Professional Documents
Culture Documents
MMiITAL IIUAGERY
i-
Department of Psychology
University of Adelaide
South Australia
Awatd¿ol rù ø Yì
SUMMARY
imagery ability.
cube as stimulus.
Authóf rs 'stäte¡nent
Signed:
(Patricia A. Quealey)
vf-l .
ACKNOÌ4¡LÐGE{ENTS
Acknowledgements vra
(ii) Perceptual 2L
(iii) Physical 23
I NATT]RE 36
I1. FIJNCTTON 39
EXPER.IMENT I 51
Met̡od 52
Results 55
Discuss Íon 62
I Tt¡e Task 67
TII. Procedure 70
l"letl¡od 77
Results 80
Discussion 84
Method 96
Results 99
Discussion ro2
Page
CHAPTER 8. 4: TIIE EFFECT OF
EJKPERII'IENT PHYSICAL
II{POSSIBILIT( ON AN IIUÀGINAL TASK I06
Method I09
Results rlt
Discuss íon 1r5
.Mett¡oil 123
SesuIts L25
Discuss ion t28
Ð(PERIIVTB'IT 6: 145
Method 145
Results and Discussion L47
Ð(PERllvlÐ{T 7 ¡ r50
-tt{ethod r50
Results and Discussion r52
General Comments 158
Page
Results L79
Discussion 183
BTBLIOGR.APHY 234
***
CHAPTER 1
r GENERAL
There has never been debate over the reality of the phenomenological
The early writers seemed to assume that their readers were farniliar
witì imagery as a phenomenon. Th¡e hundred years since then has
The last ten years has brought a resurgence of enthusiasm for these
issues - enough to fill seven rnajor booksl and a quarterly journal
devoted entirely to imagery research and discuss i-on2, to justify
so that copies of them arise again in the mind after the original
outward stímutus ís gonerr (p.44). To ,Jung (f923) an image was
nthe concentrated expression of the total psychic situatíon,
Ie78).
While it has been charged tl¡at imagists like Paivio have not
made it clear what tJ:ey do mean by imagery, it is also true
that pro¡nsítionists have not made it clear either. In abstract
logical terms, we know what a pro¡nsition is. We do not know
One way in which the view that inages are propositions could be
tested would be by determining whether images bel¡ave Ìike
propositions. For example, if images do bet¡ave like propositions,
then the one deep structure should permit different surface
structures. Moreover, irnage forrnation would be determined by
consonant rules. One would also want to know what types of
pro¡nsitions are images.
{ztyshyn (L973) in his article "What the mindrs eye tells the
mindrs brain" rejected both the perceptuar and pictoriar notions
of imagery. As sree (1976) notes, arr imagery researchers would
agree with hi¡n about the inadequacy of a purely pictorial
conception of imagery, but $zlyshyn rejected the perceptual
idea for the same reasons as he rejected the pictorial idea,
without. dístinguishing between tl¡em.
Rather than deal with imagery in its own terms (whatever they may
CHAPTER 2.
REVTEViI OF TMAGERY RESEARCH
r RESEARCH'ON'NATI'RE OF IMAGB.Y.
The ¡nss,ibility that there are specific neural areas for ìmagery,
the noisy brain and insufficiently sensitive measuring instruments
combine to make that line of research general_Iy still unproductive.
AIl that can be said is that the presence of irnagery is not
2L.
( ii) PerceÞtuaI
with each other when they were in the same modality. His
conclusions hrere restricted to spatial representations,
which he held were only one form of visual inagery.
Ertensive work using this approach was carried out by Segal and
(Cicero, de Oratore,
in Yates, 1966, p. 2'r.
Modern memory techniques stitr use the same basic idea as the
ancient orators. Ttrey form an image, emphasising the spatial
arrangement of its contents. Mne¡nonics such as rhylñing schemes
does aid recall, for concrete objects but not for words, only in
non-sequential tasksr and when items are presented slowly.
SHAPTER 3.
NATT'BE
viewpoint.
(ií) Function
P. 60
serveól no function.
percepts.
of details)
I
CHAPÎER 4
EXPERIMEI\]T I
Tt¡is research consisted of a series of exPeriments which were
concerned with the possible influence of qualitative and
ÌvlcKellar , Lg57; Buck and Geers, 1967) was one of wide individual
differences in the reported vividness of imagery.
way in the United States during the First !{orld lalar, produced
measuring scales for most recognised behaviours. Fo1J-owing
factors (sritain).1
I
Nowadays, controversies tend to be kept at home, someti¡nes
even within the same University department. Morning tea
conversations at the University of V[. Ontarío may educate
the eavesdropper as Paivio and {zlyshyn argue about the
nature of imagery.
49.
þp. 7a-75)
METHOD
kinaesthetÍc. (A¡¡Pendix I) .
53.
RESULTS
verbal scale of the IDQ, the total score for each subject on the
1
six irnagery measuresl and the Matriculation Aggregate. The
TABLE 4.1
0ùlr
.3r9
]TDQ ***
.468 .252
PIR *** **
.0 34 .067 .079
MPFB
TABLE 4.2
one negatÍvely
FACTOR SCORES
prone courses.
TABLE 4.3
ARCHITECTURE PSYCHOLOGY
* p<.05
60.
TABLE 4.4
MALES FE¡{ALES
* p=.02
TABLE 4.5
ARCHITECTURE PSYCHOLOGY
*** p ¿, .001
** p<.01
p < .0005, l-tailed; for MPFB, t = 2.5Ir 47 df, P < '0I' I-tailed) '
efficient.
62.
DISCUSSTON
tests and the total score for imagery measures (r = .499, r = .557).
compared witJl the negligible relationship between the spatial
separate faculty (p. 751 , and with Pembertonrs (1952) opinion that
spatiat tasks can be done either by imagery or by reasoning.
lftre results of the factor analysis can be compared with those from
a study by Paivio, which te reported in his 1971 book. He found
QMPFB (.71), rrDQ (.65), QMr (.55) and also on MPFts (.48)-
Ttre sigrnificant loading on botJ: factors may well have been due to
his use of only part of the MPFts test. Four imagery factors
emerged in alt: Factor 3, which was unlabelled, involved PIR, IIDQ' VID
Factor 4, also rrnlabelled, involved PfR and IIDQ.
