Perceptrml and /Motor Skills, 1964, 13, 131-138.

@ Southern Universities Press 1964

EIDETIC IMAGERY: I. FREQUENCY'

RALPH NORMAN HABER AND RUTH B. HABER'

Yale Unii~ersity
Sunmtary.-Eidetic imagery, clefined as a visual image persisting after
stimulation, relatively accurate in detail, colorecl positively, and capable of being
scanned, was nieasurecl in nearly all children in an elementary school in New
Haven, Conn. Care was taken to specify and follow precise methods for the
measurement, ancl strict criteria were used for the discrimination of eidetic
images from after images and from memory. Discontinuous distributions of
scores were founcl on all of the measures used to define eidetic images, with
12 Ss ( 8 % ) falling into this qualitatively different eidetic category on each
measure Thc~rimages lasted as long 3s 4 min.. during which nearly a11 of the
details of the stimulus could be reproduced. However, rheir memory of the
stimulus was not much better after their imagery hacl faded than that of Ss who
did not have eidetic imagery. These results were discussed in terms of models
of translation of stimulation into memory. and further research nr3s outlined.

Some children ( a n d a very few adults) are able co maintain a complete

visual image of a s c i m ~ ~ l ufrom
s which they can describe the stimulus in detail.
This eidecic imagery is notably different from most adult perception, in which
the visi~alimage of a scirn~~lus fades almost immediately and any description
of the stimulus must be based on a normally incomplete memory. These two
characteristics of eidetic imagery-long-term visual imagery and accurate re-
ports without memory-should have given i t great theoretical interest. How-
ever, current theories of perception have not considered visual imagery to be
very important. Sperling ( 1 9 6 3 ) , in perhaps the most comprehensive model
of the translation of stimulation into memory, does not mention visual imagery,
with the exception of after images. For him, the offset of the scimulus energy
is the end of the stimulus, and all f~rrcherprocessing of t h x stimulation uses
different aspects of encoded memory. W h i l e he shows (Sperling, 1960) that a
dark pre-exposure field increases the duration of what he calls short-tern1 mem-
ory (presumably because of the long after images produced), in ocher places
both he (Sperling, 1963) and Averbach (Averbach 8: Coriell, 1 9 6 1 ) talk of
short-term memory as already having undergone some encoding.
Eidetic imagery is i~nportnntin this connection because it is sufficiently
long and accurate to allow extensive and complece reports of the s t i n i ~ ~ l uto s
be made without benefit of the intervening processes described i n Sperling's
model. Indeed, if eidetic imagery is as prevalent as the l i t e r a t ~ ~ rindicates,
e then
'This research was supported in part by Grant MH-03244 from the National Institute of
Mental Health to Yale University, under the direction of the first author. We would like
to thank Miss May White. the Superintendent of Schools for Elementarv Educat~onof the
Ciry of New ~ a v e n Miss
, ~ a r ~ a ' eR.
t Fitzsimons, Principal, and each'of the teachers of
the Roger Sherman Elementary School of New Haven, for their permission to test the chil-
dren, and for rheir kind support throughout the project.
'Now a t the University of Rochester.
132 R. N. HABER & R. B. HABER

