Content uploaded by Paola Palladino
Author content
All content in this area was uploaded by Paola Palladino on Oct 05, 2017
Content may be subject to copyright.
Decline in working memory updating through ageing:
Intrusion error analyses
Rossana De Beni
University of Padova, Italy
Paola Palladino
University of Pavia, Italy
The present research examines the decline in working memory updating through age. Two experiments
compared groups of participants in different age ranges (young-old, 55±65 years, old, 66±75 years and old-
old, more than 75 years and, in Experiment 2 only, young,20±30 years). Memory updating tasks were
administered, which required participants to remember the smallest items in each list. To perform the task
correctly, participants had to update information efficiently, reducing interference from items no longer
relevant. Intrusion errors were computed and in the first experiment these were described as ``intrusions
of irrelevant items'' (immediate exclusion) and ``intrusions of once relevant items'' (delayed exclusion).
The oldest adults performed worse in memory updating and made a greater number of intrusion errors of
once relevant information. In the second experiment results showed that increases in memory load
(number of items that had to be remembered) and updating demand (number of potentially relevant
items) impaired performance. The oldest adults had greater difficulty when the task demand was
increased. Furthermore, they produced a higher number of intrusion errors, particularly when the
updating demand was increased. It therefore appears that elderly people have specific difficulty in
updating information in working memory by excluding irrelevant information.
Numerous studies on age-related decline have
focused on cognitive functions, such as working
memory skill. Working memory is generally con-
ceived as a limited-capacity system for the tem-
porary storage and processing of information
(Baddeley, 1986; Baddeley & Hitch, 1974). Some
studies have been driven by the idea that working
memory skill is critical in explaining the decline in
cognitive performance of older people in com-
parison to young adults.
A well-known measure of working memory is
the Reading Span Test, RST, (Daneman & Car-
penter, 1980). It requires participants, after read-
ing (RST) or listening (LST) to series of
sentences, to recall the final word of each sentence
in sequential order. A typical version of this test
also requires the participant to verify the truth-
fulness of each sentence as a control of complete
sentence processing.
Since their introduction, working memory tests
have been shown to be related to the execution
of a range of cognitive tasks such as learning,
reasoning, and reading comprehension (e.g.,
Baddeley, 1986). A deficient working memory
performance has been found in groups with cog-
nitive deficits, such as poor comprehenders (for a
meta-analysis see Daneman & Merickle, 1996),
poor problem-solvers (Passolunghi, Cornoldi &
De Liberto, 1999), and older adults (see Ehrlich,
BreÂbion, & Tardieu, 1994; Gick, Craik, & Morris,
MEMORY, 2004, 12 (1), 75±89
#2004 Psychology Press Ltd
http://www.tandf.co.uk/journals/pp/09658211.html DOI:10.1080/09658210244000568
Correspondence should be addressed to Paola Palladino, Dipartimento di Psicologia, UniversitaÁdiPavia, p.zza Botta 6, 27100
Pavia, Italy. Email: paola.palladino@unipv.it
The present research was partially supported by grants to the first author. We thank Stefano Lo Presti for his help in data col-
lection.
1988; Light & Anderson, 1985; Pratt, Boyes,
Robin, & Manchester, 1989; Stine & Wingfield,
1987). Various hypotheses have been proposed to
explain these group-specific deficits, focusing on
the nature of the relation between working
memory and cognitive performance. Some
authors have suggested that older participants'
memory problems may be attributed to a deficit in
the processing component (e.g., sentence proces-
sing) (Craik & Jennings, 1992; Craik & Rabino-
witz, 1984), the storage component (e.g., the
demand to maintain information throughout the
task) (Babcock & Salthouse, 1990), or in the
control of resource allocation between storage
and processing (e.g., the dual task demand)
(Salthouse, 1990).
Hasher and Zacks (1988) proposed a theore-
tical framework to explain the poor performance
of older adults in working memory tasks and
consequently in cognitive tasks, such as reading
comprehension. They argued that good cognitive
functioning is related to a working memory
representation of the current task goal. They
suggested that inhibitory processes play an
important role in regulating the access of infor-
mation to the working memory. This occurs by
either blocking the activation of irrelevant infor-
mation (inhibition) or suppressing once relevant,
but currently irrelevant, information (suppres-
sion) while still allowing relevant information to
enter the working memory.
Hasher and Zacks' inhibition/suppression view
is supported by data that specifically show older
adults' difficulty in inhibiting distracting text
mixed with target text (i.e., poor blocking
mechanisms), (Connelly, Hasher, & Zacks, 1991),
in suppressing activated information that becomes
irrelevant (Hamm & Hasher, 1992; Hartmann &
Hasher, 1991), and in suppressing irrelevant
thoughts activated by experimental material at
both encoding and retrieval levels (Gerard, Zacks,
Hasher, & Radvansky, 1991; Radvansky, Zacks, &
Hasher, 1996).
Working memory is supposed to have limited
resources, so for it to function efficiently a con-
stant updating process is required. Resources are
usually postulated as the central, or generic,
mental capacity that must be allocated to the
various operations involved in processing, retain-
ing, and reporting information (Hasher & Zacks,
1979). Working memory updating is an extremely
frequent everyday memory process that allows
memory refreshing and learning. Updating con-
sists of a change in memory content and relevance,
reducing the activation of information that is no
longer significant. Updating is required, for
example, when a sudden change in weather causes
weekend holiday plans to changeÐupdating the
original schedule, destination, and list of things to
take. Morris and Jones (1990, p. 112) defined
memory updating as ``the act of modifying the
current status of a representation of schema in
memory to accommodate new input''. Various
studies on reading comprehension and working
memory demonstrate that mental representations
or mental models are updated during reading in
order to understand, memorise, or learn new
information (e.g., de Vega, 1995; Glenberg &
Langston, 1992; Morrow, Bower, & Greenspan,
1989). For example, de Vega (1995) observed that
a reader updates the spatial model according to
the last part of the text supplied, thus making
more accessible information coherent with the
current spatial model. Moreover, Robertson and
Gernsbacher (1996) found that when a new char-
acter is introduced, the reader updates reference
to the previous character and the capacity to recall
the previous character's name is therefore
reduced. Updating implies efficient suppression of
information that is no longer relevant in order to
devote memory resources to new information.
