You are on page 1of 14

Developmental Psychology

2014, Vol. 50, No. 3, 714 727

2013 American Psychological Association


0012-1649/14/$12.00 DOI: 10.1037/a0034293

Benefits of Training Visuospatial Working Memory in


YoungOld and OldOld
Erika Borella, Barbara Carretti, Alessandra Cantarella, Francesco Riboldi, Michela Zavagnin,
and Rossana De Beni

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

University of Padova
The purpose of the present study was to test the efficacy of a visuospatial working memory (WM)
training in terms of its transfer effects and maintenance effects, in the young old and old old. Forty
young old and 40 old old adults took part in the study. Twenty participants in each age group received
training with a visuospatial WM task, whereas the others served as active controls and completed
alternative activities. Training benefits were examined, considering (a) the specific training-related gains
in a visuospatial WM task (criterion); and (b) the transfer effects on measures of verbal WM, visuospatial
short-term memory, inhibition, processing speed, and reasoning. Maintenance of training benefits was
also assessed after 8 months. Results showed that the trained groups (both young old and old old), but
not the control groups, performed better in the WM measures and preserved these gains after 8 months.
Some transfer effects were found, but only in the young old-trained participants, and they were not
maintained at the follow-up. These results are discussed in terms of the efficacy of WM training for older
adults when a visuospatial modality is used.
Keywords: working memory training, aging, transfer and maintenance effects

(2010), both in normal aging (Borella et al., 2013; Carretti,


Borella, Zavagnin, & De Beni, 2013) and in patients with amnestic
mild cognitive impairment (Carretti, Borella, Fostinelli, & Zavagnin, 2013). Though it is not yet clear which variables are crucial
in favoring WM training-related gains in older adults (e.g., Morrison & Chein, 2011), Borella et al. suggested that their encouraging results were related to the following aspects: (a) the training
procedure and (b) the participants age range.
Concerning the first point, unlike other WM training studies,
Borella et al. (2010) used a procedure that combined an adaptive
procedure (the difficulty of the training task was increased if
participants were successful at a given level; if not, the lowest level
was presented) with a constant variation in the maintenance and
processing requirements of the trained task (to avoid practice
effects). This training regimen may consequently produce generalized effects on untrained tasks, because it combines the involvement of multiple cognitive processes (i.e., attentional control and
shifting) with an adaptive procedure (which enables participants to
be trained at a level of difficulty coming close to the limits of their
capacity). Because the requirements of the training task are always
novel and challenging, participants maintain their interest in the
activities, and their cognitive flexibility is stimulated. The training
schedule, with sessions arranged at fixed intervals, may be another
feature that made the training more successful because it left
participants sufficient time to consolidate the skills they acquired
while, at the same time, it reduced the risk of their losing any
beneficial effects of having practiced with the task (e.g., Cepeda,
Pashler, Vul, Wixted, & Rohrer, 2006).
As for the second point (the participants age range), to date
only the studies by Borella et al. (2010) and Brehmer, Westerberg,
and Bckman (2012) have considered WM training for participants
with ages in the range of 6575 in the former case, and 60 70 in

The crucial role of working memory (WM; i.e., the ability to


process and retain information temporarily for use in other cognitive tasks; Miyake & Shah, 1999) in explaining individual and
age-related differences in cognitive performance is well documented. Given the involvement of WM in everyday activities,
aging research is now focusing on whether WM training benefits
(a) can improve WM performance; (b) can be generalized to
untrained tasks (transfer effects); and, most importantly, (c) are
maintained in time.
Where studies have documented that older adults benefited from
WM training in a criterion task (i.e., the task in which participants
were trained), the benefits were very limited in terms of any
transfer effects, and they were generally not maintained over time
(see Borella, Carretti, Zanoni, Zavagnin, & De Beni, 2013; MelbyLervg & Hulme, 2013). In contrast with said results, Borella,
Carretti, Riboldi, and De Beni (2010) provided verbal WM training for older adults and found specific training gains, as well as
both near and far transfer effects, and the benefits of training were
maintained at 8-month follow-up. Their results, in terms of shortterm and maintenance training benefits, were also recently confirmed in other studies using the same procedure as Borella et al.

This article was published Online First September 23, 2013.


Erika Borella, Barbara Carretti, Alessandra Cantarella, Francesco Riboldi, Michela Zavagnin, and Rossana De Beni, Department of General
Psychology, University of Padova, Padova, Italy.
The study was partially supported by University of Padova Grant
CPDA087750/08, awarded to Barbara Carretti.
Correspondence concerning this article should be addressed to Erika
Borella or Barbara Carretti, Department of General Psychology, University
of Padova, Via Venezia, 8, 35131 Padova, Italy. E-mail: erika.borella@
unipd.it or barbara.carretti@unipd.it
714

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

VISUOSPATIAL WORKING MEMORY TRAINING

the latter, that is, the so-called young old (e.g., Baltes & Smith,
2003; Magaziner, 1989). Both studies showed specific gains in the
trained task with transfer gains to different cognitive processes/
mechanisms. Brehmer et al. (2012) found transfer effects on sustained attention, short-term memory, and on a measure of cognitive functioning (self-rating scale), but not on interference control,
reasoning, or episodic memory. In the study by Borella et al.
(2010), however, verbal WM training produced transfer effects on
visuospatial WM, verbal short-term memory, processing speed,
inhibition, and fluid intelligence.
Apart from the above two exceptions, all other WM training
studies focused on old old (individuals over 75; Buschkuehl et al.,
2008; Zinke, Zeintl, Eschen, Herzog, & Kliegel, 2012), or on a
broad age range (both young old and old old; Richmond, Morrison, Chein, & Olson, 2011). Indeed, the few studies examining
transfer effects in old old participants found transfer effects mainly
for tasks similar to those used in the training, and less so for different
tasks (e.g., Borella et al., 2013; Buschkuehl et al., 2008; Zinke et al.,
2012). Borella et al. (2013) focused on the potential role of age in
determining the success of training, applying the Borella et al. (2010)
training procedure to the old old. Their results show that there is
room for improvement in WM performance even in advanced old
age: The old olds performance improved in the criterion task (verbal
WM task), and this improvement was also maintained in time (at 8
months), but transfer effects (regarding some aspects of inhibitory
control) were more limited than those seen in the young old.
The main aims of the present study were therefore twofold.
First, we wanted to establish the efficacy of the Borella et al.
(2010) WM training (given its promising results) using a visuospatial WM training tasks. It is noteworthy that, among the few WM
training studies in aging, which identified few or no transfer
effects, the training task was either spatial (Li et al., 2008) or visual
(Buschkuehl et al., 2008), or combinations of verbal and visuospatial (Richmond et al., 2011; Zinke et al., 2012) WM tasks. In
addition, the results that emerged were irrespective of whether the
training procedure used was adaptive (Buschkuehl et al., 2008;
Richmond et al., 2011; Schmiedek, Lvdn, & Lindenberger,
2010; Zincke et al., 2012) or one that required practice (Li et al.,
2008) with the training task. Hence, our choice of a visuospatial
WM task with processing requirements were similar to those of the
verbal WM task used by Borella et al. (2010) (see Carretti, Mammarella, & Borella, 2012). Our second aim was to examine the
training-related benefits both in 64- to 75-year-olds (i.e., the socalled young old) and in 76- to 84-year-olds (i.e., the old old).
Transfer effects were examined for tasks theoretically related to
WM and were classified along a continuum from nearest to farthest transfer tasks (see Borella et al., 2010), following suggestions
from Noack, Lvdn, Schmiedek, and Lindenberger (2009). A
verbal WM task (the Categorization Working Memory Span
[CWMS] task) was used to assess nearest transfer effects because
it measured WM, but using a different task content from the
training task. The forward and backward Corsi span tests, both
usually considered as measures of visuospatial short-term memory
(e.g., Cornoldi & Mammarella, 2008), were used to assess near
transfer effects because they reveal the relationship between shortterm visuospatial memory and WM (e.g., Miyake, Friedman, Rettinger, Shah, & Hegarty, 2001). Tasks assessing reasoning ability
(the Cattell test), processing speed (the pattern comparison task),
and inhibition (the Stroop color task) were used to measure far

