Experimental Aging Research, 31: 409–420, 2005
Copyright # Taylor & Francis Inc.
ISSN: 0361-073X print/1096-4657 online
DOI: 10.1080/03610730500206725
OLDER ADULTS USE MENTAL REPRESENTATIONS THAT
REDUCE COGNITIVE LOAD: MENTAL ROTATION UTILIZES
HOLISTIC REPRESENTATIONS AND PROCESSING
Itiel E. Dror
Ina C. Schmitz-Williams
School of Psychology, University of Southampton, England, UK
Wendy Smith
School of Human Science, St Mary’s College, England, UK
Thirty-two participants (16 younger adults, mean age of 18, and 16 older
adults, mean age of 70) were examined to determine whether older adults
adopt mental representations and processes that are less taxing on the
cognitive system. Specifically, they were asked to mentally rotate a
variety of images with different levels of complexity to examine whether
they mentally rotate stimuli holistically or piecemeal; that is, whether
they rotate the image as a single undifferentiated unit or as a collection
of segments that are connected together to form the image. Using analysis
of variance (ANOVA) the authors observed that younger adults found the
more complex images harder to rotate, whereas the older adults rotated
the complex images with the same effort as the simple images. The data
reflected that older adults used holistic representations and processing in
visual mental rotation. This information-processing schema reduces the
use of cognitive resources as its underpinning because it is less computationally intensive. Furthermore, such a schema is more robust because it
is not dependant or affected by the complexity of the image. The younger
adults used piecemeal representations and processing. In contrast to the
holistic strategy, the piecemeal schema is more volatile because it entails
that the demands on the cognitive system vary with different images.
Received 26 January 2004; accepted 13 June 2005.
This research was supported by a research grant from the Nuffield Foundation. The authors
also want to thank Ailsa Peron, Mirja Petri, and Romola Bucks for their comments.
Address correspondence to Dr. Itiel Dror, School of Psychology, University of Southampton,
Southampton SO17 1BJ, England, UK. E-mail: [email protected]; http:==www.ecs.soton.ac.
uk=! id
410
I. E. Dror et al.
As we age different aspects of the brain deteriorates and our cognitive
resources decline (e.g., Briggs, Raz, & Marks, 2001; Dror & Kosslyn,
1998). Such reduction in resources does not necessarily manifest itself
as degraded cognitive performance. The biological and cognitive
systems dynamically compensate so as to adapt to the changes that
occur with aging (Reuter-Lorenz, 2002). For example, computer
neural network models show that at a biological level, with the functional decline of some neurons, other neurons take over and ‘lend
support’ in order to avoid degradation in processing capacity (Dror
& Morgret, 1996). The cognitive level also responds to challenges that
are presented with further decline in resources by adapting the cognitive processing strategies themselves (e.g., Salthouse, 1984; Raedt &
Ponjaert-Kristoffersen, 2000). The adaptation and compensation at
the biological and cognitive levels can be characterized relatively well
as a hierarchy of changes that occur within and between processing
modules (Dror, 1997).
In the present study, we investigated age-related changes in cognitive processing strategies used in mental imagery. Mental imagery
encompasses the use of images that are in the ‘‘mind’s eye’’ (Shepard
& Cooper, 1982). One of the most studied and well-documented
imagery processes is mental rotation. Mental rotation involves mentally transforming an image by rotating its position in space. The
hallmark of mental rotation is the linear increase in response times
with greater angular rotations (Shepard & Metzler, 1971). Detailed
research into mental imagery has demonstrated that mental rotation
can be achieved either via a piecemeal or via a holistic rotation
(Kosslyn, 1981; Dror, Ivey, & Rogus, 1997; Smith & Dror, 2001;
Sharps & Nunes, 2002).
Piecemeal rotation means that objects are represented and processed as a collection of segments that are connected to each other.
For example the letter ‘‘L’’ is comprised of two segments connected
at a right angle, and when rotated, each segment is manipulated separately, while keeping the correct right angle alignment between
them. In contrast, holistic rotation means that objects are represented and processed as a single undifferentiated unit (Kosslyn,
1981).