Although these tasks require updating, our
main goal was to examine working memory
updating ability with a basic task closely related to
the traditional working memory task (Daneman &
Carpenter, 1980; Siegel, 1994). The working
memory updating task should not simply focus on
remembering, but also on updating which is
explicit in the instructions (to actively select
relevant information after processing all poten-
tially candidate information). In the literature on
ageing, working memory updating tasks have not
been examined directly (but see Zacks, 1982),
whereas particular attention has been devoted to
memory or working memory processes. Working
memory tasks, such as RST, require suppression
of a large volume of information in order to
maintain and recall the last word only of each
sentence. If suppression is poor, then irrelevant
information will not be eliminated from the active
working memory, overloading it and creating
interference with the relevant information that is
competing for a limited amount of activation
(Hasher & Zacks, 1988; May, Hasher, & Kane,
1999). The number of intrusion errors has been
considered an index of poor suppression
(Chiappe, Hasher, & Siegel, 2000, De Beni, Pal-
ladino, Pazzaglia, & Cornoldi, 1998). In RST, for
76 DE BENI AND PALLADINO
example, intrusions are recalled words that were
embedded in the sentences studied, but were not
in the final position. Intrusion errors reflect the
presence of irrelevant information in the active
working memory. De Beni et al. (1998) devised a
working memory task that requires participants to
listen to series of word lists in order to detect
animal nouns and remember the last word of each
sentence. The authors found that poor compre-
henders had a lower working memory perfor-
mance and made a higher number of intrusion
errors than good comprehenders, particularly in
the case of more processed information and ani-
mal nouns. Palladino and De Beni (1999) found
that in performing LST, participants older than 75
(``old-old'') had a poorer recall and a higher
number of intrusions than participants aged
between 55 and 65 (``young-old''). These studies
did not investigate working memory updating
directly, although they analysed suppression
efficiency in working memory.
A direct measure of working memory updating
has been proposed by Morris and Jones (1990),
following a pattern already proposed (see Pollack,
Johnson, & Knaft, 1959). The test requires the
participant to listen to a sequence of letters
(varying in length from 4 to 10), and then to
remember only the last four elements. Since the
length of the series is unknown, each element is
initially held in memory and can only be updated
when its relevance has been clarified by the con-
tinuation of the presentation. Morris and Jones
(1990) consider this task to be a measure of the
central executive component of working memory,
which is a system that controls and coordinates
resource distribution between task demands
(Baddeley, 1990).
Following Hasher and Zacks' view (1988),
older people could be expected to have problems
in working memory updating tasks and to produce
a higher number of intrusion errors than younger
people. The aim of the present study was to
examine the working memory updating ability of
older adults and analyse their performance and
errors, qualitatively and quantitatively, comparing
participants of different age ranges. A further aim
was to investigate the role of memory resources
and suppression mechanisms in determining older
participants' working memory decline. The pre-
sent study aimed to verify that the deficit found in
older adults' working memory performance could
be due, at least in part, to inefficient inhibition and
suppression mechanisms. These operate to reduce
the activation of irrelevant information to avoid
interference and intrusion errors. Updating tasks
make it possible to study directly the age-related
decline in working memory updating ability and to
test the inhibition/suppression hypotheses
(Hasher & Zacks, 1988).
In the first experiment an updating task using
different materials (object and animal nouns and
digits) was devised. The participants were
required to listen to lists of items and at the end of
each list to remember only the items smallest in
size. When updating lists, items can appear
immediately irrelevant after processing (if they
are very big in size) and so they are immediately
excluded without being excessively activated. On
the other hand, items can seem relevant and then
later become either irrelevant (suppression
mechanisms are required) or maintain relevance
until the end of the list. In fact, in updating tasks it
is necessary to suppress activation of items
immediately classified as being irrelevant as well
as those no longer relevant. Items are processed in
order to evaluate their size and then either
excluded immediately or maintained for a while
(and probably rehearsed) and then eliminated.
Summarising, it is generally possible to identify
two different situations requiring suppression. In a
first situation the item is processed, compared,
and immediately excluded because the three
smaller items have already been presented. The
item immediately excluded should not reach a
high activation level, and therefore should not
require strong suppression. In a second situation
the item is maintained for a while because it is one
of the smallest ones and is then excluded when a
smaller item is presented. The items excluded
later are highly activated and, thus, require
stronger suppression.
Furthermore, in updating tasks the decision/
comparison task is not just a secondary task,
irrelevant to the final goal of remembering the
three smallest items. A size comparison process is
necessary (though not sufficient) to remember
only the correct items. In a working memory task,
such as the RST, processing efficiency does not
directly influence memory performance. The two
task components are separate and participants
could concentrate most of their resources on
remembering the last words and give only the
extra resources to the secondary task. By contrast,
participants in updating tasks should compare
each item and decide to keep or eliminate them in
order to remember only the smallest ones cor-
rectly. Consequently, in a working memory test it
is not possible to exclude the possibility that the
INTRUSION ERRORS IN AGEING 77
younger participants' better performance may be
due to strategic resource allocation that benefits
memory, drawing resources from sentence ver-
ification (Ehrlich et al, 1994; Turley-Ames, 1997).
Instead, in updating tasks this alternative inter-
pretation of the data can be tested. The relevance
of correct size comparisons to the final memory
demand makes the updating task performance less
sensitive to the participants' strategic ability to
devote limited resources to the processing task
component and concentrate resources on the
memory component.
The second experiment was aimed at extending
the results of the previous one, introducing a
younger group of participants. A further aim was to
examine the updating task in order to evaluate the
role of the memory resources required and the
quantity of potentially relevant information, and
therefore the suppression required. The material
from the previous updating tasks was combined in
order to produce lists with high/low maintenance
demands and high/low updating demands. To
produce these lists, filler items were introduced
and relevant items were reduced compared with
the lists in Experiment 1. The smaller number of
relevant items therefore induced us to consider
intrusions from the same list all together. However,
the higher number of lists allowed us to compute
intrusions from previous lists as well as inventions.
Intrusions from previous lists indicate that sensi-
tivity to proactive interference is related to LTM
inhibitory control. Inventions are the consequence
of difficulty in controlling information coming
spontaneously from semantic memory (LTM)
which had not been presented during the task.
In summary, the main goal of the present
research was to study working memory updating
in adults aged 55 to 80 and over. The interest in
this working memory task derives from the
assumption that working memory success is linked
to remembering pertinent information. Working
memory is efficient if information is continuously
updated on the basis of its relevance. Therefore, it
may be experimentally interesting to examine
whether and how working memory updating
declines with age. Furthermore, it is also clinically
important to examine this process not just by
comparing young and older adults, who differ in
cohort provenience (for a methodological over-
view see Fozard, Metter, & Brant, 1990), but also
by comparing older adults of contiguous age ran-
ges. In this way, it is possible to obtain indices of
decline in working memory updating within the
ageing process.
EXPERIMENT 1
Method
Participants.Asample of 60 old adults, 30
men and 30 women, agreed to participate in this
experiment. They were recruited from several
senior universities and recreational structures in
the north of Italy. Participants attended a short
interview which provided us with the information
necessary to exclude older people with serious
health problems or those on medications that
could induce drowsiness or influence cognitive
functioning. All participants were living indepen-
dently. In each of three age categories 20 partici-
pants were tested: young-old (range 55±65, X=58
years and 8 months), old (range 66±75, X=69
years and 8 months), and old-old (range 76±88, X
=77years and 4 months). The groups were mat-
ched for social and cultural levels; professional
qualification and/or years of education were
matched between groups (Baltes & Baltes, 1990).
