method), reported that most dreams are devoid of colour, and Garma (1961) even went as far as to
say that Dreams are like the old silent lms, without sound or technicolour.
The present study was designed to investigate whether the differences in reported frequency of greyscale dreaming can
be attributed to the differences in methodology used to nd out about dream colour or to the actual differences in the type of
dreaming experienced. This aim was achieved by comparing two cohorts with presumably different experience of coloured
and black and white media: people under 25 years of age (similar to the groups previously studied) and people over 55 years
of age. A dream diary method was adopted to nd out about the colour qualities of dreams.
The potential problem with using an older population lies in the different sleep and dreaming characteristics of this
group. Fein et al. (1985) found that elderly people (aged 6974) recalled dreams less often than young participants (in
71% of REM awakenings, as compared to 90% for younger group) and the dream reports of older participants contained signicantly less visual imagery words. Waterman (1991) also found a correlation between dream length and visual memory,
which decreases with age. Older people were also found to value their dreams less and recall dreams less frequently (Herman & Shows, 19831984) although the drop-off in recall frequency appears at the age of 2535, and is apparently not related to aging (Funkhouser, Hirsbrunner, Cornu, & Bahro, 1999). Because of the apparent age-related decrease in visual
imagery in dreams, any age-differences in the recall of colour in dreaming have to be taken cautiously, since poor visual recall can lead to mislabelling dreams as greyscale.
If media experience is indeed responsible for reports of greyscale dreaming, the following predictions can be made: people who were exposed to black and white media should report more greyscale dreams than people with no such exposure.
This difference should be evident both in dream diaries and questionnaires. The main difculty with this hypothesis in the
present design is that it makes the supposition that people who had such experience, and now are over 55 years of age, have
retained at least some of their greyscale dreaming patterns despite a long and intense colour media exposure. This is not
completely unlikely if a developmental approach to dreaming is considered. Since dreams are in development until around
12 years of age (Foulkes, 1999), it is possible that these early years are important in setting the form of dreams. Thus, a secondary hypothesis can be stated: people over 55 will report still having greyscale dreams only if childhood exposition to
media is more important than adulthood exposition.
If methodological and belief issues are the only reason why people reported black and white dreaming, different predictions can be made. Firstly, people who claim to have black and white dreams should report them more frequently when
tested with a questionnaire than when asked directly after the dream takes place. Secondly, there should be age differences in the type of dreaming that is reportedbut only for questionnaires. This is because people brought up on black
and white media might harbour the concept that dreams should be in black and white, and consequently will claim to
have greyscale dreams. This misconception should be cleared (at least to some extent) by administering a more direct
measure of dream colour. In parallel to the previous set of hypotheses, older people will only report greyscale dreaming
if the assumed beliefs regarding the nature of dreaming are quite stable in time. Finally, since misconceptions are more
likely to arise when memory falters, poorly recalled dreams might be labelled as black and white more often than wellrecalled dreams, because of a lack of detailed visual recall from the dream. Caution is needed here, however, because any
differences in memory quality between coloured and greyscale dreams might also stem from the intrinsic qualities of
these dreams.
2. Method
2.1. Design
The main variables investigated were the frequency of coloured and black and white dreams, measured with a questionnaire and a dream diary, as well as dream recall quality. The independent factors were age, experience with colour and black
and white media and attitude towards dreams. Attitude towards dreams was considered important because in previous
studies it has been found to relate to reported dream recall frequency (Beaulieu-Prvost & Zadra, 2005) and quality (Wolcott
& Strapp, 2002).
2.2. Participants
Participants were recruited through the University of Dundee and community centre advertisements as well as by contacting volunteers in existing research databases. All participants knew they were recruited for a study of dreams and dream
colour, but they were not told about the specic hypotheses. This way of recruitment has the potential to introduce a bias
into the results, as volunteers will be more interested in dreams than an average person, and possibly have a higher dream
recall frequency. However, a higher dream recall frequency would be in fact benecial for the present study. If greyscale or
mixed colour dreams are infrequent, obtaining more dream samples from each person will reduce the chances of missing
these dreams because of a narrow 10-day testing period. Also, recall frequency is positively correlated with better dream
detail recall (Wolcott & Strapp, 2002) which can be benecial to the present study.
There were 30 males and 30 females in the study; half of each group were under 25 years of age (mean = 21, SD = 2.5) and
half were over 55 years of age (mean = 64, SD = 8.0). The volunteers were paid 8 for their participation.