715

transfer effects because these mechanisms have been shown to


correlate with, or explain, the age-related decline in WM.
Consistently with the literature, we expected to see specific
training benefits in the criterion task as well as their maintenance
in the trained groups, irrespective of their age, by comparison with
the active control groups (see Melby-Lervg & Hulme, 2013).
Concerning the transfer effects, we expected to find the same
transfer effects, and maintenance effects, as were seen after administering the verbal WM training by Borella et al. (2010) to
young old. We hypothesized, however, that the gains would be
less broad because of the greater age-related memory decline in
visuospatial than in verbal tasks (Bopp & Verhaeghen, 2007;
Jenkins, Myerson, Joerding, & Hale, 2000; Myerson, Emery,
White, & Hale, 2003). In this sense, if the training task modality has
a crucial role, then transfer and maintenance effects should be less
obvious or lacking, consistently with other WM training studies using
nonverbal material.
For old old adults, also in the light of the results obtained with
the old old by Borella et al. (2013) (see also Buschkuehl et al.,
2008; Zinke et al., 2012), limited transfer effects were expected
because of the decline in flexibility with aging (Schmiedek et al.,
2010), and any benefits of training were expected to be influenced
by age, with greater gains in the young old than in the old old
(e.g., Borella et al., 2013).

Method
Participants
Forty young old (age 6575 years) and 40 old old adults (age
76 84) took part in the study. Twenty participants from each age
group were randomly assigned to receive training, whereas the other
20 in each age group formed a control group.1 All participants were
healthy, native Italian speakers who lived independently, and who
volunteered for the study. They were selected on the basis of a
physical and psychological health questionnaire. Exclusion criteria
were as follows: Use of benzodiazepines in the previous 3 months;
visual, auditory, and/or motor impairments; problems or diseases
potentially causing cognitive impairments (Crook et al., 1986); a score
below the age- and education-matched norms in the Wechsler Adult
Intelligence ScaleRevised (WAISR; Wechsler, 1981) vocabulary
test crystallized intelligence measure (Italian norms by Orsini &
Laicardi, 2003); a cognitive dysfunction such as mild cognitive impairment or Alzheimers disease. Participants were screened for cognitive impairments using the short version of the Italian Checklist for
the Multidimensional Assessment (SVAMA) of the elderly used in
the Veneto region (Gallina et al., 2006).
Moreover, all participants performed above the cutoff for their
age and education in the CWMS task (see Italian norms, De Beni,
Borella, Carretti, Marigo, & Nava, 2008).
Within each age group, the participants in the trained and active
control groups responded correctly to the 10 items of the SVAMA,
1
After a plenary session where the elderly adults were told about the two
different programs, both involving memory, the 40 young old and 40
old old volunteers were randomly assigned to two groups, 20 (for each age
group) to receive training and the other 20 (for each age group) to form the
control group.

BORELLA ET AL.

716
Table 1
Characteristics of the Groups
Years of
education

Age
Group

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Trained
Youngold
Oldold
Control
Youngold
Oldold

Vocabulary

SD

SD

SD

69.90
79.60

2.79
2.28

10.65
8.75

2.50
1.33

49.25
50.15

5.82
4.57

69.55
79.70

2.89
2.30

10.65
8.90

2.96
1.41

48.80
49.95

4.72
5.37

indicating a good cognitive functioning.2 One-way analyses of


variance (ANOVAs) demonstrated that they did not differ in age,
years of formal education, or WAISR vocabulary score (all Fs
1) (see Table 1).

Materials
Criterion task: The matrix task (adapted from Carretti et
al., 2012; Cornoldi, Bassani, Berto, & Mammarella, 2007).
The material consisted of 60 4 4 matrices (black lines against a
white background; see Figure 1). Series of matrices were presented
in increasingly long sets from two to six. In each series, three black
dots appeared in different positions in the matrix one after the
other for 1,000 ms, separated by an empty matrix appearing for
500 ms. At the end of each series, a gray screen appeared for 500
ms. Participants had to recall the position of the last dot they had
seen in the series of matrices (maintenance phase). In each series,
one row and one column were always shaded in gray, and participants had to press the space bar when the dot they saw occupied
a gray cell (processing phase). A random procedure was used to
decide which row and column to shade gray in a given set. When
a set ended, participants were shown a screen with an empty
matrix, and the experimenter asked them to use a mouse to indicate
the positions of the last dots (from two to six) seen in each of the
sets of matrices presented in the series. This empty matrix remained on the screen until the positions had been chosen. The start
of a new series was marked by the presentation of a fixation point.
Two trials were completed before the experiment began.
The total number of correctly recalled dot positions was considered the
measure of the participants WM capacity (maximum score 60).
Nearest transfer effects: Verbal WM task.
Categorization Working Memory Span Task (De Beni et al.,
2008). The task consisted of 20 lists of words organized into sets
of word lists of different length (from two to six). Each list
included five words of high-medium frequency according to the
Italian repertoire by Barca, Burani, and Arduino (2002). Participants listened to a set of audio-recorded word lists, presented at a
rate of one word a second and were asked to tap on the table with
their hand whenever they heard an animal noun (processing
phase). The interval between word lists was 2 s. At the end of a set
of word lists, participants were asked to recall the last word in each
list in consecutive order (maintenance phase); that is, they had to
remember from two to six words, depending on the level of
difficulty of the set of lists. Two practice trials consisting of
two-word lists (and requiring the recall of two words) were administered before the experiment began.

The total number of correctly recalled words was used as the


measure of WM performance (maximum score 20).
Near transfer effects: Short-term memory tasks.
Forward and backward Corsi tasks (adapted from Corsi,
1972). These tests consist of a series of nine blocks randomly
arranged on a wooden tablet. The cubes are numbered on the
experimenters side of the board to facilitate the tests administration. The blocks are tapped by the examiner in a random sequence,
and participants are asked to reproduce the same sequence. Increasing numbers of blocks are tapped, and participants have to
repeat the sequence in the same (forward) or reverse (backward)
order. The sequences are presented at a rate of one cube per
second. The tests started with three and increased to eight cubes
being tapped for the forward task, and from two to seven cubes for
the backward task. Each level contained two sequences of the same
length. After two consecutive recall errors, the task was discontinued. A practice trial with two sequences was administered for
each type of task before the test started. One point was awarded for
each correctly recalled sequence.
The final score corresponded to the total number of correctly
recalled sequences (maximum score 12, for both tasks).
Far transfer effects: Processing speed (pattern comparison
test), inhibition-related processes (Stroop color task) and fluid
intelligence (Cattell test).
Pattern comparison task (adapted from Salthouse & Babcock,
1991). In this task, participants were asked to decide whether arrangements of line segments, presented on two pages, were identical
or not. The stimuli for pattern comparison consisted of two pages,
each containing one column of 30 items. The stimuli were constructed
of three-, six-, or nine-line segments. The items of different difficulty
were counterbalanced so that 10 items of three, six, or nine segments
were presented on each page. The experimenter used a stopwatch to
record the time to complete each page. Three practice trials were
given before the experiment started.
The dependent variable was the total time taken to complete
responses for the two pages.
Stroop color task (adapted from Trenerry, Crosson, De Boe, &
Lever, 1989). This paper task consists of six cards with lists of
names of colors printed in incongruent colors (incongruent condition), names of colors printed in congruent colors (congruent
condition), and color patches (control condition), with two cards
containing 20 stimuli each for each of these conditions. Participants were asked to name the color of each stimulus and to process
the stimuli as quickly and as accurately as possible. Using a
stopwatch, the experimenter recorded response latencies for all
conditions (the times elapsing between naming the first and last
stimuli), and noted accuracy by hand on a prepared form.
To control for individual differences at the baseline (see Ludwig, Borella, Tettamanti, & de Ribaupierre, 2010), an interference
index was calculated as the time spent on the incongruent condition minus the time spent on the control condition divided by the
time spent on the control condition, so that a higher score implied
a greater difficulty in controlling the prepotent response in the
incongruent condition.
2
This cognitive assessment generates a score between 0 (perfect cognitive
functioning) and 10 (presence of a cognitive disorder). All participants obtained the
maximum score (i.e., 0), indicative of a good cognitive functioning.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

VISUOSPATIAL WORKING MEMORY TRAINING

717

Figure 1. Example of the stimuli used in the visuospatial working memory task (the matrix task). Participants
were shown black dots occupying different positions in a series of matrices consisting of 4 4 squares; six dot
positions are presented in this example. Participants were asked (a) to press the space bar whenever a dot
occupied a gray cell and (b) to indicate the position of the last dot seen in each matrices in an empty matrix
appearing immediately afterwards.