Materials and procedure.Inthe updating task,
20 lists, each comprising 10 items, were presented
to participants. Ten lists were composed of nouns
(five lists contained familiar animal nouns and five
lists contained familiar object nouns). Ten lists
contained two-digit numbers. In order to check
potentially ambiguous items, we submitted the
lists of animal and object nouns to 30 under-
graduate students. They were required to compare
sizes and, for each list, to select the three smallest
items. An item was included only when it was
selected by more than 90% of the group. In each
list, the participants were asked to remember the
three smallest items, which were distributed ran-
domly in the list. Presentation order of the lists
was fixed: first the animal noun lists, then object
nouns, and finally number lists. Within each list
category (animal nouns, object nouns, or num-
bers) list presentation was randomised. An
example object noun list is as follows: orologio
(watch), furgone (van), borsetta (purse), dizio-
nario (dictionary), armadietto (cabinet), anello
(ring), telefono (telephone), cappotto (overcoat),
autobus (bus), occhiali (glasses).
Participants were tested individually. The sub-
ject items were presented aurally at a rate of 3
seconds per word. Participants started with a
warm-up trial and were asked to listen carefully
to the items and to recall freely, at the end of the
list, the three smallest items. To discourage gues-
sing, participants were not urged to remember all
78 DE BENI AND PALLADINO
target items; if unable to recall a word they were
to omit it.
Results
In scoring this test, the following different indices
were considered:
(1) The total number of items recalled; the total
number of critical items at the end of each
string recalled during the whole test.
(2) The total number of intrusions; the number
of items that were incorrectly recalled, as
they were not the three smallest ones.
Between intrusion errors, we distinguished:
(a) items that should have been immedi-
ately rejected because they were bigger
than the items already in memory; (b)
items that should have been maintained at
first and then rejected. Between them we
distinguished: (b1) items that should have
been maintained because they were the
first in the list, and then successively
excluded because of the presentation of
smaller items; (b2) items that should have
replaced bigger ones but should then have
been replaced by smaller items.
For example, considering the list presented
previously, the target items ``watch'', ``ring'', and
``glasses'' have to be remembered at the end of the
list presentation. ``Van'' has to be considered for a
short while within the pool of relevant items
because it is one of the first items presented, but it
has then to be excluded. If it is wrongly recalled it
is a b1 type intrusion. ``Purse'' and ``dictionary''
have to be considered for a longer time within the
pool of relevant items, but have then to be
excluded when smaller items are presented
(otherwise it is a b2 type intrusion). Instead
``cabinet'', ``phone'', ``overcoat'', and ``bus'' have
to be considered immediately as being inap-
propriate (type a). Items aare wrongly recalled if
they are allowed to reach too high an activation in
working memory. By contrast, b1 and b2 items,
which are potential targets, have to be maintained
for a while gaining activation, and then require
suppression to be eliminated from working
memory. Furthermore, b1 items are probably less
activated than b2 items; they are maintained
merely by reason of being the first, waiting to see
what will happen, whereas b2 items are picked up
to substitute bigger items and are believed to be
highly probable target items. Intrusion errors in
participants with poor suppression mechanisms
are expected to be more frequent for b2 than b1
items, and for b1 and b2 items than for aitems.
The missing answer index is the number of
missing items. It is calculated by subtracting the
total number of ``words recalled + inventions +
intrusions'' from the total number of ``words to be
recalled''. This index is useful for analysing par-
ticipants' attitudes regarding the omission of
unknown or uncertain answers and the production
of intrusion errors. In particular, in ruling out the
hypothesis that intrusions are produced as a con-
sequence of a completion strategy, the missing
answer index may related to intrusion index.
Data were collapsed across the type of material
(nouns and numbers) for easier presentation. No
differences were found between materials, and a
correlational analysis showed that memory indices
were strongly related despite the type of material
used, nouns or numbers (animal nouns vs object
nouns, r= .61, p<.001; animal nouns vs numbers, r
= .37, p<.01; object nouns vs numbers, r= .47, p<
.001).
Two analyses of variance with Age as a
between-participants factor with three levels
(young-old,old, and old-old) were conducted on
the number of recalled items and number of
intrusions (for mean scores see Table 1). Results
show a main effect of Age on the number of
recalled items, F(2, 57) = 29.29, p<.001, and
number of intrusions, F(2, 57) = 11.29, p<.001;
old-old participants recalled a lower number of
items and produced a higher number of intrusions
than old and young-old participants (post hoc with
Tukey's method, p<.01). Working memory
updating performance decreased with age: the
oldest adults recalled a lower number of target
items and showed difficulty in avoiding intrusion
errors.
TABLE 1
Mean scores for the different age groups in the indices of the
updating task
Updating Recalled items
max = 60
Intrusion errors
Young-old (55±65) X= 51.25 X= 6.20
SD = 4.12 SD = 2.78
Old (65±75) X= 47.80
SD = 1.96
X= 7.00
SD = 1.72
Old-old (> 75) X= 41.30
SD = 4.51
X=10.35
SD = 3.18
INTRUSION ERRORS IN AGEING 79
As the total numbers of possible intrusion
errors of type a,b1, and b2 were different, the
total percentage of each type of intrusions was
computed. A 3 (Age) 63 (Type of intrusion, as a
within-particular factor, a,b1,andb2) analysis of
variance was conducted on the percentages of
intrusion errors. Results showed a main effect of
Age, F(2, 57) = 8.84, p<.001, a main effect of the
factor Type of intrusion, F(2, 114) = 44.52, p<
.001, and an interaction effect between Age and
Type of intrusion, F(4, 114) = 2.98, p<.05. Post
hoc comparisons, conducted with Tukey's
method, showed that the oldest participants pro-
duced the highest number of b2 intrusion errors (p
<.05) and old participants produced a greater
number of b2 intrusion errors than young-old
participants (p<.05). No differences between age
groups were found for intrusions aand b1 (see
Table 2).
A correlational analysis between age and
memory updating indices was conducted to test
the hypothesis that intrusion errors are an index of
suppression problems and not just the con-
sequence of a completion strategy (when subjects
complete lack of memory with recently processed
but irrelevant information). A negative significant
correlation between age and recall (r=7.74, p<
.001), a positive significant correlation between
age and intrusions (r= .57, p<.001), and a sig-
nificant negative correlation between recall and
intrusions (r=7.84, p<.001) were obtained.
Furthermore, a positive correlation between
intrusions and missing answers was found (r= .55,
p<.001). Participants who made more intrusions
also had a poorer recall performance, omitting
more information. In order to determine the
degree of specificity of age effects on intrusions, a
correlation between age and intrusions was con-
ducted after partialling out missing answers (r=
.40 p<.01). Correlational analyses show that
intrusions are related to memory problems, and
both are related to age. A positive correlation of
intrusions with missing answers demonstrates that
participants are producing both intrusions and
missing answers rather than producing intrusions
in place of missing answers, in order to fill the lack
of memory. Age effects on intrusions seem to be
independent of age effects on missing answers.