Perhaps the most important outcome of the present study was that it
confirmed the finding of previous research which had proven difficult
to replicate (Di Vesta, Ingersoll and Sr:nshine, '1I; Richardson, 77) t
that the e>çerience of imaging and imagery manipulation ability
were independent. lltris independence of Factors I and 2 is relevant
to tl¡e central issues of ttre nature and frrnction of imagery, which
were introduced in Chapter I. fhe ex¡rerience of imagery, Factor 1,
\^¡as reported by a1l subjects, which supports Doob's (1972) belief
that imagery appears universally among all peoples, and Marks'
(J-972) view that at least tJle potential for imagery is universal.
Ttre independence of the e>çerience of imagery and the ability to
use an image , Faciuor 2, implies that having an image is not
sufficient for successful performance at an imagery task. For
visualization was ín solving the task to have been higher than it was
Tt¡e above findings suggested that the arousal of imagery \âras not a
sufficient condition for the effective us¿of imagery: the mere presence
of an image did not guarantee its useabitity. lllre findings also
indicated that, while people cannot be readily distinguished in their
experience of imageryr they can be distingrrished in their.success in
using imagery. FurtJrermore, although the e>çerience of imagery was not
accessible for objective measurement, ttre use of imagery did appear to
lend itself to experimental measurement.
Thre focus of the research was on the factors which influence the use of
imagery. Itre approach taken in the ensuing e>çeriments was e>çerimentally
to manipulate possible influential factors. As the e>çeriments were
based on a task requiring the use of an image, sr:bjects \^/ere selected on the
both Space Relations and MPFts tests. Thris method of selection was used,
CHAPTER 5
I. TTIE TASK
simpre enough to be within the range of imagery abitity for most peopte.
Íhe task involved the use of a memory image, which made for maximum
probabry accounted for his finding that some subjects who performed
adequately on t-Jee task claimed to have used no visual imagery.
Arternative strategies, using logic or matÌrematics, were equally
available -
then to decide how many cubes had three, two, one and no
red painted faces.
FIG. 5.1
STANDARD SÎIMULUS
Ttre aim was to determine how the two types of data were associated,
and especialty how qualitative differences in imagery related to
performance.
70.
III. PROCEDURE
Test and the lvlinnesota Paper Form Board Test (as described
in Chapter 41.
For all but one experiment, the upper and lower quartiles were
the warls and no windows to break the monotony. The desk had
tJ:e cube a¡¡d asked how many cuts would be needed to cut the
cube into 27 smaller cubes of equal size. Vfhen it had been
established that two cuts in each dimension would be necessary,
and tJ'e subject professed abitity to visualise the dissected
cube, the cube \^ras removed from sight.
of the cube. fhe task itself v/as concerned with how many
cube; t.J:e second tlzpe of question asked how many cubes had
specific features or combination of features.
llhre time taken to the nearest second was also recofded by the
experimenter, using a stopwatch. The time \das recorded at two
intervals: after the general questions abouÈ the properties of the cut
cube, and ttre total time taken for the task. Because subjects tended
CHAPTER 6
EXPERIMENT 2 IDIAGE S I ZE
dimensions.
by the fÍnding that true and false responses took the same
}{ETTTOD
The stimuli were two wooden cubes, painted black and white, as
FIG. 6.1
STII{ULI FOR EI(PERIMENT 2 (II'{AGE SIZE)
79.
RESULTS
TABLE 6.1
SUBJECTS
HIGH IMAGERS
TIME (secs. ) 388 343
LOW IMAGERS
TIME 560 st2
Data from the question about image size for the four conditions
are presented in Tabre 6.2. standard deviations are incruded
as there were differences between experimental groups.
TABLE 6.2
STIMULUS
LARGE SMALL
HIGH 3.9u 3.5"
s = 2.0 s = 3.3
IMAGERY
o 2 4 6 8 10 12 14
INCHES
FIG.6.2
REPORTED IMAGE SIZES FOR SUBJECTS OF HIGH AND LOI{ IMAGTNG ABILITY
83.
were associated with rnore errors. When high and low imaging
subjects vJere considered separately, it emerged that image
size did not affect the error rate for high imagers
(rs= -.06, p = .65, N = 52). For low imagers, small images
Image size did not influence the time taken for the task for
the total group, nor for imagery levels considered separately.
(For fuII correlation results' see Appendix I0)'
84.
DISCUSS ION
finding suggested that the small image may have been Loo small
for the operations which the task demanded' as Kosslyn (1975)
predicted:
nade with the small image and that it took more time to
evaluate because the image had to be enlarged is internally
inconsistent in that the two factors of time and error should
have reflected, according to his explanation' a time-error
trade-off. Neither the present experiment nor Kosslynrs found
an inverse relationship between time and error scores.
Why small images should produce more errors, but not take
Ionger to process, is not clear. One possibility is that
successful performance is a matter of ability to use appropriate
techniques, and that investing longer time does not counteract
Iack of ability. This question of the relationship
between time and error will be discussed later in Chapter 12.
86.
The main finding of the experiment, that image size can affect
performance on an ímaginal task, could be considered as
support for the pictoriar view of imagery. rt wourd appear to
be inconsistent with the idea of imagery as an epiphenomenon:
CHAPTER 7.
TABLE 7.1
perhaps the clearest aspect of the results was the different patterns
whích ernerged for vivid and poor imagers. ft¡eir performance with
stimuli of low and medium levels of complexity are reversedr though
was to determine whether the arrows would meet when the cube
was built. The faces which were not folded, but attached
half and one second to processing time, which was less than
the 1.3 seconds added on for each relevant face. This result
was presented as an interesting side issue. The authors did
with tess visible elements are less vivid, but that recall
is not so1ely a function of vividness. Concealed images
analysis of varíance.