Culture fair test, scale 3 (Cattell & Cattell, 1963). Scale 3 of


the Cattell test consists of two parallel forms (A and B), each
containing four subtests to complete in 2.5 4 min (depending on
the subtest). Participants were asked to complete incomplete series
of abstract shapes/figures by choosing from among six options that
best completed the series (Series-Subset 1). Then they had to solve 14
problems involving abstract shapes/figures, and had to choose which two
of the five differed from the other three (Classifications-Subset 2). In
the Subtest 3 (Matrices), they were presented with 13 incomplete
matrices containing four to nine boxes of abstract figures and
shapes plus an empty box and six choices: Their task was to select
the answer that correctly completed each matrix. Participants were
then presented with 10 sets of abstract figures and lines, and a
single dot, along with five options, and they had to assess the
relationship between the dot and the figures and lines, then choose
the alternative in which a dot could be positioned in the same
relationship (Conditions-Subset 4).
The dependent variable was the number of items correctly
answered across the four subsets (maximum score 50).
For each of the tasks presented, two parallel versions were devised and
administered in a counterbalanced order across training sessions.

Procedure
Participants in the trained and control groups attended five
individual sessions: the first and last sessions were for pretest and
posttest purposes; in the other three, each control group was
involved in alternative activities (see Table 2), while the trained
groups received training. Each session lasted about 60 min and was
conducted by the experimenter, who explained the activities in-

volved and presented the materials. The training was completed


within a 2-week time frame, with a fixed 2-day break between one
training session and the next. Sessions ended with the experimenter asking participants how they felt about the activities conducted and reminding them of the date of their next meeting. The
schedule was identical for the trained and control groups, thus
enabling the amount of social interaction to be matched.
The training consisted of three sessions (Sessions 2, 3, 4) in
which participants received training with versions of the visuospatial WM taskthe matrix task(see below), modified in terms of
the amount of information to recall and the processing and maintenance requirements presented on a computer screen (see Table
2). The experimenter showed participants a series of matrices
organized in the same way as for the matrix task and asked them
(as in the matrix task) to recall the positions and to press the space
bar whenever a dot fell in a gray cell.
In the trained task, participants always had to follow the same
basic instructions (i.e., to press the space bar whenever a dot fell
in a gray cell), and the subsequent steps/manipulations were introduced during the three training sessions, as in Borella et al.
(2010), to facilitate transfer effects and limit the adoption of
task-specific strategies: (a) The maintenance demand was manipulated by increasing the number of positions to recall when participants correctly recalled the positions in a given series, or
presenting a lower memory load if they made mistakes (Session 2);
(b) the task was varied, requiring the recall of the last or first
position (Sessions 2 and 4); and (c) the processing requirement
(pressing the space bar when a dot appeared in a gray cell) was
manipulated by varying the number of gray cells (Session 3). The

BORELLA ET AL.

718
Table 2
Description of the Session Content by Group
Session
1. Pretest

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

2. Training

3. Training

4. Training

5. Posttest
6. Follow-up (8 months)
Note.

Trained group

Control group

Health interview, vocabulary test, forward and backward Corsi tasks, pattern comparison test, CWMS task, Stroop
color task, matrix task, Cattell test
WM training: Increasingly long series comprising from two to five sets
Autobiographic memory questionnaire
of 4 4 matrices, presented one after the other. The matrices
(from De Beni et al., 2008)
always had one row and one column shaded in gray. Participants
had to recall the positions of last presented dots and press the space
bar when a dot occupied a gray cell. The WM training task included
three phases, completed sequentially for each level of difficulty (or
length of the series): In Phase 1, participants had to recall the
position of the last dot in each series of matrices; in Phase 2, they
recalled the position of the first dot in each series of matrices; and
in Phase 3, they recalled the position of the last dot again. In each
phase, if the position of the dots to recall was correctly remembered,
the tasks difficulty was increased, up to sets containing five series
of matrices. If a mistake was made in one of the three phases,
participants were presented with the next set of matrices, starting
from the easiest level, and asked to recall either the first (Phase 2)
or the last position (Phase 3) of the dot.
WM training: Four sets of matrices for each different length (from two
Psychological well-being questionnaire
to five). The complexity of the task was manipulated by reducing or
(from De Beni et al., 2008)
increasing the number of gray cells.
For each matrix, participants had to press the space bar whenever a dot
occupied a gray cell, as well as remembering the position of the last
dot displayed in each matrix.
WM training: Four sets of different difficulty (involving from two to five
Memory sensitivity questionnaire
positions to recall). Participants were asked to press the space bar
(from De Beni et al., 2008)
whenever a dot occupied a gray cell, and had to recall in (a) the first
set, the last positions displayed; (b) the second, the first positions; (c)
the third, the last positions; and (d) the fourth, the first positions.
Forward and backward Corsi tasks, pattern comparison test, CWMS task, Stroop color task, matrix task, Cattell test
Forward and backward Corsi tasks, pattern comparison test, CWMS task, Stroop color task, matrix task, Cattell test

CWMS Categorization Working Memory Span task; WM working memory.

trained group was given no instructions on specific strategies, and


no feedback was provided.
The participants in the active control group met the experimenter for the same number of sessions as the trained group and
for approximately the same amount of time. Sessions 1 and 5 were
the same as for the trained group, while participants were asked to
complete different questionnaires on their autobiographical memory, memory sensitivity, and psychological well-being during Sessions 2, 3, and 4 (see Table 2).
All participants attended a follow-up session 8 months later to
assess maintenance effects, when the tasks were administered
again, in the same versions and in the same order as at the first
(pretest) session.

Data Analyses
First, preliminary analyses were conducted at the pretest session
to identify differences within each age group. Second, to examine
training-related gains, a 2 (group: trained, control) 3 (session:
pretest, posttest, follow-up) mixed-design ANOVA was run separately for each age group on all the measures of interest.
Third, to compare the benefits of training in young old and
old old, we calculated the standardized gains (e.g., Buschkuehl et
al., 2008; Schmiedek et al., 2010) to check for individual and
age-related differences between the two age groups at the pretest

session. Standardized gains3 were computed for short-term gains


([individual posttest score individual pretest score]/pretest SD)
and for maintenance gains ([individual follow-up score individual pretest score]/pretest SD).
Two (age group: young old, old old) 2 (training group:
trained, control) ANOVAs were run on the standardized gain
scores to identify significant short-term and maintenance effects.
Interactions were analyzed using post hoc analyses that consisted
of both within-group, paired sample t tests (two-sided), and across
group, unpaired t tests. Post hoc analyses were corrected for
multiple comparisons, with Bonferronis adjustment for multiple
comparisons. The value was set at .05 for all statistical tests and
at .014 for interactions.
3
Standardized gain scores can be considered as equivalent to effect sizes
and are commonly used to assess training-related gains (see training studies
by Buschkuehl et al., 2008; Schmiedek et al., 2010). They offer the
advantage of allowing for variability at the pretest stage to be controlled,
whereas classical measures of effect size adjust for the difference in means
by pooling the pre- and posttest standard deviations. We could not use
these scores to compare our young old results with those of the Borella et
al. (2010) study, however, because the latter article used Cohens d. The
only way to compare the two study populations was therefore for us to
calculate the classical measure of effect size (Cohens d).
4
For the interactions, the alpha value was set at .01 because nine
comparisons were conducted (.05/9 .006).

VISUOSPATIAL WORKING MEMORY TRAINING

Finally, to clarify the influence of the training task modality, the


benefits of training for the young old in the present study were
descriptively compared with those reported for the young old sample
in Borella et al. (2010) using Cohens d. The two studies considered
a sample of young old with similar demographic characteristics.5

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Results
The preliminary analyses one-way ANOVAsrevealed no
significant differences within each age group at the pretest session
(for all the measures within each age group, except for the forward
Corsi span task in the old old group), F(1, 39) 2.03, p .16
(F 1; see Table 3).