Holistic representations and processes are less flexible than piecemeal. They are less flexible in the sense that they have a much more
limited scope of use. At the same time they are also less volatile and
demanding on the cognitive system. They are less volatile in the sense
that they are less dependant on the actual input; whereas piecemeal
processing is highly dependant on the number of pieces=segments
Older Adults Use Holistic Mental Rotation
411
that comprise the stimulus, holistic processing is relatively unchanged
and unaffected by the details of the input.
Some types of stimuli lend themselves to either a more piecemeal or a holistic representation and processing. For example,
impossible objects are harder (if at all viable) to be dealt with,
or mentally rotated in a holist fashion (Dror et al., 1997); and
Sharps and Nunes (2002) used stimuli that had visual characteristics that made them either more appropriate for piecemeal or for
holistic visual mental rotation. In this study we set out to investigate whether the age of the participants, rather than the stimuli
characteristics, would mediate the mechanism of visual mental
rotation. Namely, whether older people would compensate for
decline in resources by using the more economical cognitive representation and processing.
Empirically, piecemeal and holistic mental rotations have different
patterns of performance. In both cases response times increase linearly with greater angular rotations. However, in a piecemeal rotation
the rate of linear increase (the slope) of response time depends on the
complexity of the image being rotated. Namely, with greater complexity of the images, the harder it is to rotate them greater angular
distances (Cooper & Podgorny, 1976). In contrast, a holistic rotation
entails that the same cognitive effort is needed to rotate images of different complexity (effort in general refers to the time needed to perform a task at certain levels of accuracy; in the context of mental
rotation, effort is reflected by the slope of mental rotation, that is,
the rate of rotation as measured by ms per angle of rotation).
Although the initial encoding may take longer for the more complex
images (reflected by the intercepts of the function), once they have
been encoded the rotation process itself (reflected by slope=rate of
the function of how response time increases per each angle of
rotation) is much less dependent on their complexity. Hence, the
intercepts of the linear function may be different, but both the simple
and complex images will have comparable rates of linear increase in
response time as a function of angle (the slope) regardless of the complexity of the image (for full details, see Cooper & Podgorny, 1976;
Folk & Luce, 1987).
This makes the holistic processing schema more robust because it
is relatively less dependent upon or affected by the images, whereas the
piecemeal processing schema varies much more in its computational
demands with different images. The present study examines the linear
function of mental rotation for simple and complex images in younger
and older adults.
412
I. E. Dror et al.
METHODS
Subjects
Thirty-two participants took part. Both the young and old participants were recruited from a village in Hampshire (near Winchester).
The initial criteria for recruitment was based on age (i.e., between 16
and 25 and over 55 years) and that all participants were healthy and
active (i.e., the older people did not live in a nursing home and were
actively and independently running their life). Furthermore after
initial recruitment, we made sure that none of the participants was
taking any medication that could affect their ability to perform the
task we used. From the 32 participants, 16 were older adults, with
a mean age of 69.9 years (age range from 59 to 83 years, SD ¼ 7.8
years). The other 16 participants were younger adults, with a mean
age of 18.1 years (age range from 16 to 20, SD ¼ 1.4 years). The ratio
of male to female was comparable in both age groups. All participants filled in a questionnaire that provided information about their
educational background, mobility, independence, emotional state, as
well as about their physical and mental health. None of the younger
and older participants reported any serious health, mental,
or emotional problems that would prevent them from partaking in
this experiment. Education was also comparable between the two
age groups; in both groups all participants had completed primary
and secondary schooling, and about half the participants continued
their education at university. All participants were paid for their
participation.
Materials
The stimuli were 24 drawings of objects, similar to those used by an
established rotation design (Smith & Dror, 2001). The stimuli varied
in their complexity: half were simple and the other half complex. The
difference between simple and complex stimuli was quantified by calculating the compactness of the drawing (e.g., Cooper & Podgorny,
1976; Podgorny & Cooper, 1983). The simple stimuli had a high compactness value and the complex stimuli had a low compactness value.