perhaps initially in the life of rhe child, all translations from stimulation to re-
ports may be through eidetic images, and not until later are the elaborate en-
coding and rehearsal processes necessary. Of course, any report in words of n
visual s t i m u l ~ ~requires
s encoding into words, but if the image of the stim~ilus
is long, this encoding does not have to be accomplished from an imperfect
memory.
W i t h these issues in mind, this research was designed first to ascertain
whether eidetic imagery occurs, and then to examine the nature of the imagery,
specifically how i t is used by the perceiver in reporting what he sees. This re-
port presents data relevant primarily to the first question.
Eidetic images have been distingilished from memory by cheir preservation
of fine detail (which is iisunlly lost in memory), by S's report that a visual
image still persists afcer the stimul~ishas been removed, and by behavior which
indicates that S is indeed attending ro such an image. Eidetic images have been
distinguished from after images by cheir persistence (afcer images fade rather
rapidly), by cheir reliability of evocation from even low-contrast stimuli (after
images are ~ ~ s u a l ldifficult
y to arouse from such stimuli), by their positive rep-
resentation of color (after images, especially long ones, are usually negative),
by their independence of visual fixation (after images usually require fixation
to form, while eidetic images do n o t ) , and by the lack of effects of eye-move-
menrs during report (after images move with the eye, while eidetic images can
be scanned visually ) .
Perhaps 200 empirical, sen~i-empirical,and clinical studies, mostly German,
have been made of chis phenomenon, although by 1925 as many were being
done in this country as abroad. Kluever has been the major reviewer of this
work, with extensive reviews in 1928, 1931, and 1932, while Jaensch has been
the primary systematizer, as a representative of the Marburg School, with a
major book in 1925 and another in 1930. By 1937, however, interest and re-
search, as judged by publications, had nearly ended, with only 12 papers listed
in Psychological Abstrac~sduring the past 25 years. T h e reasons for so sharp
a change probably included the lack of a sound theoretical base, the behaviorisric
climate against this introspective subject, and the strangeness and unusualness of
the behavior, at least as viewed by adult psychologists. N o serious doubts were
raised about the validity of eidetic imagery as a phenomenon, even tho~lghthe
methodology of assessment has been both poorly described and poorly executed.
Eidetic imagery just ceased to excite scientists. The majoriry of the dozen ref-
erences since 1937 are clinical reports of individual eidetic persons, usually
patients.
A review of all of the research findings of eidetic imagery would be our
of place here, in view of able earlier reviews of Kluever, and because of the
many serious ~nethodologicaldeficiencies of the early work. Percentages of chil-
dren said ro possess some form of eidetic imagery ranged from 30 to 90, depend-
ing upon the age and population sampled, with a r o ~ ~ gaverage
h of all studies
around 50%. Nearly every investigator has reported that eidetic imagery was
common, and that eidetic Ss could easily be found among any population of chil-
dren. Different investigators have reported different peak ages; some have
indicated a negative correlation with age, while others have pointed to puberty
or shortly before as the age of greatest prevalence. All investigators have re-
ported zero or near-zero frequencies among adults, although, as far as is known,
no longin~dinalstudies have been reported.
METHOD
Szbbjects
All Ss were students in the Roger Sherman Elementary School of New
Haven, Conn., during the academic year 1961-1962. T h e school had 245 children
registered, of whom 179 were tested. Those missed were either consistently ab-
sent ( N= 14) or because of time pressures were not included in the random
samples drawn from the lower grades. Of the 179 Ss tested, 28 were not scor-
able, due to ~nalfunctionsof the tape recorder, leaving 151 in the sample.

S was brought into a small room which contained a table with an easel on it.
The easel (30 in. wide by 24 in. high, in a n e ~ ~ c r grey
a l finish) was tilted away
from S slightly, and had a narrow ledge along the bottom on which the pictures
were rested. S was seated 20 in. away from the easel, his eyes level with the
middle of it. Room illumination was normal, with strong sunlight blocked by
curtains when necessary. A tape recorder transcribed both S's and E's voices.
Procedure
The sequence of events was the same for each S. H e first was shown a 4-in.
red square, mounted on a board 10 in. by 12 in., of the same material as the
easel. E placed the stimulus on the easel, left it there for 10 sec., and then re-
moved it rapidly. S reported what he still saw on the easel. Three other colored
squares (blue, black, and yellow), always in this order, were presented in a sirni-
lar fashion. After the fourth square was shown, f o ~ l rpictures were presented
for 30 sec. each, in the same manner.
T h e following instructions were given to S a t the beginning.
W e are going to play a game with colors and with pictures. Here on this easel I am
going to show you some colors and some pictures, and then we are going to talk about
rhem. When I put a colored square here (pointing), I want you to stare at the center
of it as hard as you can, and try not to move your eyes at all as long as I leave the square
there. When 1 take the square away, I want you to continue to stare as hard as you
can where the square was. If you stare hard enough, you will still be able to see sorne-
thing there. It is very much like when you stare hard at a light bulb, and then look
away-you can still see something out there in front of your eyes. (If any child acted as
if he was unfamiliar with this demonstration, he was instructed to try it then with one
of the overhead lights in the room.) The important thing is ro stare hard at the colored
square when I put it on the e a s e l 4 0 not take your eyes away or move them around.
134 R. N. HABER & R. B. HABER

When I remove the square, d o not look at me, or follow the color as I take it away, but
keep staring at the place where it was on the easel. As soon as I take the color away, I
want you to tell m e what you still see there, if you see anything. You d o not have to wait
until I ask you-you can begin telling me right away. OK, here is the first colored
square.