Training-Related Gains in Each Age Group


The results of the ANOVA are presented in Table 4 and summarized in Table 5.
Young old.
Criterion task: The matrix task. The main effects of group and
session were significant. The trained participants correctly recalled more
positions than the controls (MDiff. 4.83, p .001), and all participants
performance improved from pretest to posttest (MDiff. 6.00, p
.001; MDiff. 3.80, p .001, respectively) but decreased from
posttest to follow-up (MDiff. 2.20, p .001).
The Group Session interaction was also significant. Post hoc
comparisons showed that the trained participants performed better
at the posttest and follow-up sessions than at the pretest session,
t(19) 18.26, p .001, r .87; t(19) 8.45, p .001, r .79,
respectively, but worse at the follow-up than at the posttest session, t(19) 11.02, p .001, r .92. The control group
showed a significant improvement from pretest to posttest, t(19)
4.49, p .001, r .92, but decreased significantly from posttest
to follow-up, t(19) 4.46, p .001, r .94. The trained group
outperformed the control group at both the posttest, t(38) 5.88,
p .001, and the follow-up sessions, t(38) 5.69, p .001.
Transfer effect.
Nearest transfer effect.
CWMS task. The main effects of group and session were
significant. Trained participants recalled more words correctly
than controls (MDiff. 3.42, p .001); participants performance
improved consistently from the pretest to the posttest and
follow-up sessions (MDiff. 2.80, p .001; MDiff. 2.37,
p .001, respectively), which did not differ. The Group Session
interaction was also significant. Post hoc comparisons indicated
that trained participants performed better at posttest, t(19) 8.73,
p .001, r .27, and follow-up, t(19) 9.27, p .001, r .51,
than at pretest. This group also maintained its better performance
from posttest to follow-up. No significant difference was found for
the control group. The trained group outperformed the control
group at both posttest, t(38) 7.54, p .001, and follow-up,
t(38) 10.56, p .001.
Near transfer effects.
Forward Corsi task. The main effect of session, but not of
group, was significant. All participants performed better in the
posttest than in the pretest session (MDiff. 0.88, p .001),
whereas at follow-up, they did worse than at the posttest session
(MDiff. 0.75, p .001), and not differently from their
performance at the pretest session. The Group Session interac-

719

tion was also significant. Post hoc comparisons showed that


trained participants performed better at the posttest than at the
pretest session, t(19) 6.11, p .001, r .28, and at the
follow-up, t(19) 5.48, p .001, r .20, when they returned to
the same level of performance as at the pretest stage. No significant differences emerged in the control group. The trained group
outperformed the control group at the posttest, t(38) 3.72, p
.001, but not at the follow-up session.
Backward Corsi task. The main effects of group and session
were significant. Trained participants performed better than controls (MDiff. 0.52, p .001); all participants performance
improved consistently from the pretest to the posttest sessions
(MDiff. 0.68, p .001), but returned to pretest levels at the
follow-up session, with a significantly worse performance than at
the posttest session (MDiff. 0.43, p .001). The Group
Session interaction was also significant. Post hoc comparisons
showed that trained participants did better at the posttest than at the
pretest sessions, t(19) 6.57, p .001, r .50, but not at the
follow-up. Their performance declined from posttest to follow-up,
t(19) 4.66, p .001, r .37. No significant differences came
to light in the control group. The trained group only outperformed
the control group at the posttest session, t(38) 6.09, p .001.
Far transfer effects.
Pattern comparison test. The main effect of session and the
Group Session interaction were significant. The main effect of
group was not significant. All participants improved consistently in
their performance (taking less time to complete the test) from the
pretest to the posttest and follow-up sessions (MDiff. 11.15, p
.001; MDiff. 5.58, p .01, respectively), though they took longer
at the follow-up than at the posttest session (MDiff. 5.58, p .01).
As for the interaction, post hoc comparisons that trained participants
completed the task faster at the posttest and follow-up than at the
pretest session, t(19) 9.96, p .001, r .82; t(19) 4.45, p
.001, r .73, respectively, but they were slower at the follow-up than
at the posttest session, t(19) 5.64, p .001, r .89. The control
group revealed no significant differences across the sessions. The
training group only outperformed the control group at the posttest
stage, t(38) 5.64, p .007, r .89.
Stroop color task. No significant effects were found.
Cattell test. The main effect of session was significant: Performance was better in all participants at the posttest than at the pretest
session (MDiff. 1.28, p .001). However, follow-up performance
did not differ from pretest and was lower than that at posttest
(MDiff. 0.90, p .01). No other effects were significant.
Old old.
Criterion task: The matrix task. The main effects of group
and session were significant. Trained participants recalled more
dot positions correctly than controls (MDiff. 3.12, p .01). All
participants performance improved consistently from pretest to
posttest and follow-up (MDiff. 4.22, p .001; MDiff.
3.55, p .001, respectively), with no difference between the
latter two. The Group Session interaction was also significant.
Post hoc comparisons indicated that the trained participants performed better at the posttest than at the pretest session, t(19)
5
The present sample of young old did not differ in terms of demographics (age, education, and vocabulary level) from the one in the Borella
et al. (2010) study.

BORELLA ET AL.

720

Table 3
Descriptive Statistics for the Measures of Interest by Group (Trained and Control) for Each Age Group (YoungOld and OldOld)
Young old
Trained
Variable
Matrix task (max score 60)
CWMS test, correct recall (max score 20)

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Forward Corsi span test (max score 12)


Backward Corsi span test (max score 12)
Pattern comparison test
Stroop color test

Cattell test (max score 50)

Old old
Control

Trained

Control

Group

SD

SD

SD

SD

Pretest
Posttest
Follow-up
Pretest
Posttest
Follow-up
Pretest
Posttest
Follow-up
Pretest
Posttest
Follow-up
Pretest (in s)
Posttest (in s)
Follow-up (in s)
Pretest
Contr. RT (in s)
Incong. RT (in s)
Index
Err. Incong
Posttest
Contr. RT (in s)
Incong. RT (in s)
Index
Err. Incong
Follow-up
Contr. RT (in s)
Incong. RT (in s)
Index
Err. Incong
Pretest
Posttest
Follow-up

25.05
35.30
32.15
9.05
14.15
13.80
4.70
6.05
4.80
4.65
5.90
5.10
100.33
83.38
92.20

4.67
5.04
3.67
1.76
2.45
1.77
0.73
0.94
0.62
0.59
0.72
0.64
12.02
13.74
9.13

25.25
27.00
25.75
8.75
9.25
8.75
4.70
5.10
4.85
4.65
4.75
4.70
101.20
95.85
98.18

4.46
3.80
3.43
2.17
1.55
1.21
0.66
0.64
0.49
0.81
0.44
0.47
18.48
13.71
6.32

13.85
20.50
18.90
4.65
5.70
5.60
3.65
3.95
3.80
3.45
3.70
3.55
137.28
127.93
130.58

3.65
3.68
2.83
0.75
0.47
0.50
0.59
0.60
0.62
0.51
0.47
0.51
12.45
10.55
8.29

13.35
15.80
15.40
4.85
5.05
4.90
3.40
3.75
3.40
3.30
3.65
3.50
138.10
135.20
131.98

2.28
2.44
2.01
0.81
0.39
0.45
0.50
0.55
0.50
0.47
0.49
0.51
8.43
7.56
6.57

22.53
34.34
0.54
0.60

2.77
6.72
0.24
0.82

20.42
32.05
0.63
0.95

4.76
5.11
0.37
0.89

32.69
46.42
0.43
0.55

2.62
3.58
0.16
0.83

31.63
47.30
0.50
0.70

2.55
3.55
0.16
0.66

20.40
30.11
0.51
0.40

4.93
6.14
0.24
0.68

20.41
32.43
0.64
0.60

3.92
6.62
0.45
0.75

26.89
42.01
0.58
1.20

2.51
2.86
0.20
1.00

27.02
43.99
0.64
1.10

2.25
3.47
0.22
0.97

21.54
32.13
0.50
0.30
17.50
19.25
17.75

3.70
5.71
0.20
0.47
3.62
4.34
3.58

20.61
32.38
0.60
0.45
16.95
17.75
17.45

3.27
4.60
0.29
0.69
2.65
2.24
1.79

28.90
43.73
0.52
1.20
11.55
12.55
11.90

2.43
2.36
0.15
0.77
2.06
2.26
2.00

28.85
44.89
0.56
0.70
11.35
12.20
11.90

2.05
2.78
0.17
0.73
2.08
1.70
1.59

Note. CWMS Categorization Working Memory Span task; Contr. control condition; RT response time in seconds; Incong. incongruent
condition; Index interference index; Err errors.