6 8 12 16 24
Fig. 7 .L
TABLE 7.2
6 11.58
8 5.62
L2 7.52
16 Lr.27
24 10.95
one considers the many ways in which an image and the use of an
evídencer ê.9. Sheehan (L967), one could predict that high inagers would
have a detailed, quasi-photographic type of image. All the experiments
considered so far have required reproduction of the total image'
and have left it possibte for the subject to use alternative coding
strateg ies .
image task where the total stimulus was not necessary. One might
imaginal task?
METHOD
{,ã
# # 4-!
-f94.
FIG. 7.2
STI}{ULI FOR EXPERIMENT 7 (IMAGE COMPLEXITY)
97.
TABLE 7.3
SUBJECTS
HIGH
I}4AGERY 32 25 57
LOW
IMAGERY 28 24 52
TOTAL 60 49 I09
I
The slight unevenness in number of subjects per group invites
some explanation. The usual one of subjects being lost in
laboratory fires would be exciting reading, but untrue here.
It proved difficult to manage simultaneously a double-blind
procedure and equal numbers in the four groups. The data were
reanalysed using analysis of variance, randomly discarding excess
numbers to equalise the number of subjects per group, with no
alteration of results. For this reason, and because researchers
deplore wastage, the original numbers were kept.
98.
Credit was given for each item recalled and for accuracy.
For example, a response that there were seven dots (instead
of 5) scored one point. It was decided to consider only
details recalled spontaneously, as prompting would
have made ít a recognition task, on which high and low
I
Slee (L976) found no differences between high and low
imagers on pattern recognition but significant dÍfferences
on pattern recall, which suggests that 1ow imagers had in
storage quite accurate infornation on the visual features
of patterns. Low imagers did not--have a storage problem
but an alternative coding strategy which was less appropriate
for free recall of patterns. (p. I50).
99.
RESULTS
TABLE 7.4
MEAN ERROR AIID TIME RESULTS FOR HIGH AT{D LOW IMAGING
SUBJECTS GIVn{ SIMPLE AND COMPLÐ( ST]MULI
SUB'fECTS
ERRORS 5.1 3.6
HIGH II4AGBY
TIME (secs. ) 363 374
the cube during the task, whether they attempted to use the
details, and whether the details interfered with the task,
are presented in Table 7.5. Tf:ey are expressed in terms
TABLE 7.5
LOW
IMAGER.S 7 I 2
il .", i.
:',...''
lhe error and time scores were compared for subjects aware of
the detaits during the task with those not aware. llt¡ere was
no difference betl^¡een them on errors (t = 0.6I' 47 df', P <.2O)
or on time (L = 0.24, 47 df., p < .20). Results vJere not
compared for the other two response categories as the frequencies
were too low.
TABLE 7.6
HIGH
IMAGERS 9.64 4.84 9 -L2 13.96 23. 80
LOVù
IMAGERS 9.04 4.46 6. 88 1I.33 20 .38
DISCUSS ION
The most striking result is that high imagers made less errors
with a complex stimulus than they did with a simple one (Table
1.41. Tt¡is can be seen as consistent with sheehanrs finding
(:-967) that high imagers performed better with a stimulus of
.If this were in fact the case, the images of subjects in the
Íhe equivalence of error and time results for both stimulus conditions
suggests that tfiis may have been so for low imagers. Tfhe equivalent
higher recall rate for high imagers on such details. Subjects could
results obtained here support Sheehan's finding ttrat high imagers store
a literal image whereas to* i..g.rs code more verbally and hence lose
details
105 _
This set of results v¡ould suggest that the image in problem solving
tasks is not quite a memory image, nor a reconstructed perceptual image'
but something more like the product of a decision-making process on
which features are relevant to the task. Thris implies some degree
of control over what the image will be like, for both levels of imagery
ability. The finding that discarded informatíon is still available for
a complex image for tl.e task. The irrelevant details, while onritted
from the image and stored, appeared not to interfere wit]- perfolmance'
Threse findings suggest that high imaging subjects may be more motivated
were not included in the image, but were stored and were availal¡le for
recall. subjects presented with the complex stimulus had more information
stored about t].e image than did subjects presented with the simple
stimulus. Yet the performance of low imaging subjects in simple and
complex stimulus conditions \^7as ind.istinguishable in both performance
accuracy and time taken- Such a finding suggests support for the
imagery.
106.
CHAPTER 8
ON AN IMAGINAL TASK
METTTOD,
Ftg . 8.1
Tt¡e ínslde of the Eugar cube was left white. lft¡e cubes
emphasised that the sugar cube I4Ias being cut by a knife and
also asked whether the inage he used in the task was a sugar
cube.
RESULTS
llLre mea¡r error and time rates for the four groups are presented
in ttre following tabIe.
TABLE 8.
SUBJECTS
made with the sugar cube (F = 5.79, I df , p¿.02) and longer time
was taken (F = 5.65' 1 df, p = 'o2)' As in the previous experiments'
between imagery level and stimulus type for errors. fh¡ere was
truo conditions.
Tfhe final question subjects were, asked was whether their image was
TABLE 8.2
SUBJECTS
HIGH
IMAGERY 2 2L
LOW
IMAGERY 7 16
sugar cube was compared with those who reported an image not
made of sugar. Table 8.3 presents the mean error and time
scores totalled across imagery leveIs.
TABLE 8.3
I\ÍEAN SCORES
SUBJECTS
* p..05
** p<.02
r14.
Despite subjects reporting that they were using an abstract image '
the type of stimulus with which they had been presented also
influenced performance. subjects who had been presented with a
fhere was no interaction between imagery level and stimulus type for
errors. There \^las some suggestion, however' of a possible inter-
action between imagery level and stimulus tlpe for time (p = .08) '
with low imaging subjects tending to be more hampered by the sugar
cube stimulus than were high imaging subjects'
DISCUSSION
subjects of bot]- high and. low imagery levels made less errors with
the sugar cube as stimulus and took longer for the task (ta¡te 8'I) '
In the two experiments previously reported' high and low imagers
tended to differ in ttre extent to which the stimulus influenced their
performance. Íhre pattern I^Ias that low imagers \^rere more likely to
beaffectedbystimulusproperties'sltggestingthat,unlikehigh
imagers, they could not choose the information content of their
images. In Experiment 4, the stimulus property which was varied
affected the error rates of high and low imaging subjects similarly.