Following the paradigm of Smith and Dror (2001), each set of simple
and complex stimuli included equal numbers of meaningful and
meaningless images. The meaningless stimuli were created by
rearranging the parts of the meaningful stimuli (see Figure 1 for
examples).
Older Adults Use Holistic Mental Rotation
413
Figure 1. Example of stimuli. Top row depict simple stimuli and the bottom
row complex stimuli. Stimuli in the left column are meaningless, the right
column shows the corresponding meaningless stimuli; at the top, a simple
meaningless stimulus based on the left top stimulus of a house, and on the
bottom a complex meaningless stimulus based on the left bottom stimulus of
a helicopter.
This procedure enabled us to remove the meaning from the stimuli
while ensuring that the meaningless stimuli had the same components
and complexity as the meaningful stimuli, and indeed both had equal
measures of compactness values. Then, for each stimulus we created
a slightly distorted version to be used for the ‘different’ trials. The distortion encompassed slightly changing a part of the object (Figure 2).
Each of the 48 stimuli (24 objects and 24 distortions of them) were
rotated clockwise in the two dimensional plane to three different orientations: 0 (upright), 50, and 100 degrees (see Figure 1 for examples).
We constructed 144 experimental trials. Each trial consisted of a
pair of objects matched together. Each of the original 24 upright
objects (12 simple and 12 complex) was matched to objects rotated
to the three different orientations. In each of those orientations it
was matched to two objects, once to an identical image (for a ‘same’
trial) and once to the slightly distorted image of that object (for the
‘different’ trial’). Each slightly distorted image was modified in a different location, so participants could not anticipate where such discrepancies would appear and hence we need to encode and rotated
the entire stimulus (see Figure 2 for an example). This way we
produced a total of 144 mental rotations trials to be preformed by
each participant in the experiment (24 objects # 3 orientations # 2
414
I. E. Dror et al.
Figure 2. An example of two nonidentical stimuli. The two were presented
sequentially (see Figure 2); in this example the second stimulus is presented at
a 50$ orientation. In the first stimulus (left) the windows of the passengers and
cockpit are separate, but not in the second where they are together (right).
Each stimulus had six such modifications at different locations for each of
the nonidentical trials. The stimuli were presented sequentially and the modification made at different location so as to encourage participants to encode
and rotate the entire image.
same=different judgments). The order of the 144 experimental trials
was counterbalanced to avoid any order effects.
We also created some practice trials with additional simple and
complex objects that were not used during the actual experiment.
The full practice sequence began with practicing the actual use of
the ‘different’ and ‘same’ response buttons. The word ‘different’ or
‘same’ appeared on the screen 32 times (16 times each, in random
order) and participants needed to press the correct key. Feedback
was given if they pressed the wrong key. Then instructions were given
on the screen about what constitutes ‘different’ and ‘same’ stimuli,
and that followed by 16 trials (8 of each, in random order) of upright
stimuli, and participants practice determining if they were ‘same’ or
‘different.’ Feedback was provided if they made a mistake. Then,
after participants were well acquainted with the response buttons
and what constitutes ‘different’ and ‘same’ stimuli, they continued
to practice with 8 additional trials of stimuli that were actually
rotated as would be in the actual experimental trials. The stimuli that
were used during practice included all orientations, and all types of
stimuli that would be used in the actual experiment (meaningful
and meaningless, simple and complex), and all combination of factors
(angle of rotations) and possible responses (same=different) were
included to equal levels. Furthermore, participants were allowed to
ask questions only during the practice trials. The experiment was
administered using the experimental package SuperLab, version
1.75 (Cedrus Corporation, 1991).
Older Adults Use Holistic Mental Rotation
415
Procedures
Each participant completed a questionnaire about their background,
health, lifestyle, and other personal information. They were then
given instructions and the practice sequence. Following the practice
participants were administered with the 144 experimental trials. Each
trial began with the word ‘‘ready’’ appearing on the computer screen.