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

1231

2.3. Measurements
2.3.1. Estimated dream colour type
Participants were asked to estimate what their dream colour type is prior to starting the dream diary. They could choose
one of the following options: colour, black and white, both, neither and dont know.
2.3.2. Media experience
Exposure to coloured and black and white media was estimated with a short questionnaire, adapted after Schwitzgebel
(2006). It probed the age of rst regular exposure to black and white media and coloured media (on a following scale: 03
years old, 46, 710, 1114, 15 years and over, never), number of hours spent currently watching TV, and percentage of programmes watched in black and white. An additional nominal variable which represented exposure to black and white media
only was arrived at by subtracting the age category at which colour media were introduced from the age category for rst
access to black and white media and collapsing the result to either 0 (no exposure to black and white media prior to colour
media) or 1 (exposure to black and white media prior to colour media).
2.3.3. Dream colour type
This variable was measured with a dream diary. Every morning for 10 days the participants recorded the number of
dreams they remembered and answered six yesno questions about the colour qualities of each dream. The question structure of the diary was designed to minimise the chances of people mistakenly reporting dreams with unrecalled colouration
as black and white and to avoid assuming the colourfulness of a dream without remembering particular colour instances.
The answers to these questions would later allow classifying the dream into one of four main categories: colour, greyscale,
mixed (containing both coloured and greyscale elements) and neither. Two additional categories were also includedone
for dreams where the answer pattern was inconsistent with any interpretation (unclear type) and one for dreams where participants could not answer any of the questions concerning colour in their dreams (no recall type). Two types of variables
were extracted from this data: one which classied how each participant dreams in generalonly colour, only black and
white or both; and the other was a set of 6 variables which indicated how many dreams (in percent of the total dream count)
a participant experienced in any of the above categories.
2.3.4. Dream recall quality
This set of variables was also measured with the dream diary. The participants rated how well the remembered the ve
dream constituents identied by Hall (1953), namely the characters, plot, setting, actions and visual scenes, for each dream
they recalled. The ve separate ratings were used because breaking up dream recall into smaller and better dened categories was thought to yield more accurate results. Additionally, more specic comparisons between memory for components
could be made. The ratings were made on a one to ve scale, where 1 meant no recall of that aspect, and 5 meant the person
remembers it almost as well as a waking event. General recall quality of a dream was obtained by averaging the ratings for
the ve components. Both individual dream ratings and ratings averaged across dreams were used in this study.
2.3.5. Attitude to dreams
Attitude to dreams was measured with a Likert scale which consists of eight items, each rated on a 1 (completely disagree)
to 5 (completely agree) scale. There are ve positively and three negatively worded items, three of these items reect the
behavioural aspect of attitude, three reect the affective component and two look at the cognitive component. Cronbachs
alpha coefcient for the scale items is 0.80. Attitude score was calculated by reverse scoring of the negative items, adding
all the item scores and dividing by the number of items to achieve a score with a maximum of 5 and minimum of 1. A higher
score reects a more positive attitude towards dreams, while a lower score reects a neutral or disinterested attitude.
2.4. Procedure
The session started with administering the media experience and dream colour questionnaire. After this, the author presented the structure of the dream diary, explaining its features and making sure the participant knows how to interpret and
answer the questions. The participants were informed to ll the diary as soon as possible after waking up, and were told to
keep it for ten consecutive days. They were also informed that there was no pressure on them to provide many dream reports
and that no dream recall was also a valuable result. The diaries were returned either in person or by post and the participants
was paid 8 on returning the diary.
3. Results
3.1. Analysis
Two types of analyses were conducted. The rst one investigated individual differences in the frequency of coloured and
greyscale dreaming and looked at possible relationships with age and media experience, through a set of independent t-tests.