17.52, p .001, r .89; t(19) 10.89, p .001, r .82, but this


benefit was not maintained at the follow-up stage, t(19) 4.46,
p .001, r .91. The control group performed better at the
posttest and follow-up (which did not differ) than at pretest,
t(19) 15.39, p .001, r .98; t(19) 4.15, p .001, r .47,
respectively. The trained group nonetheless outperformed the control group at both the posttest and the follow-up sessions, t(38)
5.34, p .001; t(38) 4.51, p .001, respectively.
Transfer effect.
Nearest transfer effect.
The CWMS task. The main effects of group and session were
significant. The trained participants correctly recalled more words
than the controls (MDiff. 0.38, p .01). All participants
performance consistently improved from the pretest to the posttest
and follow-up sessions (MDiff. 0.63, p .001; MDiff.
0.50, p .01, respectively) (remaining the same at the latter two
time points). The Group Session interaction was also significant. Post hoc comparisons showed that trained participants performed better at both the posttest and follow-up than at the pretest
sessions, t(19) 7.76, p .001, r .59; t(19) 6.19, p .001,

r .45, respectively, maintaining their improvement 8 months


after receiving training. No significant differences were seen in the
control group. The trained group outperformed the control group at
both the posttest, t(38) 4.74, p .001, and the follow-up
sessions, t(38) 4.65, p .001.
Near transfer effects.
The forward Corsi task. The main effects of group and
session were significant. Trained participants outperformed
controls (MDiff. 0.28, p .05). All participants performance improved from the pretest to the posttest session
(MDiff. 0.32 p .01), whereas follow-up performance did
not differ from the other two sessions. The Group Session
interaction was not significant.
The backward Corsi task. The main effect of session was
significant: All participants performed consistently less well at the
pretest than at the posttest session (MDiff. 0.30, p .05). No
other effects were significant.
Far transfer effects.
Pattern comparison test. The main effect of session was significant. Participants performance was consistently better, with

VISUOSPATIAL WORKING MEMORY TRAINING

721

Table 4
Mixed-Design 2 3 ANOVA Results for the Measures of Interest, With Group (Trained and Control) as the Between-Subjects Factor
and Session (Pretest, Posttest, and Follow-up, Respectively) as Repeated Measures for Each Age Group (YoungOld and OldOld)
Young-old
Variable

df

Old-old
n2p

df

MSE

n2p

0.27

1,38

13.46

21.64

0.26

2,76
2,76

0.80
0.69

112.88
32.82

1.83
1.83

0.75
0.49

7.45
7.45
7.45
1.48
1.48

0.55
0.55
0.55
0.62
0.56

1,38
1,38
1,38
2,76
2,76

8.09
8.09
8.09
18.34
10.72

0.54
0.54
0.54
0.24
0.24

0.18
0.18
0.18
0.33
0.22

0.69

0.09

1,38

4.92

0.48

0.12

MSE
Specific effect

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Matrix task
Between subjects
Group (G)
Within subjects
Session (S)
GS

1,38
2,76
2,76

14.42

153.72
83.02

48.61
2.40
2.40
Transfer effects

CWMS test, correct recall


Between subjects
Group (G)
Within subjects
Session (S)
GS
Forward Corsi task
Between subjects
Group (G)
Within subjects
Session (S)
GS
Backward Corsi task
Between subjects
Group (G)
Within subjects
Session (S)
GS
Pattern comparison test
Between subjects
Group (G)
Within subjects
Session (S)
GS
Stroop color interference index
Between subjects
Group (G)
Within subjects
Session (S)
GS
Cattell test
Between subjects
Group (G)
Within subjects
Session (S)
GS

1,38
1,38
1,38
2,76
2,76
1,38

46.98
46.98
46.98
61.42
49.17
3.87

2,76
2,76

23.97
8.50

0.37
0.37

0.39
0.18

2,76
2,76

5.04
0.47

0.23
0.23

0.12
0.01

1,38

15.13

0.53

0.29

1,38

1.12

0.18

0.03

2,76
2,76

14.43
10.56

0.32
0.32

0.28
0.22

2,76
2,76

3.28
0.12

0.27
0.27

0.08
0.003

416.85

0.07

1,38

1.43

38.05
38.05

0.46
0.19

2,76
2,76

1,38
2,76
2,76

2.99
32.67
8.88

24.68
5.99

0.04

21.26
21.26

0.39
0.24

0.11

0.03

0.21
0.03

0.63
0.03

1,38

1.99

0.34

0.05

1,38

2,76
2,76

0.18
0.08

0.01
0.05

0.05
0.02

2,76
2,76

1,38

0.73

0.02

1,38

0.09

10.23

0.003

2,76
2,76

7.00
1.64

0.16
0.04

2,76
2,76

12.94
0.47

0.66
0.66

0.25
0.01

25.1
2.45
2.45

1.28

201.79

64.40
0.96

Note. ANOVA analysis of variance; CWMS Categorization Working Memory Span task.

p .05. p .01. p .001.

shorter completion times, at the posttest and follow-up sessions


than at the pretest (MDiff. .6.13, p .001; and MDiff.
.64, p .001). No other effects were significant.
Stroop color task. The main effect of session was significant:
The interference index was higher at the posttest and follow-up
sessions than at the pretest session (MDiff. 0.14, p .001;
MDiff. 0.78, p .001, respectively); the difference between the
posttest and follow-up indexes were statistically significant
(MDiff. 0.07, p .001). No other effects were significant.
Cattell test. The main effect of session was significant: In both
groups, performance was better at the posttest than at the pretest

session (MDiff. 0.93, p .001), but at follow-up it no longer


differed from the pretest session (i.e., it was worse than at the posttest
session; MDiff. 0.48, p .01). No other effects were significant.
Comparing the influence of training-related benefits on the
standardized gains by age group. Standardized gains are presented in Figure 2.
Because significant benefits of the training were only seen for
both age groups in the matrix task (the criterion task) and the
CWMS task (see Table 5), standardized gain scores for short-term
and maintenance effects were only analyzed for these two tasks
(see Figure 2 and Table 6).

BORELLA ET AL.

722

Table 5
Summary of Short-Term Effects (Pre- vs. Posttest) and Maintenance Effects (Pretest vs. FollowUp) for Each Task for Trained YoungOld and OldOld Participants

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Trained young old

Trained old old

Variable

Short-term
effect

Maintenance
effect

Short-term
effect

Maintenance
effect

Matrix task
CWMS test
Forward Corsi span test
Backward Corsi span test
Pattern Comparison test
Stroop Color test
Cattell test

X
X

X
X
X
X
X

X
X
X
X
X

X
X
X
X
X

Note.

CWMS Categorization Working Memory Span task.