E was watching carefully during the exposure to be sure S did not move
his eyes. If S reported that he saw nothing at all after the square was removed,
he was encouraged by being assured that it was all right to see things after the
color was removed. If he still said he saw nothing, he was reminded to stare
hard, and not to move his eyes at all, and he was questioned again as to whether
he knew what these instructions meant. Then E presented the next square, in-
creasing the duration by 10 sec. over the previous exposure.
If S said he saw something, lie was allowed to report spontaneously. W h e n
he stopped, he was questioned on whichever of the following items he had not
reported: W a s the image still visible? W h a t was its color and shape? Did
color and shape change, and i f so, how? In what direction did the image move?
H o w did it disappear? Did it move when the eyes moved ( S was instructed to
try to move his eyes to the top of the easel) ? After these points had been cov-
ered, and the image had faded completely, E gave the initial instructions again,
and showed another square. T h e same procedure was followed for the four
squares.
After the last square was shown and S had finished his response, the instruc-
tions for the pictures were given.
Now, I am going to show you some pictures. For these, however, I do not want
you to stare in one place, but to move your eyes around so that you can be sure you can
see all of the details. When I take the picture away, I want you to continue to look hard
at the easel where the picture was, and tell me what you can still see after 1 take it away.
After I take ic away, you also can move your eyes all over where it was on the easel. And
be sure, while the picture is on the easel that you move your eyes around it to see all of
the parts.

All Four pictures were presented for 30 sec. each. E watched closely to be
sure the picnlres were scanned and not fixated. The first picture was of a fam-
ily scene, black pictures pasted on a grey board to form a silhouette. The sec-
ond, constructed in the same way, was of an Indian hunting, with a deer, other
animals, and some birds. The third, in full color, showed an Indian fishing in
a canoe, with many fish in the water. T h e fourth, also in color, from Alice in
Wonderlarzd, depicted Alice standing at the base of a large tree staring u p at the
Cheshire cat. A number of other similar pictures had been used in pretesting
and in extra testing with some of the same Ss.
After the first picture was removed, S was told to continue to look at the
easel, and to tell E whatever he could still see. S was reminded that he could
move his eyes. If S reported seeing something, E asked if he was actually seeing
i t then or remembering it from when the picture was still on the easel. E asked
 EIDETIC IMAGERY 135

frequently if he was still seeing it, since Ss often would not report the fading of
the image but would continue reporting it from memory. If S stopped his
report, E asked if he could see anything else. If S said no, but said he was still
seeing an image, E asked if he could describe anything else about that image.
E probed for further description and attributes of all objects still visible in the
image. S also was asked to move his eyes if he had not done so spontaneously.
E noted the relation between direction of gaze and details of report. Whenever
S said the entire image had faded, E asked him to describe the picture from
memory, wich as many details as possible. T h e same kind of probing questions
were asked.
If, after the picture was removed, S said that he saw nothing, E asked him
to describe the picture from memory with as many details as possible. The prob-
ing qi~estionsdescribed above were asked.
This process was repeated for all four pictures. T h e average time for test-
ing varied from 10 min. wich a young S having no visual imagery to more
than 30 min. for an older S with extensive imagery.
A t the beginning of the next academic year, 34 Ss were retested using the
same procedures and the same pictures and colors. T h e 34 were selecced so
as to include most Ss who had earlier produced long-term and accurate images,
some Ss with partial images, and a few Ss who never had reported an image to
any of the pictures.
Scoring
The tape recordings were encoded onto specially prepared data sheets,
which indicated the content of all responses (images and memory). T h e relia-
bility of this condensation of the data was nearly perfect, since the coding sheets
had categories for every object and mosc of their attributes for each stimulus;
the coder rarely had to make any scoring decision. All furrher scoring was done
fro111 these data sheets except the durations of responses, which were taken
directly from the tape recordings.
For the results to be reported here, Y scores were assigned to each S by the
first author. His scoring reliability was checked by a second judge who also
scored each S on the same S indices. In each case, the reliabiliry exceeded
u = +O.Sl for the continuous ratings, and 99% agreement for the dichotomous
ratings. T w o scores indicated whether or not the images for the squares and
the pictures were colored positively. T w o scores indicated the durations of the
images for the pictures and for the squares. T w o scores, o n a 5-point scale, indi-
cated the accuracy of the coloring of the images for the pictures and for the
memory of the pictures. Finally, two scores, on a 9-point scale, indicated the
accuracy of the details of the images for the picti~resand the memories for the
pictures. The four scores assigned to rating scales were summary ratings made
of the completeness of the images and memories.
136 R. N. I-IABER a R. B. HABER