1232

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

The second one employed analysis of variance and concentrated on the properties of individual dreams and investigated
whether there are any differences in recall quality between different colour types of dreams. Both recall of visual scenes
and general recall quality of the dream were used in the analysis, since people are better at encoding and retrieving coloured
visual stimuli (Spence, Wong, Rusan, & Rastegar, 2006; Suzuki & Takahashi, 1997) but there is more to the recall of dreams
than just the visual aspects.
3.2. Media experience
Analysis of the dream recall and media experience questionnaire revealed the predicted age differences. Among the younger age group, all participants had frequent access to colour TV and lms by the age of 6 (with 21 participants gaining such
access before the age of 4). In contrast, none of the participants in the older group had access to coloured media before the
age of 7 (with 17 participants gaining such access only after the age of 15). Moreover, 8 participants had black and white
media access by the age of 6, and further 19 by the age of 14. Most importantly, in the older sample there was a subgroup
of 22 people who had had gained access to black and white media before they experienced coloured ones.
3.3. Attitude towards dreams
There was a statistically signicant age difference in attitude towards dreams (t(58) = 4.90, p < .001).: the average score in
the younger group was 3.9 (SD = 0.59), while the older groups score was 3.2 (SD = 0.63). While people in the younger group
were more favourably inclined towards dreams than people in the older group, both groups scored close to the positive end
of the scale. In further analyses attitude to dreams was found to be correlated with variables that displayed age differences
(such as dream recall quality). However, since there were no corresponding intra-group relationships, attitude towards
dreams was not used in subsequent analyses.
3.4. Frequency and quality of dream recall
There were no signicant age differences in the average number of dreams experienced during the testing phase, with the
younger and older groups reporting a mean of 11 dreams in the 10-day period (SD = 4.6 and SD = 6.7, respectively). People,
on average, report one dream every two days (Goodenough, 1991), so this can be seen as evidence of a volunteer bias or of
situational demands. Nonetheless, it is important to keep in mind peoples accuracy in reporting the number of dreams has
never been examined (for various methodological reasons). There were some statistically signicant (t(58) = 3.33, p = .001)
age differences in the quality of dream recall: the general recall quality in the younger group was 3.1 (SD = 0.58) as compared
to 2.5 (SD = 0.69) in the older group. This was also the case with the specic measure of memory for visual aspects of the
dream (t(58) = 2.58, p = .012), since the younger group rated their memories as better than the older group (mean of 3.1 as
compared to 2.7). This result is in line with previous research on age-related differences in dreaming and is important because it suggests that any age differences in recalled dream colour can be also attributable to visual memory issues.
3.5. Questionnaire dream colour versus dream colour as examined by dream diary
Analysing the questionnaire item concerning participants beliefs on whether they dream in colour, black and white or
otherwise yielded the age differences predicted by both sets of hypotheses. In the young group, 21 participants indicated
that they dream in colour, 6 indicated they have both colour and greyscale dreams and 3 did not know how they dream.
In contrast, only 8 people in the older group indicated they dream in colour, 4 said they only have greyscale dreams and
12 mentioned having both types. 5 people from the older group did not know how they dream, and 1 person indicated having dreams neither in colour nor greyscale.
The dream diaries revealed a somewhat different pattern. In both younger and older age groups the number of people
reporting only coloured dreams decreased and the number of participants reporting having both colour and greyscale dream
increased. No people had only greyscale dreams. However, a McNamara related groups test revealed that there were no signicant differences in the reporting of coloured dreams (N = 50, p = .290) and greyscale dreams (N = 50, p = .450) between the
questionnaire and the dream diary. Importantly, there were also no signicant differences when individual age and media
experience groups were examined. In the young group 7 participants reported at least a single black and white dream (as
compared to 6 who initially stated they have both colour and greyscale dreams); in the older group 17 participants had
at least a single greyscale dream (as compared to 16 in the questionnaire). Thus, the number of people reporting greyscale
dreaming has not changed when a different research method was employed.
An additional analysis of peoples accuracy of dream type assessment was performed by comparing the dream type as
claimed in the questionnaire with the dream type obtained from the 10-day sample of dreams in the diary. As no age-related
differences were found, the results are presented for the whole participant set. The questionnaire data were in accordance
with the diary data for 75% of the participants for reporting of colour dreaming, and for 70% of participants when greyscale
dreaming was examined. Table 1 shows more details
It is important to note that it is likely that in this short study period the participants might not have had the chance to
experience particular dream types, especially if they happen rarely. On the other hand, there is the issue of consistency bias.