Tlrere a tendency, however , for low imagers to be more adversely
r^¡as
affected than were high imagers in the time taken for the task'
115.
Ttris was consistent witl the finding that, generally, the sugar
cube stimulus seemed to influence them similarly'
the sugar cube stimulus are not hard to find. Factors such as
made the stimulus more interesting, the image more robust, and
sugar cube image for most subjects. (Table 8.2) More low imagers
than high imagers, however, did employ a sugar cube image for the
task. fhese sr:bjects were trling to do with an imaginal replica
of the stimulus sometJring which they believed could not be done
with the physical stimulus. To realise that the sugar cube could
ability.
those who did not' t].Ie former group took longer time for the task'
116.
though they did not make significantly more errors. (Table 8.3)
physicallyimpossible,whichsuggeststhatthedifferencebetween
tlresugarcubeimagersandthenon-sugarcubeimagerswasnot
their und.erstanding of the problem. Most subjects reported using
an image of a cube that was wit̡out texture, that was almost a'bstract'
The descriptions given by subjects of their images in both stimulus
difficult was transferred into the imaginal task for all subjecÈs '
lfhrose who incorporated the sugar texture into their image performed
Ì¡rorse tJean subjects who had the information stored but not represented
in the image.
fhreresultsofthise>rperimentextendthoseoftheprevioustwo
in that they support the notion that information about the image,
thoughnotÏepresentedinit'caninfluenceperformanceonan
imaginaltask.Mostsubjectsclaimedtohaveusedanimagenot
made of sugar, yet their performance differed in
errors and time
CHAPTER 9
illustration in a novel.
during imaging, even when signal and image were not in the
same modalíty, suggested that attention was a contributing
factor. The explanation in terms of distraction, however,
could not account for the finding that the interference was
that the pictoriat features of the image \^tere necessary for task
performance. such a finding would be incompatible with the
the task. Th¡e rnoÇal rate of slide presentation and tape volume
was then utilized in Experiment 5. The calibration was thus set
METHOD
task for the second time. To allow for the effect of repeated
The standard procedure was followed for aII groups, with one
RESULTS
TABLE 9.I
PER.EORMANCE ON TRIALS UNDER VTSUAL INTERFERENCE,
AUDITORY TNTERFERENCE AND CONTROL CONDITIONS FOR
HIGH AND LOW IMAGING SUBJECTS
MEA}I ERRORS
SI'BJECTS
HIGH IMAGERY 3.5 3.8 4.L 3.0 3.5 2.5
LOVS IMAGERY 8.0 8.0 7.4 6.0 7.6 4.2
TOTAL 5.8 5.9 5.8 4.5 5.6 3.4
STJBJECTS
It was found that, on the test trial, high imaging sr:bjects made
indistinguishable from the time taken when repeating the task under
control conditions. TLre above effects all reached .05 level 0f
significance.
TLrese results indicated ttrat both high and low imaging subjects
performed best when merely repeating the task, and worst when repeating
the task during visual interference- Thre performance accuracy of low
DISCUSS ION
equivalent degree.
The control group overall performed better on the test trial than
on the initial trial: they made less errors and took less time.
TL¡is was to be expected as the task would have had some familiarity.
this difference between trials was significant only for low imaging
subjects. For high imaging subjects the improvement was not
sufficiently large to reach conventional levels of significance.
Ttris may well have been because high imagers performed adoquately
on the initial trial, which meant less scope for improvement.
conditions, the tendency was for the control group to make less
errors.
CHAPTER ]- O
One distinction tlat has been made about imagery has been on an
time.
AIso of interest was the amounL of the cube they could see and
most often worked with: did they work with individual cubes,
I
134 .
RESULTS
subject who could not answer Question I and two who could not
ans\^'e! the second question were omitted from that item.
TABLE 10.I
LOV{
]MAGERY 66 32 35 63 34 55 IO
SUBJECTS
a
r35.
The frequencies in the above table were anal-ysed using chi-
TABLE 10.2
Less errors were made by high imaging subjects (F = 69.37 t L/L88 df'
p = .001) by subjects who used an undivided image (F = 5.55,
'
1/I88dfr p = .0I9), by subjects who kept their images still while
counting (¡ = 14.94¡ L/I88 df' p = .00I) r and by subjects who
worked with more than one face of the cube at a time (î = 17.77,
2/L93 df, P = .00I).
Tt¡e technique used did not influence time taken for the task.
Imagery levet was the only main effect for time. (In analysÍs I,
F = 47.'76, l/I88 df r p = .00I; ín analysís 2, î = 47.89t
L/L93 df, P = .001).
137.
DISCUSSION
face of the cube' more often than did high imaging subjects.
(p.866)
oneaspectofthedataisthewayitgivesrisetoquestions
of introspection that are not easily answerable ' The
experimenter exPectations ..
r4t.
CHAPTER I1
EXPERIMENITS 6 ANd 7
Evidence from genetics thus suggests that males begin life with
more aptitude for spatiat visualization than do females. This
and more fully in boys than in girls (Maccoby and Jacklin, 1975)
and vocational counseIIOrS note that more men than women choose
Students who score low on the initial testing (using the spatial
visualization subtest of the Primary Mental AbiIíties Test) are
counselled to transfer to engineering courses where spatial
ability is not so important, such as electrical or solid state
engineering. This nust reflect a belief on the part of the
staff that formal training is of restricted va1ue. The belief
would appear to be confirmed by the finding that initial tow
scorers irnproved less with the training than did studehts who
EXPERIMENT 6
TT4ETHOD
Ihe procedure for both conditions was the same, in all but one
On the test trial, both groups were given the more difficult
task, i.e. the task requiring the cube to be cut inEo 27.