DISCUSSION
In this first experiment older participants per-
formed a particular working memory task, that of
memory updating, an extremely frequent every-
day memory process. It permits refreshing of the
memory and learning, and can typically be
exemplified when a person moves to a new house
or town and must update address, phone number,
the way home, etc.
It has been hypothesised that memory decline
observed with ageing is, at least in part, due to a
decrease in the efficiency of inhibition and sup-
pression mechanisms (Hasher & Zacks, 1988).
The updating task is a working memory task that
requires subjects to refresh memory directly and
it, too, involves inhibition and suppression
mechanisms. Participants should process each
word presented in order to select and recall some
and eliminate the other information according to a
criterion. The updating task used in this experi-
ment required participants to remember three
items only, a memory load compatible with older
adults' span (see, for example, Craik, 1994; Pal-
ladino & De Beni, 1999). However, a further
processing of size comparison was associated to a
three-item memory load and an active exclusion
of irrelevant items was required. Comparison in
TABLE 2
Mean scores for the different age groups regarding types of intrusion in the updating task
Updating
Types of intrusions (%)
Intrusions a Intrusions b1 Intrusions b2
Young-old (55±65) X=0.77 X= 3.57 X= 6.50
SD = 1.20 SD = 3.04 SD = 8.13
Old (65±75) X=0.64 X= 5.00 X=10.5
SD = 1.13 SD = 2.43 SD = 6.86
Old-old (< 75) X= 1.80 X= 6.43 X= 16.00
SD = 2.21 SD = 4.44 SD =10.95
80 DE BENI AND PALLADINO
size was evaluated in order to be as evident as
possible, thus further demand on resources came
from the suppression and inhibition of irrelevant
items (Conway, Tuholski, Shisler & Engle, 1999).
Older adults showed a decline in the working
memory updating task and this decline was clearly
evident through the age groups considered. The
oldest adults performed worse than both old and
young-old adults. These results indicate that age-
related memory decline involves working memory
updating ability and is more marked at over 70
years of age.
An analysis of errors was carried out to inves-
tigate the role of suppression mechanisms in
working memory updating tasks. Intrusions are
considered as an indirect index of poor updating
ability and are supposed to be due to inefficient
suppression mechanisms (Chiappe et al., 2000; De
Beni et al., 1998). Results demonstrate an age-
related increase in intrusion errors. The oldest
adults produced more intrusions than younger old
adults and intrusions were produced more fre-
quent by the participants with worse recall per-
formances. However, the intrusions observed did
not appear to be just a tendency to avoid omis-
sions. First, the instructions discouraged guessing,
as is evident from the fact that participants, and
particularly the older ones, did not recall at the
items expected (the oldest adults recalled a mean
percentage of about 85, computed as targets plus
intrusions). Second, intrusions were directly rela-
ted to missing answers, showing that the more a
participant made intrusions, the more recall was
lacking.
In order to analyse the nature of intrusions
and underlying mechanisms further, intrusions
were distinguished on the basis of the initial acti-
vation potentially associated. Intrusions of
immediately excluded, thus less activated, items
were distinguished from intrusions of delayed
excluded, and thus more activated, items that
required stronger suppression. Results show that
even the oldest participants were able to compare
and exclude immediately irrelevant items. No
differences were found in the number of intru-
sions of immediately irrelevant items (less acti-
vated information) between age-groups. There
was no age-related increase in intrusions requir-
ing limited suppression. Furthermore, this result
allows us to exclude the possibility that older
adults have difficulty in size comparisons. They
were as accurate as younger participants in com-
paring size when the three smallest items were
already in the memory and the new item had to
be immediately excluded. Differences in number
of intrusions between groups were observed
when the task required the exclusion of no longer
relevant items. These items were maintained for
a while, gaining relevance and activation, and
were considered to have some probability of
being target items. The oldest adults had pro-
blems when they were required to update more
activated information either with phonologically
and semantically more distinguishable material
(object and animal nouns) or with numbers that
produced higher phonological and semantic
interference.
These results sustain the hypothesis that the
oldest adults have less efficient working memory
updating. This deficit seems to be due, at least in
part, to poor suppression mechanisms. However,
although it is possible to highlight the role of
suppression mechanisms in working memory
updating, recent research sustains the fact that
memory resources play an important role. A
smaller number of resources available would
reduce memory capacity, processing, and sup-
pression mechanism efficiency (Conway et al.,
1999). However, if working memory updating
performance is mainly due to the quantity of
resources available, increases in memory and
updating demands should impair performances in
a similar way.
A second experiment was conducted to further
investigate the decline of working memory
updating and to analyse memory and updating
demands separately. An updating task was
devised with new lists balanced between partici-
pants for maintenance and updating demands. It
was hypothesised that memory and updating
demands would produce a partially independent
effect on recall performance and on intrusion
errors in particular.
The increased number of lists/trials with similar
materials (animal and object nouns) also gave all
opportunity to look for proactive interference,
measured as the number of intrusions of items
from previous lists. However, the smaller number
of relevant items in each list, due to the intro-
duction of filler items, induced us to consider
intrusions from the same list all together. Three
different types of errors were computed: AB
intrusions (non-target items from the last list), C
intrusions (items presented in previous list), and
inventions, D (items that were not presented in
any list).
INTRUSION ERRORS IN AGEING 81
EXPERIMENT 2
In this experiment a memory updating task was
proposed in order to examine the effects of
memory and updating demands on accuracy and
errors, with groups differing in age, including a
control group of younger adults.
The updating task was similar to the task pro-
posed by Palladino, Cornoldi, De Beni, and Paz-
zaglia (2001) for poor and good comprehender
young adults. It comprised lists with high/low
updating demands and high/low memory
demands, balanced within participants. A high
updating demand means that the participant has
to extract target information from a larger pool of
relevant information. As a result, more informa-
tion must be activated and temporarily main-
tained in the memory and consequently it has to
be processed and excluded. A high memory
demand means that participants have to remem-
ber a higher number of target items; consequently
they have to hold in the memory and/or extract a
greater number of elements, with an increase in
the amount of activity required.
Material from the updating task proposed in
Experiment 1 was re-combined in Experiment 2
to produce lists with high/low maintenance
demands and high/low updating demands. To
compose these lists, filler items were introduced
and relevant items were reduced compared with
the lists in Experiment 1, as can be seen for the
low-updating lists (see also Table 3). Therefore,
the smaller number of relevant items, and rela-
ted potential intrusions, did not allow a distinc-
tion between types a,b1,andb2,asin
Experiment 1, but induced us to consider intru-
sions of the same list all together. However, the
higher number of lists in this new task (10 in
the first experiment vs 24 in the second), six
per category, allowed us to compute intrusions
from previous lists (due to the increased proac-
tive interference). Both intrusions of items
never presented and items from previous lists,
which should no longer concern working mem-
ory but rather long-term memory, were con-
sidered. In a study with older participants, May
et al. (1999) found that they were more affected
by proactive interference produced by items
from preceding lists. If intrusion errors gener-
ically involve all memory systems and, there-
fore, all kinds of information, the distinction
between items belonging to the same or preced-
ing lists should have no influence for older
adults but only for younger ones.