RESULTS
Eighty-four of the 15 1 Ss ( 5 5 % ) reported images of at least one of the
pictures. As might be expected, a positive relationship between accuracy and
duration was found, although the only Ss who had both very high accuracy and
duration scores were those who saw images of all four pictures. The 12 most
extreme Ss in chat group were discontinuous from the remaining 72 Ss on
several measures. They were the only Ss who saw four images, all 4 8 of which
lasted over 40 sec., all of which had an accuracy of 6 or greater ( t h e majority
were 8 or 7 ) , 70% of which were positively colored (as compared to 345% for
the remaining Ss), and 100% of which could be scanned with the eyes (as
compared to 2 % ) . Since these latter two scores had been proposed as criteria
to distinguish eidetic images from after images, and because of their far super-
ior accuracy and duration, these 12 Ss seemed to be reporting eidetic images of
[he pictures, while the remaining 72 Ss seemed to be reporting after images o r
weak visual images of some other kind. Given this discontinuity on nearly
every measure relevant to a definition of eidetic imagery, these 12 Ss appeared
to possess an imagery which was qi~alitativelydifferent from that of all of the
other Ss in the sample.
T h e most striking aspect of the eidetic child's report was the vividness and
completeness of an image that was "out there" in front of him. There was no
qualification in his speech, such as "I think I see," nor did he ever use the past
tense as he might have if he were combining imagery and memory. H e was
able to report very fine detail, such as the number of feathers worn by each of
the 10 Indians in one pretest picture, the different colors in a multi-colored In-
dian blanket, the expressions on the faces, and the various poses of the persons,
and all from the same image. Even if these reports were based o n memory,
which they did not seem to be, i t would be quire unusual. One of the clearest
examples of eidetic imagery occurred when E showed the next picture, mistak-
enly chinking that S had indicated that che image to the previous one had faded.
After the second picture had been removed, S described her eidetic image, which
was clearly a fusion of the images of the two stimuli. She said that she knew
this was happening, but was still seeing it.
In addition to the differences in acc~lracy and duration between the 12
eidetic Ss and the 72 other Ss who produced some kind of imagery (non-eidetic
image), the eidetic Ss also had Inore positive (i.e., black) images to the two
silhouette pictures ( 7 2 % vs 2S%, I = 6.53, p < .001), and greater accilracy of
color in their images of the two remaining pictures than the non-eidetic-image
Ss ( t = 4.30, p < .001). For the responses to the colored squares, and for the
memory of the pictures, the eidetic Ss were compared with the 72 non-eidetic-
image Ss who produced some images to the pictures, and with the other 67 Ss
who never reported an image of any of the pictures (no-image Ss). T h e dif-
ferences between the eidetic Ss and the other two groups were not quite so srrik-
 EIDETIC IMAGERY 137

ing, but they were still highly significant (all t tests exceeded 3.00, p < .01).
In no instance did the non-eidetic-image Ss and the no-image Ss differ signifi-
cantly. T h e eidetic Ss saw more after images to the colored squares (3.8 as com-
pared with 3.2 and 3.0),more of which were positively colored (43% as com-
pared to 19% and 1 4 % ) , and which lasted more than twice as long as those of
either of the non-eidetic groups (35 sec. as compared with 14 and 13 sec.) . T h e
differences in memory among the rhree groups were much smaller than ex-
pected, although for borh the accuracy of detail and of color, the eidetic Ss were
significantly superior to borh of the other groups (all IS > <
2.10, p .05).
T h e eidecic 5s did not differ from rhe school population in their sex or
race. Their ages varied from 8 to 12 yr. Since no children older than these
were tested, no indication was available of the upper age limir. N o eidetic chil-
dren were found below the second grade. - However, since there were so few
eidecic children at all, conclusions regarding their distribution throughout the
population must be very cautious. Furrher, clear evidence of diffic~llcy was
found in the younger children (ages 4 through 6 or 7 ) in understanding what
was expected of them and in their conimunication of what they were seeing and
remembering. O n the basis of these observations and results in chis age mnge,
previous posirive r e s ~ ~ l of
t s eidetic imagery in children chis young or younger
should be viewed with extreme caution.
Nine of the 12 eidetic Ss were retested 8 mo. later, along with 25 other Ss.
None of the non-eidetic Ss, based o n the original classification, produced any
images in the retest to any of the pictures that were as long or as accurate as the
poorest image of the eidetic Ss. Of rhe original 9 who were retested, all but
one showed the same type of imagery. While the relationships between original
and retest scores for these 9 Ss were generally low, they were all off the distribu-
tions on both testing. One S failed to p r o d ~ ~ cany
e eidetic images on the retest,
even though she had an average of nearly 3 nlin. of imagery originally, with
very fine accuracy. E reported that this S seemed extremely anxious on the re-
test, and was very concerned as to the reason she was being tested again. W i t h
this one exception, the retesting further supported the classification of Ss into
eidetic and non-eidetic.