On the test trial, the procedure detailed in Chapter 5 was followed
for both groups. There was a break of five minutes between initial
and test trials for both groups. To minimise rehearsal of the
task during this interval, t̡e experimenter engaged subjects in
conversation on a range of topical issues, including recent
electíons and approaching exams.
The mean results for errors and ti:ne on initial and test trials
TABLE II.I
TRIÄL I 2 T 2
27 cubes.
The concern of the experiment was with the effect of the two
improved performance.
14 8.
tl¡e few ele¡nents in the easier imaging task may have been
one problem in this design was that there v¡as no check that
subjects assigned to the two groups would have performed
equivalently on the initial run. Subjects were assigned
to groups randomly, however, whích should have made non-
equivalence unlikely. It wou1d, of course, have been
impossible to give both groups t̡e same task without defeating
the pur¡nse of the experiment, which was to provide a.
EXPERII"IENT 7
METTIOD
TABLE II.2
U\,IAGERY LEVEL
ERRORS
TIYIE (secs. )
for both errors (F = I2I .59, .L/36 df , P < .001) and time
(F = 20.23, Iß6 df.t p < .00I) . Less errors were made on
the test trial than on the initial trial. There was a
their error rates more than did any other group (p = .05).
Unlike subjects of high and medium imagery levels, however, who
took less time for the task on the test trial in both conditions,
Iow irnaging subjects took tonger to do the task following
training
154 .
TABLE 11.3
HIGH
II\4AGERS 0.6 2.0 2L5 L44
MEDITM
IMAGERS 2.3 3.I 260 136
LOW *
IMAGERS 2-0 5.9 15 -L02
an intermediate degree.
156 .
4 HIGH IMAGING SUBJECTS
MEDIUM IMAGING SUBJECTS
LOW IMAGING SUBJECTS
U)
3
()
ul
cl
fU,
2 I
l!
o I
/
ct I
I
1 I
AI
I
/ /
\
\ I
/\
/\ \
-2 o246 I 10
REDUCTION IN ERRORS
F'IG. II .I
CHANGE IN ERRORS FOLLOWING PRACTICE
4 {
,t
,l
t,
3 I
ct) I
o
l¡J
I
I I I
m t
-t I I
U) 2
I
l!
o x I
c;
rl
z 1 I \, /---t
/ \, \
I \
I
I
/ I
-2 o2468 10
REDUCTION IN ERRORS
FIG. TL.2
CHANGE IN ERRORS FOLLOWING TRATNTNG
Is7 .
GСERÄL COMMEI\]TS
GIAE{TER 12
ST{IDY OF ERRORS
tirne and error scores for the two groups. Ihese patterns
demanded further analysis as they seemed to provide evidence
of just the kind of qualitative differences in the imaging
process for high and low imagers whÍch might be informative.
I. TYPES OF ÐRRORS
how many had only one white face, how many had one white and
16I.
TABLE ].2.I
STA}¡DARD
DEVTATION 14 .5 29.I
RANGE -38 to +42 -34 to +I35
considers that 27 cubes'in aII were involved, and a sum total of 65.
40
I
U) /
C)
ul
I
ìq¡ 30 I
Ð
U' I
\
I
l¡.
I \
o I
\
c; 20
z \
\
\
I \
10
DIFFERENCE SCORES
Fig. 12.L
distr ibution.
163.
TABLE 12.2
TJNDER-
ESTIMATTON 45 27 72
OVER-
ESTIMATION 5I 72 L23
TOTAL 96 99 195
subjects were asked about the number of small cubes which had only
one white face. To guote one horrified subject - "Oh, there must be
Data from the same I99 subjects were used to examine the
relationshíp between error and time scores for performance
Data frorn the two imagery leveIs were then analysed separately.
For high imaging subjects, a correlation of .04 was obtained
(p = .68), as compared with a correlation of .28 (p = .004)
for low imagers. The correlation coefficíents for the Lwo
imaging task was not only associated with less errors¡ but
with errors of a different kind. The finding makes it likeJ-y
CHAPTER 13.
percept, then it, would follow that the image must be different
from a percept.
example.
t7r.
analogous:
It
the functions are surprisingly alike
and equalJ-y linear in both cases
It¡ese various results strongly
suggest that the mentally transformed
internal representation is acting as.
a kind of holistic templatç that can
be matched against the test stimulus
about as effectively as can an untransformed
perceptual-memory image " .
pp.223-224
are inter-related.
M.ETHOD
FrG. t3 . I
STIMULT FOR TETRAHEDRON TASK
Ltl .
FTG. T3 .2
SOLUTTON OF TETRAHEDRON TASK
178.
TABLE 13.I
Ð(PERIMM{TA], GROUPS
HIGH
IMAGER.S 27 28 55
LOVt
IMAGERS 25 27 52
TOTAL 52 55 r-0 7
Subjects were told that the two identical pieces fitted together
to form a tetrahedron, which was described as an object consisting
of four equilateral triangles, one on each of three sides and one
on the base. It was also described as rese¡nbling a p1'ramid, the
only difference being that a ¡¡¡ranid has a square base, whereas a
RESULTS
TABLE 13.2
SUCCEÐÐ 23 20 11 I8 72
FAfLED 4 I I4 9 35
TOTAL 27 28 25 27 107
succeed more often when the stinulus was smaII. High imaging
subjects, however, were unaffected by stimulus size.
t8I.
TABLE 13.3
HIGH IMAGERS
Succeeded 3I L2 43
FaiIed 2 10 12
LOW IMAGERS
Succeeded L4 I5 29
FaiIed 2 2l 23
TOTAL 49 58 107
succeeded at the task within the time limit. lltre time taken to
TABLE 13.4
SUBJECTS
DISCUSSION
bxperiencedr
'which
might imply that they were of high
imaging ability. It might also imply .that for subjects
of high imaging ability, performing the relatively simple
that this may not be the case where the task involves more than rotation.