In the matrix task, the short-term and maintenance effects were


stronger for the trained groups than for the controls (MDiff.
1.75, p .001; MDiff. 1.23, p .001, respectively). In addition,
for maintenance effects the main effect of age group indicated
stronger effects for the old old than for the young old (MDiff.
0.34, p .05). The Age Group Training Group interaction was
not significant for either short-term or maintenance gains.
As concerns the CWMS, for both the short-term and the maintenance effects, the young old outperformed the old old (MDiff.
0.63, p .01; MDiff. 0.57, p .05, respectively), and the
trained groups gained more than the controls (MDiff. 1.73, p
.001; MDiff. 1.79, p .001, respectively). The Age Group
Training Group interaction was significant for the short-term and
maintenance effects, reflecting the greater gains for the trained
young old than for the trained old old, as shown by post hoc
comparisons, t(38) 3.64, p .001; t(38) 3.69, p .001,
respectively.
Comparing training-related benefits between the young old
in this study and the sample in Borella et al. (2010). Figure 3
shows Cohens (1988) d values (expressing the effect size of the
comparisons) obtained in the present study for the specific and
transfer effects detectable between the pretest and posttest sessions, as compared, at a qualitative level,6 with those reported for
verbal WM training in the Borella et al. study. The size of the
effect was large for the criterion task and nearly the same as after
verbal WM training. Similarly, the values of the nearest and near
transfer effects across the two studies were in the range of a large
effect size (see Cohen, 1988). The size of the effect was also much
the same, and again in the range of a large effect size, in the case
of the far transfer effect on processing speed. However, the two
studies differ in the short-term effects on the Stroop color test and
the Cattell test, in which there were no effects in the present study
(see Figure 3).
Concerning the maintenance effects, both studies showed a large
effect size for the criterion task (verbal and visuospatial WM task)
and for the processing speed measure.
The nearest transfer effect was maintained in this study, but this
was not the case in the Borella et al. (2010) study. Finally, large
maintenance effects were found for the reasoning task in Borella et
al., but not in the present study.

Discussion and Conclusion


This study examined the efficacy of visuospatial WM training in
the young old and old old. To our knowledge, it is the first to
assess the benefits of WM training in (a) young old and old old
participants given training as compared with active control groups
and (b) both transfer and maintenance effects.
Consistently with the literature on WM training, we found gains
in the criterion task (the task that was practiced) and in the nearest
transfer task, that is, the verbal WM task (Dahlin, Nyberg, Bckman, & Stigsdotter Neely, 2008; Li et al., 2008), in both young
old and old old trained participants. These benefits were maintained 8 months after the training. The results achieved in the
trained groups were higher than those seen in the active control
groups, so it is reasonable to assume that the benefits gained by the
trained participants were mostly attributable to the WM training.
Our results thus confirm that learning capacity, at least to a
certain extent, is preserved in aging and suggest that WM performance can be improved by even a very brief training (three
sessions), given that the specific and nearest transfer gains found
by Borella et al. (2010) were confirmed using the visuospatial WM
training considered here.
Concerning near transfer effects, only our trained young old
participants experienced an improvement in short-term memory
tasks (in the forward and backward Corsi tests) immediately after
receiving training, though this benefit was not maintained at the
8-month follow-up. A possible reason for this lies in that both
short-term memory tasks are sensitive to the use of strategies (e.g.,
Borella et al., 2010; Vranic, Spanic , Carretti, & Borella, in press),
so at the posttest stage, participants may have developed some
strategies to help them store the information and thus perform
better, whereas they returned to using automated skills afterwards
(Verhaeghen & Marcoen, 1996). Though we unfortunately have no
way to check this impression, because participants were not interviewed on their use of strategies, there are reports of strategies
being used in the two versions of the Corsi span task (Cornoldi &
Mammarella, 2008). These transfer effects were not seen in the
6
The comparison was drawn in qualitative terms because of the small
number of participants. For the same reason, the Cohens d values, as used
in Borella et al. (2010), were adjusted with the Hedges and Olkin (1985)
correction to avoid any small sample size bias.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

VISUOSPATIAL WORKING MEMORY TRAINING

723

Figure 2. Scores for short-term (pre- vs. posttest sessions) and maintenance (pretest vs. follow-up sessions)
gains in units of standard deviation for young old (A) and old old (B) trained participants and controls. Error
bars represent standard errors. CWMS categorization working memory span task.

trained old old participants, even at short term (e.g., Buschkuehl


et al., 2008), however, probably because of the more accentuated
decline in their cognitive resources, which also affects the spontaneous use of effective strategies (Verhaeghen & Marcoen, 1996),
and self-initiated encoding and retrieval operations.
Contrary to the results reported by Borella et al.s (2010) verbal
WM training study, no far transfer effects were apparent in our
participants of either age group, with the exception of a processing
speed measure showing that trained young old completed tasks
more quickly at the posttest stage, and this benefit was not maintained at the follow-up.

This happened despite the training program used in this study


being devised so as to comply with the key requirements for
effective training as suggested by Borella et al. (2010); that is, (a)
the proposed types of activity combined an adaptive procedure (the
task was made more difficult if participants were successful at a
given level; if not, only the lowest level was presented) with
variations in the corresponding maintenance requirements (to
avoid simple practice effects); (b) the training sessions were scheduled with a fixed interval between them that gave participants
sufficient time to consolidate the skills they acquired (see Carretti
et al., 2013). Given previous findings with verbal training, we did

BORELLA ET AL.

724

Table 6
Results of an ANOVA for Standardized Gains Calculated Following the Training (Pre- vs.
Posttest Sessions; Short-Term Gain) and at the 8-Month Follow-Up (Pretest vs. Follow-Up
Sessions; Maintenance Gain) for the Criterion Task (the Matrix Task) and Nearest Transfer
Effect Task (the CWMS Task), With Age Group (YoungOld vs. OldOld) and Training Group
(Trained vs. Control Groups) as the Between-Subjects Factors
Variable
Short-term gain

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Maintenance gain

Short-term gain
Maintenance gain

F(1, 76)

MSE

n2p

Matrix task
age group (AG)
training group (TG)
AG TG
age group (AG)
training group (TG)
AG TG

0.53
303.11
1.86
5.54
72.91
2.63

0.11
61.09
0.37
2.26
30.52
1.09

0
0.80
0.02
0.07
0.49
0.03

CWMS test
age group (AG)
training group (TG)
AG TG
age group (AG)
training group (TG)
AG TG

9.16
68.75
9.20
5.87
57.67
7.26

7.95
59.66
7.98
6.58
64.64
8.14

0.11
0.47
0.11
0.07
0.43
0.09

Group

Note. Training group: trained versus control. ANOVA analysis of variance; CWMS Categorization
Working Memory Span task.

p .05. p .01. p .001.

not expect these results, especially in the young old. The absence
of far transfer effects in the young olds performance in the
Stroop color and Cattell tests (which theoretically share processes
with WM) may mean that visuospatial WM training fostered an
improvement not in general flexibility, but only in WM tasks. This
impression is supported by the finding that the significant transfer
gains seen were not maintained.
The general lack of transfer effects in the old old is in line with
the findings of Borella et al. (2013), who also reported no gains in
terms of transfer effects after verbal WM training in the old old,
suggesting that they can benefit from this kind of training, but not
as much as the young old in Borella et al. (2010). At the same

time, it is also possible that similarly to young old, also for


old old the training may have elicited only specificWMand
not generalized processes. Moreover, the decline in plasticity
(Schmiedek et al., 2010) for this age group could have exhausted
these specific improvements found in the WM tasks only. It may
also be that, being less flexible, the old old improve more
gradually than the young old. It is worth emphasizing that the
duration of training is one of the aspects debated in this domain. In
addition, few studies to date have approached WM training gains
in young old and old old; future studies could examine this issue.
The results of the present study might suggest that the training
task modality (visuospatial rather than verbal) used here to train

Figure 3. Comparison between visuospatial (present study) and verbal (Borella et al., 2010 study) working
memory (WM) training gains, using Cohens d, for young old individuals. CWMS Categorization Working
Memory Span task.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