1233

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

Table 1
Accuracy of questionnaire responses when compared to dream diary data
Percentage of people who

Colour dreams (%)

Black and white dreams (%)

Correctly reported
Reported a false positive
Reported a false negative

75
6.7
3.3

70
8.3
3.3

Since the participants answered the questionnaire before lling out the dream diary it is possible that they were inuenced
to conform to their original dream colour choice when lling out the diary. Thus, this level of consistency is not decisive.
3.6. Age, exposure to media and dream colour
Two sets of independent t-tests were carried out, with age (under 25 and over 55) and black and white media exposure
(had access before or after access to colour media) as the grouping factors. The dependent variables were the percentages of
dreams dreamt in colour, greyscale, mixed, neither and unknown colours. Table 2 shows the means and standard deviations,
along with the results of the t-tests for the two age groups. During the period of the study people over 55 had experienced
signicantly less coloured dreams than people under 25, but they had signicantly more mixed and greyscale dreams.
Finally, comparisons within the over 55 age group were carried out with another series of independent t-tests. The independent variable was experience with black and white media before such access to coloured media. The group with black
and white media experience (n = 22) reported signicantly more greyscale (t(28) = 2.35, p = .030) and mixed dreams
(t(28) = 4.53, p < .001) and less colour dreams (t(28) = 2.75, p = .010) than the group with no such experience (n = 8). At
the same time, the group with no early black and white media experience was somewhat older than the group with such
experience (67.75 and 62.86, not statistically signicant, t(28) = 1.12, p = .29) and there were no differences in quality of
dream recall. Thus, reporting of greyscale dreams cannot be attributed solely to age-related visual memory problems.
3.7. Analysis of dream colour and dream recall quality
A pair of two way ANOVAs was carried out with access to black and white media and dream type (colour, mixed and greyscale) as independent variables and with individual dreams visual recall and general recall quality as the dependent variables. Planned contrasts were carried out for the interaction discovered. Dreams labelled as neither and unrecalled were
not included because of very low recall scores; unclear dreams could not be included because of classication issues.
Dreams with the average recall rating lower than or equal to 1.5 (white dreams) were removed from the analysis, as any
attributions of colour qualities would have been unreliable. 66 dreams (12%) were removed from the set, 15 from the younger age group and 51 from the older group, resulting in a nal set of 475 dreams.
The group sizes, means and the standard deviations of visual recall quality and general recall quality for each of the group
types are shown in Table 3.
3.7.1. Visual recall quality
The assumption of homogeneity of variance was satised (Levenes F(5, 469) = 1.53, p = .178).
The two way ANOVA detected a signicant main effect of dream type (F(2, 469) = 11.70, p < .001, effect size = 0.05) but the
main effect of media access was not statistically signicant (F(1, 469) = 1.15, p = .284).
A signicant interaction between dream type and age group was discovered (F(2, 469) = 3.51, p = .031, effect size = 0.02).
Planned contrasts revealed that while the effect of dream type was signicant for both groups examined (F(2, 469) = 6.91,
p = .001 for group with no access and F(2, 469) = 5.40, p = .005 for group with access to B&W media), there were some small
differences. In the group with no black and white media exposure the visual aspects of coloured and mixed dreams were
remembered equally well (p = .789), but were remembered signicantly worse in greyscale dreams (p < .001). In the group
with access to black and white media, the visual aspects of coloured dreams were recalled signicantly better than of mixed

Table 2
Age differences in relative frequencies of different colour types of dreams
Age group

%
%
%
%
%

Of
Of
Of
Of
Of

colour dreams
greyscale dreams
mixed dreams
dreams neither colour nor greyscale
dreams with unrecalled colour

df

Under 25 (n = 30)

Over 55 (n = 30)

Mean

SD

Mean

SD

68.4
4.41
8.74
1.86
15.1

31.8
9.01
13.8
4.94
23.7

33.9
22.2
22.2
1.80
17.7

36.9
27.1
15.7
5.97
25.04

58
35.3
58
58
58

3.88
3.40
2.10
0.04
0.40

<.001
.001
.040
.960
.690

1234

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

Table 3
Group sizes, means and the standard deviations for visual recall quality and general recall quality
Group