Method
Participants.Atotal of 80 participants were
selected from the population of a small town in the
north-east of Italy. The older adults attended a
senior university and all participants were living
independently. Selection followed the same
inclusion criteria as the previous experiment. In
each of four age groups 20 participants were tes-
ted: young (range 20±30 years, X=26.6), young-
TABLE 3
Examples of each type of list presented in the Experiment 2 updating task
High memory load
(5 target items to be
remembered)
High suppression
(5 relevant items to be
suppressed)
Low memory load
(3 target items to be
remembered)
High suppression
(5 relevant items to be
suppressed)
High memory load
(5 target items to be
remembered)
Low suppression
(2 relevant items to be
suppressed)
Low memory load
(3 target items to be
remembered)
Low suppression
(2 relevant items to be
suppressed)
pen bus nail shadow
relationship pencil suitcase key
wallet door cause action
cot period dictionary desk
ladder computer bookcase dice
glasses sock personality value
pain palace danger trunk
exercise book situation boot must
rubber loneliness gaze satisfaction
train hearing guitar light bulb
chair month series excuse
violin blackboard table prize
82 DE BENI AND PALLADINO
old (range 55±65 years, X= 59.2), old (range 66±75
years, X= 69.25), and old-old (range 76±85, X=
78.75). The groups were matched for social and
cultural levels. Professional qualification and/or
years of education were matched between groups
(Baltes & Baltes, 1990).
Materials and procedure.Atotal of 24 lists of
12 words were presented. Relevant information
comprised animal and object nouns distributed in
the different lists, balanced within-participants.
The nouns used were mostly the same as the
preceding experiment, or were chosen with simi-
lar characteristics. The 24 lists were divided into
four categories, according to the number of rele-
vant items (high and low suppression demands)
and the number of target or to-be-recalled items
(high and low memory demands in loading and
retrieval operations). In the low suppression lists
there were two relevant but non-target items,
whereas there were five in the high suppression
lists. In the case of low maintenance, participants
had to remember three items, and five items for
high maintenance. Abstract nouns were intro-
duced to maintain the list length of 12 nouns
invariable. Lists of each category were presented
in a block. The following is a list with high sup-
pression and memory demands: penna (pen),
relazione (relationship), portafoglio (wallet), culla
(cot), scala (ladder), occhiali (glasses), dolore
(pain), quadernone (exercise book), gomma
(rubber), treno (train), sedia (chair), violino (vio-
lin). Participants were required to remember the
five smallest items (pen, wallet, glasses, exercise
book, and rubber) and to exclude the five largest
items (cot, ladder, train, chair, and violin). An
example of each type of list is presented in Table
3. The order of the blocks of lists was balanced
between participants. There were four practice
trials for each category. Instruction, practice, and
item presentation modality were the same as in
Experiment 1.
Results
Participants' performance was analysed in terms
of both accuracy (number of correct items recal-
led) and type of error (number of intrusions,
number of items incorrectly recalled, but which
appeared among the words presented).
Accuracy. Participants' performance was
analysed in terms of percentage of items correctly
recalled, distinguishing four conditions: high
memory±high updating, high memory±low updat-
ing, low memory±high updating, low memory±low
updating. A 4 6262 ANOVA for a mixed
design was carried out with Age (young,young-
old,old,old-old)asbetween-subjects factor and
Maintenance (high vs low) and Updating (high vs
low) as within-subject factors. Significant main
effects of Age, F(3, 76) = 22.15, p<.001, Main-
tenance, F(1, 76) = 147.73, p<.001, and Updating,
F(1, 76) = 35.72, p<.001, were reported. A sig-
nificant interaction between Age and Main-
tenance, F(3, 76) = 3.72, p<.05, was found. No
other term in the analysis was significant. Post-hoc
comparisons (Tukey test) were carried out to
further explore the interaction between Age and
Maintenance. With high maintenance demand a
decline was observed between young and all three
groups of older participants (p<.05) and between
yound-old and both old and old-old (p<.05). With
low maintenance demand a decline was observed
between old-old and all other groups. Mean values
are presented in Table 4. Results show that lists
with high maintenance/updating demands were
more difficult than lists with low maintenance/
updating demands. The memory performance of
the various age groups was also influenced differ-
ently by the increase in maintenance demand. Low
maintenance lists demonstrated a memory decline
in the oldest participants only. On the other hand
high maintenance lists showed a memory decline
gradually involving all age groups considered.
TABLE 4
Mean (and standard deviations) of percentages of words recalled in the updating task in 4 conditions (Exp. 2)
High memory load Low memory load
High suppression Low suppression High suppression Low suppression
Young (20±30) 86.67 (6.31) 85.33 (7.12) 90.83 (7.49) 95.83 (4.37)
Young-old (55±65) 75 (13.66) 79 (10.49) 86.94 (9.75) 91.11 (7.73)
Old (65±75) 65.5 (10.10) 70.5 (9.87) 81.67 (14.54) 86.94 (10.71)
Old-old (> 75) 63 (8.16) 66 (8.89) 70.83 (15.39) 80 (15.02)
INTRUSION ERRORS IN AGEING 83
Errors.Asimilar ANOVA was carried out on
intrusion errors, i.e., non-target items recalled that
were present in the previous lists (not the smallest
items). Main effects of Age, F(3, 76) = 3.05, p<.05,
Maintenance, F(1, 76) = 98.88, p<.001, and
Updating, F(1, 76) = 33.90, p<.001, were
obtained. A significant interaction between Age
and Updating, F(3, 76) = 2.65, p<.05, was found.
No other term in the analysis was significant. Post-
hoc comparisons (Tukey test) were carried out to
explore the interaction between Age and Updat-
ing in greater detail. No differences were observed
between groups with low updating demand,
whereas significant differences were reported
between the group of younger and all three groups
of older participants with high updating demand.
Means are presented in Table 5. The ability to
avoid intrusions seems to be related to both
maintenance and inhibition demands. When both
the quantity of information to be remembered and
the number of items to eliminate from memory
increased, participants had difficulty in eliminat-
ing (suppressing) irrelevant information. How-
ever, the oldest participants had greater difficulty
when the updating demand increased. No similar
impairment was observed when memory load
increased.
To establish further that intrusion errors were
not produced as part of a completing strategy, a
distinct computation of intrusions from the first
and second part of the list was made. In fact, it
could be hypothesised that participants with
memory problems try to complete their poor
recall with the clearest memory traces, the last
words presented. It follows that intrusions would
be the last words heard by participants. If this
hypothesis is true, the number of intrusions from
the second part of the list should be significantly
higher than from the first part. A mixed design 4
62 ANOVA was carried out with Age as a
between-subjects factor and Part (first or second
part of the list) as within-subject factor. A main
effect of group was significant, F(3, 76) = 3.05,
p<.05. No other effect was significant, F<1. No
difference was observed between number of
intrusion errors from the first (M= 4.96 SD = 3.85)
or second (M= 4.84 SD = 3.08) parts of the list.