VISUOSPATIAL WORKING MEMORY TRAINING

participants influenced the efficacy of the training in terms of the


transfer effects. Before drawing any final conclusions, however, a
word of caution is needed regarding the potential differences
between the visuospatial and the verbal training task. Although the
WM tasks were shown to be similar in terms of their requirements
(see Carretti et al., 2012), adapting the visuospatial WM task for
the purpose of preparing training activities may have had a crucial
role. In fact, our training activities required that participants remember the positions of dots on a 4 4 matrix, together with a
secondary, perceptual task (pressing the space bar when the dots
occupied a gray cell). It may be that participants were less aware
of both their improved recall and their errors because dot positions
are less meaningful than words (e.g., Cornoldi et al., 2007), so they
were probably less engaged in activities that prevent recall errors,
such as the active control of irrelevant information. This contrasts
with what probably happens in verbal WM training, which involves having to process words according to their semantic category (identifying animal nouns): participants usually noticed any
improvement in their accuracy and the mistakes they made in
recalling animal nouns (in fact, some reported having difficulty
with blocking the animal noun that popped up in their mind). In
this sense, the processing demand of the verbal WM task (identifying animal nouns) led to a high interference in recall, and
involved engaging active inhibitory control mechanisms to prevent
intrusion errors. An analysis of intrusion errors (i.e., nonfinal
positions recalled erroneously in the WM task) would have given
us a better understanding of the mechanisms elicited by the training. Unfortunately, this measure (representing the ability to inhibit
no-longer relevant information) was not recorded in this study. It
is worth mentioning, however, that a further study on WM training
(Borella et al., 2013) in old old adults alone included this measure
and showed a significant decrease in intrusion errors in the verbal
criterion task at follow-up, suggesting that changes relating to
attentional control mechanisms of WM might have favored training gains. Future WM training studies should make an effort to
consider this variable to further clarify the mechanisms involved in
the training.
In addition, the type of activity (remembering dots) may have
been seen as too unfamiliar to older adults, and too unrelated to
their prior knowledge (Vecchi & Cornoldi, 1999). Although people are often exposed in everyday life to the need to retain and
recall verbal information for use in complex activities (e.g., reading comprehension), they are rarely asked to maintain and remember positions of abstract stimuli like dots. In this sense, motivation
for encoding and recalling meaningless material may have
played a part as well. We hypothesize that, because of the stimuli
used during the visuospatial WM task, participants do not always
see the task as novel and challengingfeatures that keep them
interested in the activities and stimulate their cognitive flexibility
(e.g., Carretti et al., 2013).
Thus, given the nature of the stimuli, it may be that the adaptive
nature of the task and the changes made to the tasks requirements
were not sufficient to favor participants awareness of an improvement in their performance, with the consequent motivation to do
better. Of course, these are mere speculations; no training studies
have analyzed the role of training task modality as yet.
At the same time, a greater impairment in visuospatial WM
tasks than in verbal tasks with aging (e.g., Myerson et al., 2003)
and a greater effort required in the identification process (Sharps &

725

Gollin, 1987) may mean that more specific processes are needed to
deal with the demands of visuospatial WM training tasks than in
the case of verbal training tasks, reducing the chances of any
transfer effect on tasks closely related to WM (such as the Cattell
test). If training programs can help the older adult brain to build
scaffolds in response to age-related changes (Park & ReuterLorenz, 2009), we suggest that such a compensatory scaffold can
only be built if the training focuses on abilities that are more
age-resilient, such as verbal skills (crystallized intelligence) (Baltes, 1997), which may indeed enable older adults to create the
foundations for supporting their flexibility in information processing, and thus compensate for their age-related decline. Alternatively, because visuospatial abilities are more sensitive to agerelated changes, it may be that the training schedule used here
would suffice when a verbal task is used (Borella et al., 2010), but
would need to be longer to achieve transfer effects when a visuospatial task is used. This interpretation would apply more to
young old, whose crystallized abilities have been shown to have
a compensatory role (e.g., De Beni, Borella, & Carretti, 2007), and
less to old old because their crystallized abilities start to decline.
These are mere speculations, though they highlight the importance
of clarifying the variables that might mediate the success of WM
training (possibly including the nature of the training task) with a
view to improving the quality of life of older adults, and especially
of the old old.
The present results would need to be replicated with a different
visuospatial WM training task (e.g., using pictures, which are more
meaningful, but retaining the present training procedure) to further
examine the role of task modality and training schedule. In future
studies, it will also be worth directly comparing WM training
programs that differ only in the nature of the tasks used, with a
view to better assessing the role of the task material in obtaining
and maintaining the benefits of training. Although the comparison
between the verbal and visuospatial training modalities is qualitative, it nonetheless suggests that a further factor (task material) is
potentially crucial in determining training gains. In this sense, an
effort should also be made to select training tasks and transfer
tasks that are more closely related to an older adults everyday life.
Finally, it would be of interest to administer the present training
program to less well-educated and also to less well-functioning
older adults, such as those with mild cognitive impairments, to
assess its efficacy and the generalizability of our findings, given
that the present sample consisted of well-educated and highfunctioning older adults who could have been more receptive to
the training. In the same vein, including a control group involved
in most effortful activities (see Brehmer et al., 2012) could confirm
the potential of the present training program.
A corollary aim of the present study was also to examine the
role of age by comparing short-term and maintenance gains between young old and old old, in tasks for which a gain after
training was found in both age groups (i.e., the criterion task and
the nearest transfer task). Our findings showed that aging did not
affect the benefits of training, in terms of standardized gains, when
the criterion task was considered: In the short term, both the
trained young old and the trained old old participants had greater
gains after receiving training than their active control counterparts.
The maintenance effect of training was also stronger for the
old old than for the young old, but this effect did not interact
with the group (trained vs. control), possibly meaning that old old

BORELLA ET AL.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

726

(both trained individuals and controls) developed task-specific


strategies or a greater familiarity with the task. When the trained
participants of each age group were compared in the task measuring nearest transfer effects, the trained young old showed a larger
improvement than trained old old in the short-term and in the
long-term effects, in terms of standardized gains. These results
suggest that even the very elderly can still improve and learn,
though the young olds cognitive skills are more flexible than the
old olds (e.g., Schmiedek et al., 2010), and cognitive plasticity
gradually declines with aging.
To conclude, the present findings show that both young old and
old old can benefit from WM training. They also suggest, however, that the efficacy of training visuospatial WM in inducing
flexibility is weaker the older the adult (using the training task
involved in this study at least). Identifying which factors do or do
not facilitate the beneficial effects of training is crucial to future
research aiming to develop successful training programs for sustaining active aging.

References
Baltes, P. B. (1997). On the incomplete architecture of human ontogeny:
Selection, optimization, and compensation as foundation of developmental theory. American Psychologist, 52, 366 380. doi:10.1037/0003066X.52.4.366
Baltes, P. B., & Smith, J. (2003). New frontiers in the future of aging: From
successful aging of the young old to the dilemmas of the fourth age.
Gerontology, 49, 123135. doi:10.1159/000067946
Barca, L., Burani, C., & Arduino, L. S. (2002). Word naming times and
psycholinguistic norms for Italian nouns. Behavior Research Methods,
Instruments, & Computers, 34, 424 434. doi:10.3758/BF03195471
Bopp, K. L., & Verhaeghen, P. (2007). Age-related differences in control
processes in verbal and visuospatial working memory: Storage, transformation, supervision, and coordination. Journal of Gerontology: Psychological Sciences, 62, P239 P246. doi:10.1093/geronb/62.5.P239
Borella, E., Carretti, B., Riboldi, F., & De Beni, R. (2010). Working
memory training in older adults: Evidence of transfer and maintenance
effects. Psychology and Aging, 25, 767778. doi:10.1037/a0020683
Borella, E., Carretti, B., Zanoni, G., Zavagnin, M., & De Beni, R. (2013).
Training-related working memory gains in old-age: An examination of
transfer and maintenance effects. Archives of Clinical Neuropsychology,
28, 331347. doi:10.1093/arclin/act020
Brehmer, Y., Westerberg, H., & Bckman, L. (2012). Working-memory
training in younger and older adults: Training gains, transfer, and maintenance. Frontiers in Human Neuroscience, 6(63).
Buschkuehl, M., Jaeggi, S. M., Hutchison, S., Perrig-Chiello, P., Dpp, C.,
Mller, M., . . . Perrig, W. J. (2008). Impact of working memory training
on memory performance in old old adults. Psychology and Aging, 23,
743753. doi:10.1037/a0014342
Carretti, B., Borella, E., Fostinelli, S., & Zavagnin, M. (2013). Benefits of
training working memory in amnestic mild cognitive impairment: Specific and transfer effects. International Psychogeriatrics, 25, 617 626.
doi:10.1017/S1041610212002177
Carretti, B., Borella, E., Zavagnin, M., & de Beni, R. (2013). Gains in
language comprehension relating to working memory training in healthy
older adults. International Journal of Geriatric Psychiatry, 28, 539 546.
doi:10.1002/gps.3859
Carretti, B., Mammarella, I. C., & Borella, E. (2012). Age differences in
proactive interference in verbal and visuo-spatial working memory.
Journal of Cognitive Psychology, 24, 243255. doi:10.1080/20445911
.2011.603695