Dream type

Visual recall quality

General recall quality

Mean

SD

Mean

SD

No early black and white media experience

Colour
Mix
Black&White
Total

289
30
12
331

3.59
3.53
2.42
3.54

1.07
1.22
0.79
1.09

289
30
12
331

3.39
3.41
2.42
3.36

0.96
1.12
0.87
0.99

Early black and white media experience

Colour
Mix
Black&White
Total

49
47
48
144

3.76
3.15
3.13
3.35

0.99
1.16
1.02
1.09

49
47
48
144

3.28
3.12
3.08
3.16

1.11
1.12
0.96
1.06

Total

Colour
Mix
Black&White
Total

338
77
60
475

3.61
3.30
2.98
3.48

1.06
1.19
1.02
1.09

338
77
60
475

3.37
3.23
2.95
3.29

0.99
1.12
0.97
1.02

and greyscale dreams (p < .001). The effect of media experience, however, was signicant only for greyscale dreams
(F(1, 469) = 4.21, p = .041), with people who had contact with black and white media remembering the visual details of such
dreams signicantly better than the other group.
This means that while both groups are better at recalling the visual aspects of coloured dreams, the group that has had
experience with black and white media rates their visual memories of greyscale dreams higher than the group without such
exposure. Fig. 1 shows the pattern of the interaction.
3.7.2. General recall quality
General recall quality was calculated by averaging the recall ratings for dream characters, scenes, actions and plot. Thus, it
represented the memory of all non-visual aspects of the dream. When using the general recall quality as the independent
variable in the ANOVA, Levenes test for equality of variances indicated that the error variances across groups were heterogeneous (F(5, 469) = 2.35, p = .040). However, a more detailed analysis revealed that this was due to a single outlier cell with

Fig. 1. Mean dream recall quality for different colour types of dreams and different levels of black and white media experience.

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

1235

a low mean and very low variance, and there was no correlation between cell means and standard deviations. Since Lindman
(1974) has demonstrated that the F statistic is quite robust in these cases, it was decided to proceed with the analysis of
variance The two way ANOVA detected a signicant main effect of dream type (F(2, 469) = 5.36, p = .005, effect size = 0.02)
but the main effect of media access was not statistically signicant (F(1, 469) = 0.37, p = .543).
There was a signicant interaction between dream type and media access (F(2, 469) = 3.04, p = .050, effect size = 0.01),
which was broken down with planned interaction contrasts. The effect of dream type was signicant only for the group with
no black and white media access (F(2, 469) = 5.51, p = .005), but not for the other group (F(2, 469) = 0.53, p = .589). Thus,
while participants with no black and white media access recalled colour and mixed dreams signicantly better than greyscale dreams, people who have had such access recalled all three types of dreams equally well. Furthermore, the effect of
media access was signicant only for greyscale dreams (F(1, 469) = 4.17, p = .042), but not for coloured (F(1, 469) = 0.53,
p = .465) or mixed dreams (F(1, 469) = 1.50, p = .221), with the recall of greyscale dreams being superior in the group with
black and white media access. Fig. 1 illustrates this interaction.