It suggests that the relationship between imagery and other aspects of
problem-solving is a more complex one.
Subjects' reports in the present e>çeriment implied that they were not
trying to solve the task mentally in the same \^¡ay as physically.
Notwittrstanding the possibfe demand situation, it would appear from
subjects I reports that quite different techniques were used to solve
the task physicatly and mentally. T'he sequence in tJ-e physical
attempt was more random than the more considered imaginal sequence.
as reported. lllre apparent discrepancy between observed physical
strategies and reported imaginal ones offers some support to the
concern of other researchers (e.9. Vüinston, 19'73; Weisstein, I973)
that physical and imaginal sequences may not bottr proceed incrementally.
suggested that not all subjects could produce an image version of the
manipulations of tlre stimuli. Thre performance of subjects in tl-is
could produce an image, they could not all manipulate their i^tg"=.1
r: REVIEW OF EXPERII"IENTS
for aII subjects between reported irnage size and errors was found'
to be negative: subjects who reported using small images made
more errors. The central finding, that differences in pictorial
qualities of the imagPwere reflected in performance, suggested'
support for the pictorial view of imagery.
l-90.
Some subjects who had taken part in any of tJre preceding three
expe riments were invited back for Experiment 5. lllrey were
required to do the task either with visual interference, or
with auditory interference, or merely to repeat tl¡e task
under the same conditions as on the initial triaI. Visual
interference disrupted the perfornance of subjects of both
imagery leveIs, which suggested that imagery was necessary
for the task. Low imagers, unlike high imagers, tended to
would have been incorporated into the image. ft was found, however,
that the reported image did not contain such information, though
there was evidence from performance measures that the information was
present in some other form. The finding from this experiment suggests
that pictorial descriptions are appropriate for the representation of
the image,
The results from the interference experiment offer some support for
stimulus \^ras for the task. Such an interpretation would account for
the equivalent performance of high imaging subjects with large and
small stimuli.
2. Dete ts of Functionalitv
the processing mechanisms being employed for the task. While there
r97.
Not atl subjects, however, could use their images effectively. The
were more successful ín their use of imagery than were low imaging
particular kind and the processes which are applied to it, rather than
to the generally accepted notion of !good.' and tpoor' images,
Both high and l-ow imaging subjects made less errors with the sugar
cube stimulus than with the wooden stimulus and took longer to do
the task. The question could be raised as to whether the exÈra
The ease with which even low imaging subjects in the training
3. Accessibilitv to Intros'pection
Subjects t reports were concerned with the techniques they used for
solving the image task and with the properties of their images.
The reports of how they used their images related directly to the
degree of success in terms of error and time rates. To the extent
that subjects r reports of how they used their images were associated
with distinct patterns in performance, it would appear that some of
the functional aspects of the imagíng process are accessible to
introspection.
imagers haye more insight into theír cognitive processes and are
may be more open to description than the more complex imaging process
for high imaging subjects. It may be, as with skills generally,
that the image construction process had become automatic for high
inaging subjects and thus less arai.lable for reporting,
The most staþIe result throughout the experiments was the consistently
reflected the influence of arry stimulus feature which would have made
5. Sex Differences
Although the present research had not set out to investigate sex
differences in,ìmaging abitity, it did provide some information
releyant Èo the issue. As discussed in Chapter 4, previous research
on the relationship had been inconclusive. Attempts were made in
this research to counteract its possibte contamination by selecting
an equal nunber of high and low imaging alrility subjects from both
sexes.
Generally, the results indicated that males performed the task more
The present research required the use of an image, not merely the
passive irnage.
imaged. One coul-d suggest, that a possible task would be one involving
an image too complex to represent pictorially, such as those reported
by de Grootts chess masters (J.965),
images and problem solVing images have in corrunon that they are the
of an image, but the type of image prresent and the techniques apptied
to ìt which deter¡nine success'Ìn the irnaging task, The key findings
which supported this interpretation \,\'ere the superior performance of
207.
In spite of the finding that much of the solution process was not
available to introspection, there would appear to be no justification
for accepting this as inevitably so. IÈ is possible, for instance,
that subjects appropriately trained may be able to report more,
discussed ìn the first chapter, that there are many types of imagery,
The two which were examined, îemory imagery and problern solving imagery,
208.
aPPeared to have more in consnon with each ottrer than with the more
perceptually based imagery þerceptual, iconic, eidetic) or with
prirnary imagery (imagination, hypnagogic, hypnoponpic). !'or
instance, both memory and problem solvíng imagery involve retríeval
processes and would appear Èo be more readily controlled than are
other forms of imagery. Ttrough more open to control ttran other
forms¡ they emezged. as being distinguishable on this dimension.
Memory irnages are more likely to arrive unbidden, are quasi-
perceptual in nature, and can be more easily reinterpreted. It is
suggested that a problem solving image is constructed anew each time
can be achieyed.
to the term ri4ag"t. The only doubt might concern the appropriateness
rV: RECOMMENDATIONS
***
2LT.
LIST OF APPENDICES
Appendix No.
P.ge
APPENDIX I
Please do not Lurn to the next page until you have completed
the items on the page you are doing, and do not turn back to
check on other items you have done. Complete each page before
moving on to the next page. Try to do each iLem separately'
indepent of how you may have done other items.
RATING SCALE
The image aroused by an item of this test may be -
Perfectly clear and as vivid as the actual experience .. Rating I
Very clear and comparable in vividness to the actual
experience ... ..... Rating 2
Moderately clear and vivid Rating 3
Not clear or vivid, but recognizable Rating 4
Vague and dim ..... ..... Rating 5
So vague and dim as to be hardly discernibl-e Rating 6
No image present at allr vou only "knowing" that
you are thinking of the object Rating 7
L.2
EXÀMPLE
1r. Sand
L2. Linen
13. Fur
L4. The prick of a pin
15. The warmth of a tepid bath
2L. Salt
22. Granulated (white) sugar
23. Oranges
24. JeIly
25. Your favourite.soup.
1.4
3r. Fatigue
32. Hunger
33. A sore throat
34. Drowsiness
35. Repletion as from a very full meal.
2I4.