A relevant aim was to investigate further the
nature of intrusion errors in order to understand
underlying inhibitory mechanisms. This experi-
ment design required a large number of lists in
order to balance the four conditions. An increased
number of trials (participants listened to 24 lists of
nouns vs 10 lists in Experiment 1), with similar
material (object and animal nouns), provided the
opportunity to investigate interference, by com-
puting intrusion errors of items presented in pre-
vious lists; previous-intrusions. These type of
errors were compared with intrusions from the
same list, the three types (a,b1,andb2) described
in Experiment 1 were collapsed as same-intru-
sions,and a third type of error, inventions, i.e.,
intrusions of items not presented at all during the
task. In fact, intrusions from the same list were not
so frequent as in Experiment 1 because in the
present lists relevant items were reduced con-
siderably in number, particularly in lists with low
updating demand. Following this distinction we
have three types of error: same-intrusions (items
that were presented in the same list as the target
items), previous-intrusions (items presented in a
previous list), and inventions (items not presented
in any list).
Same-intrusions could be related to suppression
mechanisms that operate in working memory,
controlling and reducing activation of information
that has just been processed. Previous-intrusions
and inventions are related instead to the control of
LTM, either of irrelevant thoughts in the case of
inventions,orpreviously relevant information in
the case of previous-intrusions.
A463ANOVA for a mixed design was
carried out with Age (young,young-old,old,old-
old)asbetween-subjects factor and Type of errors
TABLE 5
Mean (and standard deviations) of within list intrusion errors in the updating task in 4 conditions (Exp. 2)
High memory load Low memory load
High suppression Low suppression High suppression Low suppression
Young (20±30) 2.45 (2.09) 2.35 (1.73) 1.2 (1.15) .55 (.83)
Young-old (55±65) 4.85 (3.13) 3.45 (1.93) 1.9 (1.65) .95 (.89)
Old (65±75) 3.95 (2.68) 2.95 (2.08) 1.65 (2.08) 1.3 (1.81)
Old-old (>75) 4.6 (2.48) 2.9 (1.89) 2.8 (2.35) 1.35 (1.73)
84 DE BENI AND PALLADINO
(same-intrusions,previous-intrusions,inventions)
as within-subject factor. Main effects of Age,
F(3, 76) = 23.90, p<.001, and Type of errors,
F(2, 152) = 30.86, p<.001, were reported. A sig-
nificant interaction between Age and Type of
error, F(6, 152) = 2.70, p<.05, was obtained. Post-
hoc comparisons (Tukey test) were carried out to
explore the interaction in greater detail. As Figure
1 shows, young and young-old participants pro-
duced a higher number of same-intrusions than
previous-intrusions and inventions (p<.05).
Within the group of old participants an increase in
previous-intrusion is reliable. However inventions
are still fewer than same/previous-intrusions.
Instead, for old-old participants there was no dif-
ference between types of intrusion, but a sig-
nificant increase in previous-intrusions and
inventions. Results show that intrusions increase
with age. However, whereas intrusions of infor-
mation just processed are also significant in the
young adult group, intrusions related to the con-
trol of LTM appeared later in ageing. In parti-
cular, intrusions of information never processed
during the task increased in the oldest period only,
over 75 years.
To investigate the hypothesis of a relation
between suppression of within-list candidates
(same-intrusions), suppression of between-list
candidates (previous-intrusions), and inhibition of
never-presented candidates (inventions), a corre-
lational analysis was conducted. Results show a
positive correlation between same-intrusions and
previous-intrusions (r= .28 p<.05), and a positive
correlation between previous-intrusions and
inventions (r= .39 p<.01).
Older adults appear to be more sensitive to
interference than young participants, showing less
efficient suppression mechanisms. However, when
age advances further, the impairment of suppres-
sion mechanisms becomes more serious and
appears to involve different levels of activation
control and suppression and different memory
systems, i.e., working memory and long-term
memory.
Discussion
The issues of whether an age-related decline in
working memory updating can be observed and
whether older adults' working memory updating
performance is impaired more by the memory
maintenance demand or the updating (suppres-
sion) demand, or both, were investigated. The
Figure 1. Same-intrusions (A), previous-intrusions (B), and inventions (C) in the four age groups. Error A = same-intrusions
(intrusions of items presented in the same list as target items); Error B = previous-intrusions (intrusions of items presented in a
previous list); Error C = inventions (items not presented in any list).
INTRUSION ERRORS IN AGEING 85
results of Experiment 2 confirm the results
obtained in Experiment 1, showing an age-related
decline in working memory updating. Both
maintenance and updating demands influenced
memory performance even if an increase in
maintenance demand produced greater impair-
ment in older adults' performance. With low
maintenance demand a decline can be observed
when comparing young participants with the three
older age groups. On the other hand, regarding
high maintenance demand, a decline can also be
observed between old age groups, in particular
between old and old-old participants. By contrast,
an increase in updating demand impaired all
groups equally. Furthermore, an analysis of
intrusion errors demonstrates an interaction
between groups and updating demands. Older
adults have specific difficulty in suppressing part
of the relevant information, when this part is
conspicuous in quantity. In fact they made a
similar number of intrusion errors with high and
low memory demands. On the contrary, with high
updating lists the oldest participants made a mean
number of intrusion errors that was considerably
higher (7.4 vs 4.25) than the number of intrusions
observed in the low updating lists. This effect was
specific to intrusions: an analysis of recall perfor-
mance shows that groups were similarly affected
by increased suppression demand (see Table 3).
These data suggest that older adults' difficulty
in the task is due to the memory demand. Low
memory demand impaired groups similarly but
high memory demand affected older participants
in particular. By contrast, the difficulty in avoiding
intrusion errors is not due to an increased memory
load or retrieval operations, but is related to the
quantity of relevant information that must be
suppressed or inhibited. An increase in number of
items to be selected impairs the groups similarly,
but an increase in items to be suppressed impairs
the oldest adults specifically. The ability to select
relevant information and maintain it in memory
and the ability to suppress irrelevant information
seem to be partially independent. These results
are coherent with those obtained by Palladino et
al. (2001) by comparing good and poor compre-
henders in a similar task. The effects of main-
tenance and suppression demands appear to be
partially independent.
The nature and meaning of intrusion errors
were also investigated. Results show that intru-
sions are not due to the strategy of filling the lack
of memory with more recently presented infor-
mation. Intrusions were words from the first and
second parts of each list equally. Three types of
intrusion were distinguished: information just
processed, information from a previous list, and
information not presented at all during the task.
One is information that has just entered in work-
ing memory while the other is already in the LTM.