4. Discussion
The present study has investigated whether black and white dreaming can be explained by media inuence or research
methodology. The data provide reasonable support for the media inuence as a source of black and white dreams. First of all,
in agreement with the rst hypothesis, signicant and robust age differences in frequency of black and white dreams (as
measured with the dream diary) were discovered in the two age groups examined. At this point it can be argued that memory and sleep structure are important age-related factors that can inuence how the colour properties of dreams are recalled.
However, the differences in frequency of greyscale dreaming were still present when comparisons within the older age
group were made. Thus, it is possible to conclude that early access to black and white media (without any access to colour
media) is a key factor in the reporting of black and white dreams. It is worth noticing that even the group with black and
white media experience, the average percentage of greyscale dreams experienced in this experiment was much lower than
the proportions typically reported in the 1940s and 1950s. This difference is most likely due to the inuence from colour
media, which has been the dominant media type for at least the last 40 years. Overall, it seems that the form of dreams is not
xed in stone during that period.
The second set of comparisons was designed to see whether a dream diary, being a more immediate measure of recalled
dream colour, will yield different results than the questionnaire. It was found that there were no differences between the
questionnaire and the dream diary in the rates of reporting black and white dreaming, for either age or media experience
group. This result can bolster the condence in the early 20th century questionnaire studies of dreaming, by showing that
people can provide reliable information in questionnaires. On the other hand, there is a possibility that the diary method is
not immediate enough to counteract memory and selective recall issues (especially in the older population), and that this
convergence of methods is an artefact of a conrmation bias.
Finally, the last hypothesis that black and white dreams should be remembered less well than colour dreams has been
validated to some extent in the group with no access to black and white media, but not in the other group. These analyses
reveal two interesting effects. As far as only visual memories of dreaming are concerned, coloured dreams are uniformly better recalled than black and white dreams, although people who have had access to black and white media have better memories of black and white dreams than people with no such access. This is consistent with the previous ndings of better
encoding and recall of coloured pictures (Spence et al., 2006; Suzuki & Takahashi, 1997) and likely due to the extra detail
colours provide. It is quite interesting to see this effect in dreams. The nding that there are media access dependent differences in visual memory for greyscale dreams can be interpreted in two ways: either early exposure to black and white content enables people to encode and retrieve black ad white materials more efciently or there might be some group
differences in the experience of greyscale dreaming. The latter explanation is strengthened by the results of the second analysis. When the general memory of a dream is examined, there are no differences in recall quality between colour and black
and white dreams for the group with access to black and white media. However, in the group with no early access to black
and white media, coloured dreams are recalled signicantly better than black and white dreams, and at the same time there
are no group differences in recall for coloured dreams. These data are consistent with the hypothesis that people who have
not experienced black and white media might label poorly remembered dreams as greyscale. This result can be interpreted
in two ways: either people who claim to have greyscale dreams but have not had experience with such media are simply
mislabelling poorly recalled colour dreams or people with early black and white media access misremember the presence
of colour in their dreams more easily than people without such experience. This second option could be linked to different
expectations and beliefs about dreaming.
Another unexpected effect was found for mixed colour dreams. These are supposed to be dreams in which only a couple
of objects are coloured, and the rest is in greyscale. People who have not had sole access to black and white media, recall
mixed dreams in a pattern similar to coloured dreams while people with such access recall mixed dreams in the same
way as black and white dreams. A possible interpretation for this discrepancy is a group difference in what is labelled as
mixed colour dream. It seems that in the former group, when only some of the colour information is recalled, colour dreams
might be labelled as mixed, because a lack of memories about the colour qualities of the background is interpreted as
greyscale background. The other group, on the other hand, might misremember some black and white dreams or insert a