APPENDIX 2
INSTRUCTIONS:
***
2I5.
APPENDIX 3
RATING SCALE
Vividness of Imaqery
AUTOMOBTLB L23456 7
Low imagery High imagery
WRITER L23 456 7
Low imagery High imagery
DEMOCRACY 123 4s6 7
Low imagery High imagery
VAPOUR l-23 456 7
Low imagery High imagery
B M R -s-
_l
__-l---
E o --T P
I
1
5 v 6 A 7 F I L
w B G M N
L
Y D ) K o P
è
lo
217
APPENDIX 5
TNSTRUCTIONS
fn solving the problems of this test did you MAKE USE of whatever
visualizing was present in getting the relations between the
shapes, forms, spaces. Were visual images an essential part of
the process? Did they help you? Did you feel, rThis would be
easy if I could get a clear picture of it and hold it in my head'?
Did you try deliberatel-y to get images as your method of approach?
How important were they?
***
APPB\DIX 6
2r8.
SPACE IìELATIONS
DIIIECl'IONS
Iri'd Lhe place for Space lìelaLio's on the Answer sheet.
Example X.
B c D
In the test there will always be a row of four ñgures follorving each pattern
-
In every row there is only one correct figure
-
Now look at the pattern for Example Y and the four choices for it. Note that when the patter¡ is folded,
the frgure must have three gray surfaces. Two of these rvi]l be large surlaces either of which could be the top
or the bottom of a box. The olher is a small surlace which rvould be one end of the box.
Example Y
A B
D
20
o ) € V
I
C )
Þ
o I
t
0 f I V
¿
0 l s
t
z'9
2]-9 -
APPENDIX 7
How many cuts are needed to cut the cube inEo 27 small cubes?
Vlhere will they be?
time:
10. 1 black, I white
11. 2 black, I white
12. 3 bIack, 1 white
13. I b1ack, no white
14. 2 black, no white
15. 3 black, no white
16. 2 black, 2 white
]-7. 2 white, 1 black
18. 3 white, I black
19. I white, no black
20. 2 white, no black
2I. 3 white, no black
4 white, no black
ERRORS:
TIME:
220.
APPENDIX 8
4. When you were doing the task, did you consider the
whole cube intact, or did you work with faces, or
with indivÍdual cubes?
****
22I.
APPÐ{DIX 9: TABLE 6 1
ERRORS
SOI'RCE df MS F
Maln Effects
Imagery Level- 1 9.2 37. I ***
Condition I 0.2 I 0
Interactións
Imagery Leval x Condition I t.2 5.0
Côvariate
Sex t 3.9 15.9 ***
TTME
SOI'RCE df MS F
Main Effects
ûnagery Leve1 I 4I0.7 22.2 ***
Condition I 25.2 I.4
Interactions
ïmagery Level x Condition 1 0.2 0.0
Covariate
Sex t s3 .8 2.9
* p< .05
** p< .0r
*** p< .001
222
APPENDIX IO
ERORS
TTME
LOW IIIIAGERS t
s
.15 (p = .28')
N=52
***
223.
APPB{DIX 11
OUESTIONNATRE ON DETATLS
ERRORS
SOURCE df MS F
Mai-n Effects
Imagery Level I 7.2 33.5 ***
Condition I t.2 5.7 *
fnteractlons
Irnagery Level x Condition I .7 3.0
Cóvar iate
Sex I 4.3 19.6 ***
Residual 104 -2
TIME
SOURCE df MS F
Main Effects
Imagery Level I 496.2 33.0 ***
Condition I 2.5 o.2
Interactions
Imagery Level x Condition l_ 0.0 0.0
Covariate
Sex I 1.I 0.1
* p <.05
** p <.01
*** p < .001
p P
SUBJECÐS
D L
ll
HIGH 4.84 3.87 .77 X
rd
ll
COLOUR OF I .16 .7L9 .36 47 .722
E Þ
H
FIGURE E td
LOVü 4.46 3.59 .73 È lrl
z(,
(,
H
X
ts
HIGH 9.L2 3.37 .67 L^J
ERRORS
SOURCE df MS F
Main Effects
Imagery Leve1 t 19.s 52.2 ***
Condition I 2.2 5.8 *
Interactions
Imagery Level x
Condition I 0.2 0.6
Covar iate
Sex I 2.7 'r)**
Residual 87 0.4
TII\,IE
SOURCE ¡{f MS F
lvtain Effects
Interactions
Inagery Level x
Condition I 40.7 1.9
Covariate
Sex I 0.5 0.0
Residual 87 2I.8
* p¿ .05
** p¿ .0r
*** p¿- .00r
226a.
APPENDIX 14.22 TABLE 8.4
EXPERT¡4ENT 4: EFFECT OF PHYSICAL POSSTBILITY ON AN IMAGINAL TASK
ERRORS
SOURCE df MS F
Main Effects
Imagery level t L9.4 53.6***
Condition 1 2-O 5.6*
Interactions
Imagery level x 1 0.1 o.4
'Condiùion
Residual 85 o.4
TIME
SOURCE df MS F
Main Effects
Imagery level I 579 -A 29.2***
Condition 1 92.O 4.6*
Interactions
Imagery level x 1 62.7 3.2
Condition
Residual 85 19.9
* p<.05
** p<.01
*** p <.001
227 .
APPENDIX 15
trffi
fiíl il
z'sT
E'9T
ì
v'9f
r5.5
f
I
ffi
\
R
Effi
ffi
I
I
9'9r
228.