When the participant was ready, they pressed the spacebar to initiate
that trial. This was followed by a blank screen for 500 ms, and thereafter the upright version of an object was presented. Participants
studied the object for as long as they needed and then when they were
ready to continue they pressed the spacebar. A visual mask then
appeared on the screen for 500 ms (to avoid any after image) followed
by the next stimulus. This second stimulus was presented in one of the
three orientations. Participants were instructed to determine whether
the two objects were identical (by pressing the ‘same’ button) or
different (by pressing the ‘different’ button) regardless of orientation.
The second stimulus remained on screen until the participants made
their choice. After participants pressed either of the buttons, the next
trial began with the word ‘‘ready.’’ This continued until all 144
mental rotation trials were completed (see Figure 3).
The instructions were given to the participants on the computer
screen; this ensured that the same instructions were given to all
the participants. The instructions required participants to decide if
the two stimuli were identical regardless of orientation in the twodimensional plane. They were instructed to respond as quickly as
possible while remaining accurate in their responses.
RESULTS
The data were subject to analysis of variance (ANOVA), examining
performance as a function of angle of rotation, comparing simple
and complex stimuli within each age group. There were two variables:
angle of rotation (0$, 50$, and 100$) and complexity of the objects
(simple and complex). Error rates were very minimal and did not
permit a meaningful analysis. Response times were analyzed only
for correct responses (there is no reason to believe that the incorrect
responses do indeed reflect the process that is been examined).
Response times varied with the different angular rotations in both
age groups, F(2, 30) ¼ 20.06, p < .05, for the younger participants,
and F(2, 30) ¼ 14.58, p < .05, for the older participants (Figure 4).
We examined the linearity of the slopes of rotation and found (as predicted by reflected by the hallmark of mental rotation) that the data
416
I. E. Dror et al.
Figure 3. The sequence of events in a single experimental trial. Each trial
began with the word ‘‘ready’’ appearing on the computer screen. When the
participant was ready, they pressed the spacebar to initiate that trial. This
was followed by a blank screen for 500 ms, and thereafter the upright version
of an object was presented. Participants studied the object for as long as they
needed and then when they were ready to continue they pressed the spacebar. A
visual mask then appeared on the screen for 500 ms (to avoid any after image)
followed by the next stimulus. This second stimulus was presented in one of the
three orientations. Participants were instructed to determine whether the two
objects were identical (by pressing the ‘same’ button) or different (by pressing
the ‘different’ button) regardless of orientation. The second stimulus remained
on screen until the participants made their choice. After participants pressed
either of the buttons, the next trial began with the word ‘‘ready.’’ There were
144 trials in total.
constituted a linear slope. The younger participants required 1899 ms,
2477 ms, and 2587 ms to rotate greater angular distances, reflecting a
linear slope of 6.88 ms per each angle of rotation, with an R2 of .87.
The older participants required 2929 ms, 3573 ms, and 3744 ms to
rotate greater angular distances, reflecting a linear slope of 8.15 ms
per each angle of rotation, with an R2 of .90. A main effect of
complexity was found in the younger group of participants,
F(1, 15) ¼ 7.07, p < .05. The main effect of complexity per se does
not reflect mental rotation performance and is only manifested
in the intercept of the linear function (it may reflect initial encoding
of the image and so forth). Rotation itself is observed in the actual
linear function of the increased response time as angular rotation is
greater (the slope of the function, not the intercept). Both the simple
and complex stimuli reflected the linearity of mental rotation. Simple
stimuli were rotated at 2315 ms, 2683 ms, and 2926 ms for greater
angular distances, reflecting a linear slope of 6.12 ms per each angel
of rotation, with a R2 of .98; complex stimuli were rotated
at 2513 ms, 3368 ms, and 3405 ms for greater angular rotations,
Older Adults Use Holistic Mental Rotation
417
Figure 4. Response times as a function of angle of rotation. The older
adults rotated complex images with the same effort as rotating simple images
(as reflected by comparable slopes of rotation for both types of images),
whereas the younger adults found more complex images harder to rotate (as
reflected by the steeper slopes of rotation for complex images, relative to
the slope for rotating simple images).
reflecting a linear slope of 8.92 ms per each angel of rotation, with a
R2 of .78.