1236

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

coloured object in a central place of the dream. Alternatively, one of these groups might experience genuine mixed colour
dreams, and one of the ways of nding that out would be to study how mixed colour pictures are recalled when compared to
colour and black and white pictures.
Overall, while the results of this study strengthen the media access explanation, they do not allow to fully disambiguate
between the dream experience (people do experience greyscale dreams) and belief (people only believe they have greyscale dreams) variants. The administration of the questionnaire before the dream diary created the possibility that the agreement of diary and questionnaire data (predicted by the experience variant) is an artefact of a conrmation bias or
inadequateness of a diary measure. The dream recall quality analysis can also be interpreted as supporting both variants.
The dilemma lies in deciding which outcome is more probable: a short period of early and intense exposure to black and
white media causing a relatively stable shift in the experience of dreaming, or the exposure to black and white media
and the cultural attitudes causing the formation of equally stable beliefs that prevent people from ascertaining how they
actually dream. On one hand, it is difcult to see how a few hours of childhood black and white media exposure could have
such a profound impact on the colour of dreams, especially since waking life is universally coloured. On the other hand, why
would only people with such early exposure absorb the belief that dreaming should be black and white, when this sentiment
was present throughout the whole culture, and was not particularly emphasised in the media? And why would they retain
this belief despite the dramatic changes in the general beliefs about dreaming? Without a REM awakening study that would
probe the colour of peoples dreams more immediately than a dream diary it seems impossible to decide between the two
explanations.
Other potential limitations of this study lie in the fact that the previous research consistently found marked age differences in dream recall frequency (Funkhauser et al. 1999; Giambra, Jung, & Grodsky, 1996), which have not been found in
this study. This discrepancy might be either due to the relatively low mean age in the older group or a sampling bias that
favoured people with overall good dream recall. However, it does not seem that this could pose serious problems for the
validity and generalisability of the results. After all, the purpose of this study was to nd why some people report greyscale
dreams, and not to establish population rates of greyscale and coloured dreaming. Another issue is related to the high variances observed in the relative frequencies of various types of dreaming even when looking at separate groups (with media
experience as the grouping variable) the standard deviations were typically very high. This was not due to the impact of single outliers, but rather to a slightly bimodal distribution of frequencies.
The discovered relationship between dream colour and memory for the visual aspects of the dream is also interesting.
Previous research found that presence of colour in visual stimuli (such as pictures) improves peoples recognition memory (Spence et al., 2006; Suzuki & Takahashi, 1997). This effect is located in the encoding stage and colour understood to
play the role of additional detail which facilitates encoding and later retrieval through multiple cues. Precisely the same
effect has been found in the present study when visual memory (understood as recall of picture or movie like elements) of dreams is considered, with coloured dreams remembered better than black and white dreams. This result
can be seen as an indication that dreams (and their memories) are processed in the same way as other sensory data.
On the other hand, poor memory for visual details of a dream might have prevented people from labelling a dream
as a coloured one. In this design, the participant needed to remember at least a single occurrence of colour in addition
to having a general feeling that the whole dream was in colour for the dream to be classied as coloured. Thus, poor
visual recall would inuence the dreams classication, pushing it into inconsistent or unrecalled category. When
recalling real-life scenes we have the convenience of knowing that these were seen in full, natural colour. This is not
so with dreams, and thus to some extent this relationship between visual memory and colour might be an artefact of
the applied categorisation rules.
Acknowledgments
This study was carried out as a part of my Ph.D. research, which is funded by the University of Dundee.
I thank Prof. Trevor Harley for his continuing support and supervision of this research project, as well as his help with
writing and revising the manuscript.
References
Beaulieu-Prvost, D., & Zadra, A. (2005). Dream recall frequency and attitude towards dreams: A reinterpretation of the relation. Personality and Individual
Differences, 38(4), 919927.
Bentley, M. (1915). The study of dreams. American Journal of Psychology, 26, 196210.
Berger, R. J. (1963). Experimental modication of dream content by meaningful verbal stimuli. British Journal of Psychiatry, 109, 722740.
Calef, V. (1954). Color in dreams. Journal of the American Psychoanalytic Association, 2, 453460.
Cartwright, R., Agargun, M. Y., Kirkby, J., & Friedman, J. K. (2006). Relation of dreams to waking concerns. Psychiatry Research, 141(3), 261270.
de Martino, M. F. (1953). Sex differences in the dreams of Southern college students. Journal of Clinical Psychology, 9, 199201.
Den, V., & Bulck, J. (2004). Media use and dreaming: The relationship among television viewing, computer game play, and nightmares or pleasant dreams.
Dreaming, 14(1), 4349.
Domhoff, G. W., & Schneider, A. (1999). Much ado about very little: The small effect sizes when home and laboratory collected dreams are compared.
Dreaming, 9(23), 139151.
Epstein, A. W. (1985). The waking event-dream interval. The American Journal of Psychiatry, 142(1), 123124.
Erlacher, D., & Schredl, M. (2004). Dreams reecting waking sport activities: A comparison of sport and psychology students. International Journal of Sport
Psychology, 35(4), 301308.