Cattell, R. B., & Cattell, H. E. P. (1963). Measuring intelligence with the


Culture Fair Tests. Champaign, IL: Institute for Personality and Ability
Testing.
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006).
Distributed practice in verbal recall tasks: A review and quantitative
synthesis. Psychological Bulletin, 132, 354 380. doi:10.1037/00332909.132.3.354
Cohen, J. (1988). Statistical power analysis for the behavioral sciences.
Hillsdale, NJ: Lawrence Erlbaum.
Cornoldi, C., Bassani, C., Berto, R., & Mammarella, N. (2007). Aging and
the intrusion superiority effect in visuo-spatial working memory.
Aging, Neuropsychology, and Cognition, 14, 121. doi:10.1080/
138255890969311
Cornoldi, C., & Mammarella, I. (2008). A comparison of backward and
forward spatial spans. The Quarterly Journal of Experimental Psychology, 61, 674 682. doi:10.1080/17470210701774200
Corsi, P. M. (1972). Human memory and the medial temporal region of the
brain (Unpublished doctoral dissertation). McGill University, Montreal.
Crook, T., Bartus, R. T., Ferris, S. H., Whitehouse, P., Cohen, G. D., &
Gershon, S. (1986). Age-associated memory impairment: Proposed diagnostic criteria and measures of clinical changereport of a National
Institute of Mental Health work group. Developmental Neuropsychology, 2, 261276. doi:10.1080/87565648609540348
Dahlin, E., Nyberg, L., Bckman, L., & Stigsdotter Neely, A. (2008).
Plasticity of executive functioning in young and old adults: Immediate
training gains, transfer, and long-term maintenance. Psychology and
Aging, 23, 720 730.
De Beni, R., Borella, E., & Carretti, B. (2007). Reading comprehension in
aging: The role of working memory and metacomprehension. Aging,
Neuropsychology, and Cognition, 14, 189 212. doi:10.1080/
13825580500229213
De Beni, R., Borella, E., Carretti, B., Marigo, C., & Nava, L. A. (2008).
BAC. Portfolio per la valutazione del benessere e delle abilita` cognitive
nelleta` adulta e avanzata [The assesment of well-being and cognitive
abilities in adulthood and aging]. Firenze, Italy: Giunti OS.
Gallina, P., Saugo, M., Antoniazzi, M., Fortuna, P., Toffanin, R., Maggi,
S., & Benetollo, P. P. (2006). Validazione della scheda per la Valutazione Multidimensionale dellAnziano (SVAMA) [Validation of the
checklist for the assessment multidimensional of the elderly]. Tendenze
Nuove, 3, 229 264.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis.
Orlando, FL: Academic Press.
Jenkins, L., Myerson, J., Joerding, J. A., & Hale, S. (2000). Converging
evidence that visuospatial cognition is more age-sensitive than verbal
cognition. Psychology and Aging, 15, 157175. doi:10.1037/0882-7974
.15.1.157
Li, S. C., Schmiedek, F., Huxhold, O., Rocke, C., Smith, J., & Lindenberger, U. (2008). Working memory WM plasticity in old age: Transfer
and maintenance. Psychology and Aging, 23, 731742. doi:10.1037/
a0014343
Ludwig, C., Borella, E., Tettamanti, M., & de Ribaupierre, A. (2010).
Adult age differences in the Stroop color test: A comparison between an
item-by-item and a blocked version. Archives of Gerontology and Geriatrics, 51, 135142. doi:10.1016/j.archger.2009.09.040
Magaziner, J. (1989). Demographic and epidemiologic consideration for
developing preventive strategies in the elderly. Maryland Medical Journal, 38, 115120.
Melby-Lervg, M., & Hulme, C. (2013). Is working memory training
effective? A meta-analytic review. Developmental Psychology, 49, 270
291. doi:10.1037/a0028228
Miyake, A., Friedman, N. P., Rettinger, D. A., Shah, P., & Hegarty, M.
(2001). How are visuospatial working memory, executive functioning,
and spatial abilities related? A latent variable analysis. Journal of Ex-

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

VISUOSPATIAL WORKING MEMORY TRAINING


perimental Psychology: General, 130, 621 640. doi:10.1037/0096-3445
.130.4.621
Miyake, A., & Shah, P. (Eds.). (1999). Models of working memory:
Mechanisms of active maintenance and executive control. New York,
NY: Cambridge University Press. doi:10.1017/CBO9781139174909
Morrison, A., & Chein, J. (2011). Does working memory training work?
The promise and challenges of enhancing cognition by training working
memory. Psychonomic Bulletin & Review, 18, 46 60.
Myerson, J., Emery, L., White, D. A., & Hale, S. (2003). Effects of age,
domain, and processing demands on memory span: Evidence for differential decline. Aging, Neuropsychology, and Cognition, 10, 20 27.
doi:10.1076/anec.10.1.20.13454
Noack, H., Lvdn, M., Schmiedek, F., & Lindenberger, U. (2009). Cognitive plasticity in adulthood and old age: Gauging the generality of
cognitive intervention effects. Restorative Neurolology and Neuroscience, 27, 435 453.
Orsini, A., & Laicardi, C. (2003). Wais-R e terza eta` . La natura
dellintelligenza nellanziano: Continuita` e discontinuita` [WAIS-R and
the third age. The nature of intelligence in elderly people: Continuity and
discontinuity]. Firenze, Italy: Giunti OS.
Park, D. C., & Reuter-Lorenz, P. (2009). The adaptive brain: Aging and
neurocognitive scaffolding. Annual Review of Psychology, 60, 173196.
doi:10.1146/annurev.psych.59.103006.093656
Richmond, L. L., Morrison, A. B., Chein, J. M., & Olson, I. R. (2011).
Working memory training and transfer in older adults. Psychology and
Aging, 26, 813 822. doi:10.1037/a0023631
Salthouse, T. A., & Babcock, R. L. (1991). Decomposing adult age
differences in working memory. Developmental Psychology, 27, 763
776. doi:10.1037/0012-1649.27.5.763
Schmiedek, F., Lvdn, M., & Lindenberger, U. (2010). Hundred days of
cognitive training enhance broad abilities in adulthood: Findings from
the COGITO study. Frontiers in Aging Neuroscience, 2, 110.

727

Sharps, M. J., & Gollin, E. S. (1987). Memory for object location in young
and elderly adults. Journal of Gerontology, 42, 336 341. doi:10.1093/
geronj/42.3.336
Trenerry, M. R., Crosson, B., De Boe, J., & Lever, W. R. (1989). Stroop
Neuropsychological Screening Test. Lutz, FL: Psychological Assessment Resources.
Vecchi, T., & Cornoldi, C. (1999). Passive storage and active manipulation
in visuo-spatial working memory: Further evidence from the study of
age differences. European Journal of Cognitive Psychology, 11, 391
406. doi:10.1080/713752324
Verhaeghen, P., & Marcoen, A. (1996). On the mechanisms of plasticity in
young and older adults after instruction in the method of loci: Evidence
for an amplification model. Psychology and Aging, 11, 164 178. doi:
10.1037/0882-7974.11.1.164
Vranic, A., Spanic, A. M., Carretti, B., & Borella, E. (in press). The
efficacy of a multifactorial memory training in institutionalized older
adults. International Psychogeriatrics.
Verhaeghen, P., Marcoen, A., & Goossens, L. (1992). Improving memory
performance in the aged through mnemonic training: A meta-analytic
study. Psychology and Aging, 7, 242251. doi:10.1037/0882-7974.7.2
.242
Wechsler, D. (1981). Wechsler Adult Intelligence Scale-revised manual.
New York, NY: Psychological Corporation.
Zinke, K., Zeintl, M., Eschen, A., Herzog, C., & Kliegel, M. (2012).
Potentials and limits of plasticity induced by working memory training
in old-old age. Gerontology, 58, 79 87. doi:10.1159/000324240

Received September 2, 2012


Revision received April 15, 2013
Accepted April 18, 2013

You might also like