An interaction was observed between group and
type of error. There was a substantial increase in
current list intrusions moving from the young to
young-old ages. The addition of an average of 10
years of age yields an increase in prior list intru-
sions. A further addition of 10 years (late 70s)
yields the addition of extra-list intrusions. The
oldest adults had a general difficulty in avoiding
all three types of intrusions considered, whereas
young and young-old subjects mostly produced
intrusions from the same list. Evidence of a rela-
tion between suppression of within-list candidates
and between-lists items in the oldest adults is
coherent with data obtained by May et al. (1999)
with older adults. The age-related increase in
number of intrusions appears to be due to differ-
ent causes, not just to inefficient suppression
mechanisms operating in WM but even to
mechanisms of blocking and controlling activation
operating on LTM. The age-related decline in
working memory updating appears to be related
to both a difficulty in maintaining a large amount
of information and a general difficulty in reducing
interference and avoiding intrusions. This was also
observed for words never presented, or presented
in previous lists.
GENERAL DISCUSSION
Memory updating is a particularly important
everyday memory ability which is necessary for
learning and accommodating new information to
old representation. As far as we know, no previous
researches have investigated working memory
updating of older participants. The present
research demonstrates that ageing produces a
decline in ability to update information in order to
keep in mind the currently more relevant
``schema'' (Morris & Jones, 1990). With increas-
ing age, working memory updating performances
decreased; older participants performed worse
than younger ones.
The nature of the task was both maintenance
and updating. Updating implies the ability to
maintain the more significant information in the
memory and to suppress irrelevant information.
To allow new information to enter the working
86 DE BENI AND PALLADINO
memory, old information should be reduced in
quantity and activation. Several authors have
argued that less efficient suppression/inhibition
mechanisms could be one of the reasons for poor
performance in working memory tasks (Chiappe
et al., 2000; De Beni et al., 1998; Gernsbacher,
1993; Hasher & Zacks, 1988). Participants who
were unable to suppress no-longer-relevant
information adequately or to control and inhibit
activation of irrelevant information, maintained it
in the working memory, which was then over-
loaded and became less efficient. If this is the case,
poor working memory performance should be
associated with the production of intrusion errors.
Irrelevant and no longer relevant information is
activated, maintained, and recalled as target items
(Chiappe et al., 2000; De Beni et al., 1998). In the
first experiment the main aim was to analyse types
of intrusion in order to demonstrate that they are
due to poor suppression (reduction of activation)
mechanisms. If the oldest participants have inef-
ficient suppression mechanisms, they should pro-
duce intrusion errors of items that were processed
and activated more and then more difficult to
suppress once irrelevant. With the second
experiment, the main aim was to investigate the
relevance of maintenance and suppression in
memory updating performance. A further aim was
also to study the spread of older participants'
suppression problems: whether older adults also
have difficulty reducing proactive interference
and blocking the activation of irrelevant thoughts
from the LTM, thus avoiding previously (but no
longer) relevant and never processed information
gaining activation.
The results show that older participants pro-
duced a higher number of intrusion errors than the
younger ones. These intrusions were of more
activated information, such as items first main-
tained and later suppressed in the updating task
(Experiment 1). Older participants had difficulty
when required to remove more processed infor-
mation from the active memory, suppressing its
activation. In Experiment 2 older adults' perfor-
mance was also impaired by an increase in main-
tenance demand but not in suppression demand
although, according to Conway et al. (1999), both
require resources. In Experiment 2, intrusions
from previous lists and inventions were also
computed. Data demonstrate that the oldest
adults produced a higher number of both types of
intrusions, compared to both young adults and
young old adults. The oldest participants showed a
general impairment of suppression mechanisms,
suggesting that age-related decline of working
memory updating is related to a general difficulty
in reducing interference and avoiding intrusions.
However, intrusion errors are an indirect index
of poor suppression/inhibition mechanisms, and
other hypotheses could be proposed to explain
such intrusion errors. A general lack of memory
resources, for example, could be a possible alter-
native explanation. However, agreement about
the fact that resource deficiency plays a role in
working memory updating performance of the
oldest adults, as also shown in Experiment 2, does
not exclude the possibility that suppression
mechanisms are involved. The intrusion effect
obtained in Experiment 1 seems to be due to high
memory strength information rather than a wea-
kened trace and loss of memory discriminability.
According to the findings of De Beni et al. (1998),
intrusion errors concern information that is pro-
cessed and remembered better. Furthermore,
another control excludes that intrusions could be
due to poor memory and forgetting (Light, 1991),
or to slow processing, according to the hypothesis
of Salthouse (1996). In the first case, we should
expect intrusions to be of information more fre-
quently located in the second part of the list, the
more recently processed and thus fresher and
more memorable. In the second case, we should
expect older participants to still be processing the
first information when the last is presented and
then to produce a higher number of intrusions
from the first part of the list. Neither of these two
tendencies was verified. Intrusions were produced
equally from the first and second parts of the list
(Experiment 2).
The hypothesis that older participants could
have developed a strategy of producing intrusions
in order to complete the missing recall was ruled
out. In fact, according to this hypothesis, older
adults' recall should have been close to the recall
expected (summing correct words and errors) and
missing answers should have been inversely related
to intrusions (more intrusions to reduce the num-
ber of missing answers). Instead, it was observed
that older adults' total recall did not approach the
total expected, in either the first or second experi-
ment, and missing answers were not inversely
related to intrusions. They were instead directly
related to each other and inversely related to recall.
It also seems possible to discount that the poor
working memory performance of the oldest par-
ticipants could be related to a lack of ability in
strategic resource allocation (Ehrlich et al., 1994;
Turley-Ames, 1997). The ability to efficiently
INTRUSION ERRORS IN AGEING 87
compare objects, animals, and numbers is neces-
sary for successful completion of updating tasks.
To perform well, the participant should carry out
both tasks adequately: comparison and exclusion
or maintenance. Older participants were able to
compare and immediately exclude irrelevant
items. No differences were found in the number of
intrusions of less processed information between
the age groups. Problems emerged when older
participants were required to exclude items no
longer relevant.
In conclusion, looking at what is generally
called old age, a fine-grained ageing effect of
working memory updating ability is observed. The
present study focused on the mechanisms involved
in memory updating decline observed through
ageing, found evidence of the role of both a
reduction in memory capacity and resources and
an impairment of suppression and inhibition
mechanisms. The updating performance of the
oldest participants decreases significantly with
higher memory demand. Furthermore, the oldest
participants seem more resistant to memory
updating. They tend to maintain the old, now
irrelevant information next to the new, and
naõÈvely let irrelevant gain activation.
Manuscript received 13 March 2002
Manuscript accepted 30 September 2002
PrEview proof published online 25 July 2003
REFERENCES
Babcock, R., & Salthouse, T. A. (1990). Effect of
increased processing demands on age differences in
working memory. Psychology and Aging,5, 421±429.
Baddeley, A. (1986). Working memory [Oxford Psy-
chology series No. 11]. Oxford: Clarendon Press.
Baddeley, A. D. (1990). Human memory: Theory and
practice. Hove, UK: Lawrence Erlbaum Associates
Ltd.