The critical data were hence to examine interactions between angle
of rotation and complexity of the objects. Such an interaction was
evident only with the younger participants, F(2, 30) ¼ 3.87, p < .05.
All other main effects and interactions were not significant
(p > .05). Furthermore, we wanted to make sure that our significant
findings were not due to levels in the degrees of freedom, so we computed a Huynh-Feldt (1976) correction. This analysis confirmed our
findings by preserving our results of what was significant and what
did not reach significance.
DISCUSSION
Several studies have examined whether performance ability in mental
imagery degrades with aging (Giray, 1985; Sharps & Gollin, 1987;
Craik & Dirkx, 1992; Hertzog, Vernon, & Rypma, 1993; Dror &
Kosslyn, 1994; Brown, Kosslyn, & Dror, 1998; Band & Kok, 2000;
418
I. E. Dror et al.
Briggs, Raz, & Marks, 2001; Sharps & Nunes, 2002). Most of these
studies follow similar lines of research to other cognitive domains. Such
studies compare performance ability across age groups. If they find
comparable performance levels of older and younger participants, then
they conclude that age does not degrade performance. If they find that
older adults do not reach the performance level of younger adults, then
they conclude that aging diminishes performance ability. This framework of research assumes that older and younger adults use the same
information-processing schema to accomplish a task, and hence differences in performance levels reflect different processing capacities.
Older adults may be using different mental representations and
processes than those used by younger adults. If this is indeed the case,
then it is not so critical to examine overall performance levels between
younger and older adults. Any overall differences may reflect different processing schema rather than just degradation in processing
power. Even if the overall performance level is comparable across
younger and older adults, one cannot conclude that the process is
robust to age and is the same across the age groups. Different processing schema may be adopted by older adults that produce overall
performance levels comparable to younger adults.
To examine the underpinning of cognitive aging, researchers need
to compare the pattern of performance under different conditions
within each age group rather than just compare overall performance.
Such patterns may reveal different processing schema across different
age groups.
The study reported here examined the representations and processing involved in visual mental rotations. Rather than comparing overall performance levels, it examined the pattern of rotation performance
across different complexity of images within each age group.
The pattern found in younger adults showed that when images
were more complex, the slope of increased response time as a function
of angular rotation was steeper. This reflected an increase in cognitive
load as images varied in complexity. The underpinning of such a pattern of performance is piecemeal rotation. In this schema objects are
represented and manipulated by their parts and how they are spatially organized together to form the object (Kosslyn, 1981). Therefore more complex objects have more complex representations and
more parts to manipulate and are thus harder to mentally rotate
(Cooper & Podgorny, 1976). This was the pattern of performance
found across simple and complex objects within the younger aged
group of participants.
A different pattern of performance was apparent in the older
participants. Here, the complexity of the objects did not vary the
Older Adults Use Holistic Mental Rotation
419
cognitive demands. Regardless of their complexity, the rate of
rotation (i.e., the slope of increased response time as a function of
angular rotations) remained unchanged. The underpinning of such
a pattern of performance is holistic rotation (Cooper & Podgorny,
1976). In a holistic rotation objects are represented and rotated as single undifferentiated units. In this processing schema rotation slopes
are relatively unaffected by the complexity of the object.
Although holistic mental rotations are less flexible than piecemeal,
older adults may adopt them because they are more economical and
predictable. They are more economical as they are simpler and hence
require less demand on cognitive resources and capacity to memorize
and manipulate. They are more predictable as their computational
intensity and demand are robust and do not vary across different
objects. It should be noted that the data upon which the interpretations are based are from analyses with the mean as the measure of
central tendency for each group. This type of analysis does not indicate that no younger or older person used the strategy represented by
the means intercept and slope analyses for the older or the younger
groups, respectively.
As we get older our cognitive system does not passively degrade.
Dynamic and active adaptations and compensations take place
in face of depletion in cognitive resources and capacities. Aging
and cognition can thus be viewed as an active development in
processing schema rather than a passive progression of decline and
deterioration.