E. Murzyn / Consciousness and Cognition 17 (2008) 12281237

1237

Fein, G., Feinberg, I., Insel, T. R., Antrobus, J. S., Prici, L., Floyd, T., et al (1985). Sleep mentation in the elderly. Psychophysiology, 22(2), 218225.
Foulkes, D. (1999). Childrens dreaming and the development of consciousness. London: Harvard University Press.
Funkhouser, A. T., Hirsbrunner, H. P., Cornu, C., & Bahro, M. (1999). Dreams and dreaming among the elderly: An overview. Aging & Mental Health, 3(1),
1020.
Garma, A. (1961). Colour in dreams. International Journal of Psycho-Analysis, 42, 556559.
Giambra, L. M., Jung, R. E., & Grodsky, A. (1996). Age changes in dream recall in adulthood. Dreaming, 6(1), 1731.
Goodenough, D. R. (1991). Dream recall: History and current status of the eld. In S. J. Ellman & J. S. Antrobus (Eds.), The mind in sleep (2nd ed., pp. 143171).
New York: Wiley.
Hall, C. S. (1951). What people dream about. Scientic American, 184(5), 6063.
Hall, C. S. (1953). The meaning of dreams. New York: Harper and Brothers.
Herman, J., Roffwarg, H., & Tauber, E. S. (1968). Color and other perceptual qualities of REM and NREM sleep. Psychophysiology, 5, 223229.
Herman, S., & Shows, W. D. (19831984). How often do adults recall their dreams. International Journal of Aging and Human Development, 18, 242254.
Jankowski, W. L., Dee, S. C., & Cartwright, R. D. (1977). A distribution of colorimetric imagery in REM sleep. Sleep Research, 6, 123.
Kahn, E., Dement, W., Fisher, C., & Barmack, J. E. (1962). Incidence of color in immediately recalled dreams. Science, 137, 10541055.
Knapp, P. H. (1956). Sensory impressions in dreams. Psychoanalytic Quarterly, 25, 325347.
LaBerge, S. (1985). Lucid dreaming. New York: Ballantine Books.
Lindman, H. R. (1974). Analysis of variance in complex experimental designs. San Francisco: W. H. Freeman & Co.
Middleton, W. C. (1933). Nocturnal dreams. Science Monthly, 37, 460464.
Middleton, W. C. (1942). The frequency with which a group of unselected college students experience colored dreaming and colored hearing. The Journal of
General Psychology, 27, 221229.
Roussy, F., Brunette, M., Mercier, P., Gonthier, I., Grenier, J., Sirois-Berliss, M., et al (2000). Daily events and dream content: Unsuccessful matching attempts.
Dreaming, 10(2), 7783.
Schredl, M., & Erlacher, D. (2008). Relation between waking sport activities, reading, and dream content in sport students and psychology students. Journal
of Psychology: Interdisciplinary and Applied, 142(3), 267275.
Schredl, M., & Hofmann, F. (2003). Continuity between waking activities and dream activities. Consciousness and Cognition, 12(2), 298308.
Schwartz, S. (2003). Are life episodes replayed during dreaming? Trends in Cognitive Sciences, 7(8), 325327.
Schwitzgebel, E. (2002). Why did we think we dreamed in black and white? Studies in History and Philosophy of Science Part A, 33(4), 649660.
Schwitzgebel, E. (2003). Do people still report dreaming in black and white? An attempt to replicate a questionnaire from 1942. Perceptual & Motor Skills,
96(1), 2529.
Schwitzgebel, E., Huang, C., & Zhou, Y. (2006). Do we dream in color? Cultural variations and skepticism. Dreaming, 16(1), 3642.
Snyder, F., Karacan, I., Tharp, V. K., Jr., & Scott, J. (1968). Phenomenology of Rems dreaming. Psychophysiology, 4(3), 357.
Spence, I., Wong, P., Rusan, M., & Rastegar, N. (2006). How color enhances visual memory for natural scenes. Psychological Science, 17(1), 16.
Suzuki, K., & Takahashi, R. (1997). Effectiveness of color in picture recognition memory. Japanese Psychological Research, 39(1), 2532.
Tapia, F., Werboff, J., & Winokur, G. (1958). Recall of some phenomena of sleep. Journal of Nervous and Mental Disease, 127, 119123.
Waterman, D. (1991). Ageing and memory for dreams. Perceptual & Motor Skills, 73, 355365.
Wolcott, S., & Strapp, C. M. (2002). Dream recall frequency and dream detail as mediated by personality, behavior, and attitude. Dreaming, 12(1), 2744.