Baddeley, A., & Hitch (1974). Working memory. In G. H.
Bower (Ed.), The psychology of learning and moti-
vation (Vol. 8, pp. 47±89). New York: Academic Press.
Baltes, P. B., & Baltes, M. M. (1990). Successful aging:
Perspectives from the behavioral sciences. Cam-
bridge, MA: Cambridge University Press.
Chiappe, P., Hasher, L., & Siegel, L. S. (2000). Working
memory, inhibitory control and reading disability.
Memory & Cognition,28, 8±17.
Connelly, S. L., Hasher, L., & Zacks, R. T. (1991). Age
and reading: The impact of distraction. Psychology
and Aging,6, 533±541.
Conway, A. R. A., Tuholski, S. W., Shisler, R. J., &
Engle, R. W. (1999). The effect of memory load on
negative priming: An individual differences investi-
gation. Memory & Cognition,27,1042±1050.
Craik, F. I. M. (1994). Memory changes in normal aging.
Current Directions in Psychological Science,3, 155±
158.
Craik, F. I. M., & Jennings, J. M. (1992). Human
memory. In F. I. M. Craik & T. A. Salthouse
(Eds.), The handbook of aging and cognition (pp.
51±110). Hillsdale, NJ: Lawrence Erlbaum
Associates Inc.
Craik, F. I. M., & Rabinowitz, J. (1984). Age differences
in the acquisition and use of verbal information: A
tutorial review. In H. Bouma & D. G. Bouwhuis
(Eds.), Attention and performance,X(pp. 471±499).
Hove, UK: Lawrence Erlbaum Associates Ltd.
Daneman, M., & Carpenter, P. A. (1980). Individual
differences in working memory and reading. Journal
of Verbal Learning and Verbal Behavior,19,
450±466.
Daneman, M., & Merikle, P. M. (1996). Working
memory and language comprehension: A meta-
analysis. Psychonomic Bulletin & Review,3,
422±433.
De Beni, R., Palladino, P., Pazzaglia, F., & Cornoldi, C.
(1998). Increases in intrusion errors and working
memory deficit of poor comprehenders. The Quar-
terly Journal of Experimental Psychology,51A,
305±320.
de Vega, M. (1995). Backward updating of mental
models during continuous reading of narratives.
Journal of Experimental Psychology: Learning,
Memory & Cognition,21, 373±385.
Ehrlich, M. F., Bre bion, J., & Tardieu, H. (1994).
Working memory capacity and reading comprehen-
sion in young and older adults. Psychological
Research,56, 110±115.
Fozard, J. L., Metter, E. J., & Brant, L. J. (1990). Next
steps in describing aging and disease in longitudinal
studies. Journals of Gerontology,45, 116±127.
Gerard, L. D., Zacks, R. T., Hasher, L., & Radvansky,
G. A. (1991). Age deficits in retrieval: The fan effect.
Journal of Gerontology: Psychological Sciences,46,
131±136.
Gernsbacher, M. A. (1993). Less skilled readers have
less efficient suppression mechanisms. Psychological
Science,4(5), 294±298.
Gick, M., Craik, F. I. M., & Morris, R. (1988). Task
complexity and age differences in working memory.
Memory and Cognition,16, 353±361.
Glenberg, A. M., & Langston, W. E. (1992). Compre-
hension of illustrated text: Pictures help to build
mental models. Journal of Memory & Language,31,
129±151.
Hamm, V. P., & Hasher, L. (1992). Age and the avail-
ability of inferences. Psychology and Aging,7, 56±64.
Hartmann, M., & Hasher, L. (1991). Aging and sup-
pression: Memory for previously relevant informa-
tion. Psychology and Aging,6, 587±594.
Hasher, L., & Zacks, R. (1979). Automatic and effortful
processes in memory. Journal of Experimental Psy-
chology: General,108, 356±388.
Hasher, L., & Zacks, R. (1988). Working memory,
comprehension, and aging: A review and a new view.
In G. H. Bower (Ed.), The psychology of learning
and motivation (Vol. 22). San Diego, CA: Academic
Press.
88 DE BENI AND PALLADINO
Light, L. L. (1991). Memory and aging: Four hypotheses
in search of data. Annual Review of Psychology,42,
333±376.
Light, L. L., & Anderson, P. (1985). Working memory
capacity, age and memory for discourse. Journal of
Gerontology,40, 737±747.
May, C. P., Hasher, L., & Kane, M. J. (1999). The role of
interference in memory span. Memory & Cognition,
27, 759±767.
Morris, N., & Jones, D. M. (1990). Memory updating in
working memory: The role of central executive.
British Journal of Psychology,81, 111±121.
Morrow, D. G., Bower, G. H., & Greenspan, S. L.
(1989). Updating situation models during compre-
hension. Journal of Memory & Language,13,
441±469.
Palladino, P., Cornoldi, C., De Beni, R., & Pazzaglia, F.
(2001). Working memory and updating processes in
reading comprehension. Memory & Cognition,29,
344±354.
Palladino, P., & De Beni, R. (1999). Short term and
working memory in aging: Maintenance and sup-
pression. Aging. Clinical and Experimental Research,
11,301±306.
Passolunghi, M. C., Cornoldi, C., & De Liberto, S.
(1999). Working memory and intrusions of irrelevant
information in a group of specific poor problem
solvers. Memory & Cognition,27, 779±790.
Pollack, I., Johnson, L., & Knaft, P. (1959). Running
memory span. Journal of Experimental Psychology,
57, 137±146.
Pratt, M. W., Boyes, C., Robin, S., & Manchester, M.
(1989). Telling tales: Aging, working memory and
the narrative cohesion of story retelling. Develop-
mental Psychology,25, 628±635.
Radvansky, G. A., Zacks, R. T., & Hasher, L. (1996).
Fact retrieval in younger and older adults: The role
of mental models. Psychology and Aging,11,
258±271.
Robertson, R. R. W., & Gernsbacher, M. A. (1996).
Suppression during narrative comprehension. Paper
presented at the 68th Annual Meeting of the Mid-
western Psychological Association, Chicago, IL.
Salthouse, T. A. (1990). Working memory as a proces-
sing resource in cognitive aging. Developmental
Review,10,101±124.
Salthouse, T. A. (1996). The processing-speed theory of
adult age differences in cognition. Psychological
Review,103,403±428.
Siegel, L. S. (1994). Working memory and reading: A
life-span perspective. International Journal of Beha-
vioral Development,17,109±124.
Stine, E., & Wingfield, A. (1987). Process and strategy
in memory for speech among younger and older
adults. Psychology and Aging,2, 272±279.
Turley-Ames, K. J. (1997). Strategy use and the working
memory span test. Paper presented at the 38th
Annual Meeting of the Psychonomic Society, Phila-
delphia, PA.
Zacks, R. T. (1982). Encoding strategies used by young
and elderly adults in a keeping track task. Journal of
Gerontology,37,203±211.
INTRUSION ERRORS IN AGEING 89