Contrary to a voluminous literature, the prevalence of eidecic imagery in

an elementary school in N e w Haven was quite low-about 8%. However, the
12 Ss who were classified as eideric were not merely the end of a continuous
distribution, but rather were children who showed qualitatively different be-
havior on this simple perceptual task. Therefore, eidetic imagery does exist as
a verifiable, identifiable characteriscic in children. This first snldy has not
attempted to explore further how these children were different from the 92%
of children without this ability. That is to be followed up.
138 R. N. HABER & R. B. HABER

The most likely explanation of the discrepancy from previously reported

frequencies is in terms of the methodological differences in the techniques used
to assess the presence of eidetic imagery. This study used very strict criteria and
very careful observation of the behavior of Ss' eyes, as well as their verbaliza-
tions. Many experiments in the literature classified Ss as eidetic if they pro-
duced may images of pictures. Following this criterion, 55% of the children
would be eidetic in this study. However, ic seems apparenc that most of these
images were after images-they persisted for very short periods of time, they
were usually negative in color, they could not be scanned with the eye, and they
included very lirtle detail of the stimulus. Therefore, i t is assumed that these re-
sults represent a closer approximation to the prevalence of eidetic imagery in che
general population of children than does the previous literature.
It has not been possible to differentiate after images from other kinds
of weak imagery in this study, if in fact they can be differentiated at all. T o be
able to have done so would have required that the imagery Inst long enough so
that the effects of eye-movements and the accuracy of representation could hnve
been assessed. While neither of chese w o ~ ~ lhnve d been conclusive, it might
have been possible to offer a better discinccion than we can now.
Evidence has been presented, however, chat 5 5 % of the children had some kind of
imagery to the pictures, with 8% having eidetic imagery. These 8 % could use their
imagery to report derails of the stimulus, very much as if they had a nearly perfect memory
lasting for the duration of the imagery. It was somewhat surprising that their actual
memory for the stimulus, after the imagery had faded, was not also strikingly becrer than
that of the non-eidetic Ss. Apparently, the eidetic Ss were not using the time during
which the image was present to encode the stimulus for later Wall, nor were they taking
advantage of their practice in reporting the stimulus from their imagery.
T h e presence of imagery of this fidelity and duration certainly requires inclusion in
any kind of theory or model of the translation of stimulation into memory. Originally, it
had been hoped we could carry out extensive experiments with large groups of eidetic Ss.
so as to provide data for the construction of such a model. However, because of the small
percentage found, this has not been feasible. Therefore, work is proceeding with derailed
examination of these 12 Ss. This effort will provide a short-term longitudinal study, in
which a number of different aspects of the relationship benveen perception, irnaaery, en-
coding, and memory can be analyzed.
REFERENCES
AVERBACH.E.. & CORIELL.A . S. Short-term memorv in vision. Bell S ~ s t e n ztech. I . .
1961, 40, 1-20.
JAENSCH,E. R. D i e Eider&. Leipzig: Quell & Meyer, 1925.
JAENSCH,E. R. Eidetic imagery a71d typological nzefhods o f i11vesfigntio71. New York:
Harcourt Brace, 1930.
KLUEVER,H. Studies on the eidetic type and eidetic imagery. Psychol. BttlL., 1928, 2 5 ,
69-104.
KLUEVER, H. T h e eidetic child. In C. Murchison (Ed.), A handbook o f child psy-
chology. Worcester, Mass.: Clark Univer. Press, 1931. Pp. 643-668.
KLUEVER, H. Eidetic phenomena. PsychoL. BILL/.,1932, 29, 181-203.
SPERLINC,G. The information available in brief visual presentations. Psychol, Alon-
o g ~ . 1960,
, 74, No. 11 ( W h o l e No. 498).
SPERLING, G. A model for visual memory tasks. Hum. Fucrors, 1963, 5 , 19-31.
Accepred May 19, 1964.