APPENDIX 16
SIDE 1
SIDE 2
Advertisements I 0
MusÍc 0 14
News I 6
Music 0 31
Discussion 11 t6
Advertisements t 40
Discussion 2 18
Advertisements 0 39
Discussion 9 40
Advertisements I 25
Discuss ion I I3
229
APPM¡DIX 17
ERRORS
SOURCE df MS F
Main Effects
Imagery Level 1 9.1 t9 .3 ***
Condition 2 0.2 0.5
fnteractions
Imagery level x Conditíon 2 0.2 0.5
Covariate
Sex t 0.2 0.5
Residual s6 0.5
TIIUE
SOURCE df IVIS F
Main Effects
Tmagery Level I 548.5 29.2 ***
Condition 2 10 .0 0.5
Interactions
Inagery level x Condition 2 r3 .0 o.7
CovarLate
Sex I 72.L 3.8
Residual 56 18 .8
* p< .05
** P< .01
*** p< .001
230.
APPE.iDIX I8: TABLE 9 .I
SOURCE df MS F
Main Effects
Inagery Level I 360.5 35.1 ***
Condition 2 r8.9 1.8
Tr ials I 36.3 10 .5 **
Interactions
lmag. Level x Cond. 2 6.0 0.6
Imag. Level x Trials I 7.5 2.2
Cond. x Trials 2 14 .0 4.L *
ûnag. Level x Cond.
x Trials 2 3.7 1.1
Residual
(inag. level,
cond. stratum) 54 10. 3 3.0
Residual
( i¡nag . level , cond . , 54 L5.T2 3.4
trials stratum)
TIME
SOURCE df MS F
Main Effects
Inagery Level I 14,19.0 26.2 ***
Condition 2 38 .5 0.7
Tr ials I 395.5 13.0 **
Interactions
lnag. Level x Cond. 2 42.0 0.8
I¡nag. Level x Trials I 4L.9 r.4
Cond. x Trials 2 86. 5 2.8
Imag. Level x Cond.
x Trials 2 30.2 1.0
Residual
(imag. level, cond., s4 54. r I.8
stratum)
Residual
(inaS. 1evel, cond., 54 30 .5
trials straturn)
* P< .05
** p< .01
*** p< .001
23I.
APPENDIX 19: TABLE IO 2
ERRORS
SOURCE df MS F
Main Effects
Imagery Level I 19 .9 69.4 ***
Divis ions I r.6 5.6 *
Movenent I 4.3 1s .0 ***
fnteractions
Imagery Level x Divisions I 0.r 0.2
Imagery Level x Movement I 0.3 0.1
Divísions x Movement I 0.0 0.0
ûnagery Level x Divisions
x Movement I 0.4 1.5
TÏME
SOURCE df MS F
Main Effects
Imagery Level I 896. 4 47 .8 ***
Divis ions I 1. I 0.r
Movement I 14. 4 0.8
fnteractions
Imagery Level x Divisions I 2 .5 0.1
Imagery Level x l"lovement I 4 .5 0.2
Divisions x Movement I 3 7 0.2
Imagery Level x Divisions
x Movement I 4.9 0.3
* p< .05
** p< .01
*** p< .001
232.
APPBIDIX 2.O ^ TABLE 11.2
ERRORS
SOI'RCE df MS F
Main Effects
Imagery Level 2 I40.4 33.4 ***
Condition I 0.0 0.0
Tr ials I L46.7 L2L.6 ***
Intefactions
Inag. Leve1 x Cond. 2 13 .8 3.3 *
Imag. Level x Tria1s 2 12.3 L0.2 **
Cond. x Trials I 22.0 L8.2 **
Inag. Leve1 x Cond.
x Trials 2 4 4 3.7 'r
TIME
SOURCE df MS F
Main Effects
Imagery Leve1 2 31 I 0.5
Condition I 1. 0 0.0
Trials I 259. 7 20.2 ***
Interactions
Inag. Level x Cond. 2 39 .3 0.6
Imag. Level x Trials 2 L26.2 9.8 **
Cond. x Trials 1 56.9 4.4 *
Imag. Level x Cond.
x Tria1s 2 I 4 0.1
Residual 36 66.7 5.2 *
( irnag . level , cond.
stratum
Residual 36 L2.8
(inag. levelr cond.¡
trials stratum)
* p<.05
** P<.01
*** p < .001
233.
SOURCE df MS F
I"lairr Effects
Imagery Level I 6677 .3 0.9
Method I 30245.r 4.2 *
Interactions
Imagery Level x
Ivlethod I L0672.5 r.5
Residual 68 7257.5
* p<.05
234.
B IBLIOGRAPHY
BARRATT, P.E.H. (1956) The use of the EEG in the study of imagery.
Brit. J Ps 47 , ror-r14.
DREVER,
'f. (1955) Some observations on the occipital alpha
rhythm. Quart. ,J. Ð<p. Psychol. , 7, 9r-97.
GOLLA, F., HUTTON, E.L. & GREY WALTER, W. (1943) The objective
study of mental imagery. I. Physiological
concomitants. J. Ment. Sci. , 89, 2L6-223.
GROOT de, A.D. (1965) Tt¡ouqht and Choice in Chess ftre Hague:
Mouton.
GRIFFITTS, C.H Individual dífferences ín imagery.
(J-927 )
Psychol. Monog. , 37, No. 3, 9I.
MICTIAEL, W.G.
'
GUILFORD , J. P. , FRUCHTER, B. & ZIMMERM.AN, I{. S .
(I957) lltre description of spatial-visualization
abilities. Educ. Psychol. Measmt., 17, Ig5 | Lgg.
MILLER, G.A. (L962) psycholoqy: The Science of Men tal Life.
N.Y.: Harper & Ro\4'.
MORRTSETT, L.H. Jr. (1956) The rote of implicit practice in
learning. Unpub. Doct. Diss., yale University.
to other
cognitive processes. psychometrika I L7,
267-288.
PYLYSHYN , Z.W. (1973) What the mindrs eye tells the mindts
brain: A critique of mental imagery.
Psvchol. BulI. , 80, I-24.
SEGAL, S.J. & GORDON, P.E. (1969) flre Perky effect revisited:
Blocking of visual signals by imagery.
Perc. & Mot. Sk. , 29, 79r-797 .
23, 279-299.
Kegan PauI.