REFERENCES
Band, G. & Kok, A. (2000). Age effects on response monitoring in a mental-rotation
task. Biological Psychology, 51, 201–221.
Briggs, S. D., Raz, N., & Marks, W. (2001). Age-related deficits in generation and
manipulation of mental images: I. The role of sensorimotor speed and working
memory. Psychology and Aging, 16, 449.
Brown, H., Kosslyn, S. M., & Dror, I. E. (1998). Aging and scanning of imagined
and perceived visual images. Experimental Aging Research, 24, 181–194.
Cedrus Corporation (1991). Superlab: General purpose testing package.
Cooper, L. A. & Podgorny, P. (1976). Mental transformations and visual comparison
processes. Journal of Experimental Psychology: Human Perception and Performance, 2, 503–514.
Craik, F. & Dirkx, E. (1992). Age-related differences in three test of visual imagery.
Psychology and Aging, 7, 661–665.
Dror, I. E. (1997). Computational adaptations and cognitive strategy changes as
compensation for age-related decline in cognitive resources. Society of Neuroscience Abstracts, 23, 1457.
420
I. E. Dror et al.
Dror, I. E., Ivey, C., & Rogus, C. (1997). Visual mental rotation of possible and
impossible objects. Psychonomic Bulletin & Review, 4, 242–247.
Dror, I. E. & Kosslyn, S. M. (1994). Mental imagery and aging. Psychology and
Aging, 9, 90–102.
Dror, I. E. & Kosslyn, S. M. (1998). Age degradation in top-down processing: Identifying objects from canonical and noncanonical viewpoints. Experimental Aging
Research, 24, 203–216.
Dror, I. E. & Morgret, C. C. (1996). A computational investigation of dendritic
growth as a compensatory mechanism for neuronal loss in the aging brain.
Society of Neuroscience Abstracts, 22, 1891.
Folk, M. D. & Luce, D. R. (1987). Effects of stimulus complexity on mental rotation
rates of polygons. Journal of Experimental Psychology: Human Perception & Performance, 13, 395–404.
Giray, E. F. (1985). A life span approach to the study of eidetic imagery. Journal of
Mental Imagery, 9, 21–32.
Hertzog, C., Vernon, M. C., & Rypma, B. (1993). Age differences in mental rotation
task performance: The influence of speed=accuracy tradeoffs. Journal of Gerontology: Psychological Sciences, 48, 150–156.
Huynh, H. & Feldt, L. S. (1976). Estimation of Box correction for degrees of
freedom from sample data in randomized block and split-plot designs. Journal
of Educational Statistics, 1, 69–82.
Kosslyn, S. M. (1981). The medium and the message in mental imagery: A theory.
Psychological Review, 88, 46–66.
Podgorny, P. & Cooper, L. A. (1983). Distribution of visual attention over space.
Journal of Experimental Psychology: Human Perception & Performance, 9, 380–393.
Raedt, R. & Ponjaert-Kristoffersen, I. (2000). Can strategic and tactical compensation reduce crash risk in older drivers. Age and Aging, 29, 517–521.
Reuter-Lorenz, P. A. (2002). New visions of the aging mind and brain. Trends in
Cognitive Sciences, 6, 394–400.
Salthouse, T. A. (1984). Effects of age and skill in typing. Journal of Experimental
Psychology: General, 113, 345–371.
Sharps, M. J. & Gollin, E. S. (1987). Speed and accuracy of mental image rotation in
young and elderly adults. Journal of Gerontology, 42, 342–344.
Sharps, M. J. & Nunes, M. A. (2002). Gestalt and feature-intensive processing:
Toward a unified model of human information processing. Current Psychology,
21, 68–84.
Shepard, R. N. & Cooper, L. A. (1982). Mental images and their transformations.
Cambridge, MA: MIT Press.
Shepard, R. N. & Metzler, J. (1971). Mental rotation of three-dimensional objects.
Science, 171, 701–703.
Smith, W. & Dror, I. E. (2001). The role of meaning and familiarity in mental transformations. Psychonomic Bulletin & Review, 8, 732–741.