See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223083076

Individual differences in memory updating in relation to arithmetic

problem solving

Article in Learning and Individual Differences · December 2004

DOI: 10.1016/j.lindif.2004.03.001

CITATIONS READS
113 345

2 authors:

Maria Chiara Passolunghi Francesca Pazzaglia

University of Trieste University of Padova
69 PUBLICATIONS 4,641 CITATIONS 127 PUBLICATIONS 4,335 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Maria Chiara Passolunghi on 20 April 2018.

The user has requested enhancement of the downloaded file.

 Learning and Individual Differences 14 (2004) 219 – 230
www.elsevier.com/locate/lindif

Individual differences in memory updating in relation to

arithmetic problem solving
Maria Chiara Passolunghi a,*, Francesca Pazzaglia b
a
Department of Psychology, University of Milano-Bicocca, P.zza dell’ Ateneo Nuovo, 1 Milan 20126, Italy
b
University of Padova, Italy
Received 1 August 2003; received in revised form 16 March 2004; accepted 16 March 2004

Abstract

The study investigates the relationship between memory updating and arithmetic word problem solving. Two
groups of 35 fourth graders with high and low memory-updating abilities were selected from a sample of 89
children on the basis of an updating task used by Palladino et al. [Memory & Cognition 29 (2002) 344]. The two
groups were required to solve a set of arithmetic word problems and to recall relevant information from another set
of problems. Several span tasks, a computation test, and the PMA verbal subtest were also administered. The group
with a high memory-updating ability performed better in problem solving, recalling text problems, and in the
computation test. The two groups did not differ in the PMA verbal subtest or in the digit and word spans. Results
were interpreted as supporting the importance of updating ability in problem solving and of the substantial
independence between memory updating and problem solving on one hand and verbal intelligence on the other.
D 2004 Elsevier Inc. All rights reserved.

Keywords: Memory updating; Arithmetic problem solving; Working memory

1. Cognitive processes in problem solving

Although a comprehensive theory of problem solving is still lacking, it is generally agreed that overall
problem solving processes can be analysed as two cooperating subprocesses: ‘‘understanding’’ and
‘‘searching.’’ The understanding process generates a problem’s internal representation, whereas the
search processes generate the solution. Sometimes the understanding process starts first, and then the
search process begins. However, often, the two processes alternate or even blend together (Chi, Glaser,
& Ress, 1982; Hayes & Simon, 1974). More recently, some fundamental cognitive processes underlying

* Corresponding author.
E-mail address: [email protected] (M.C. Passolunghi).

1041-6080/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.lindif.2004.03.001
220 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230

arithmetic word problem solving have been investigated and described within this theoretical frame-
work. A general finding is that the first important stage in solving arithmetic word problems consists of a
comprehension phase, which requires a complex text-processing activity (Mayer, 1992, 1998). First, an
analytical process occurs, in which a literal phrase-by-phrase translation into a memory representation
takes place for each sentence of the problem. Second, there is a synthetic process, in which the solver
tries to integrate information for each sentence into a familiar pattern or a ‘‘schema’’ of the problem. The
second main stage refers to searching for a solution. In this stage, the problem solver establishes a
solution plan based on knowledge of appropriate solving strategies and performs arithmetical calcu-
lations using his/her knowledge of arithmetical algorithms.
On closer examination of the comprehension phase, according to Hegarty, Mayer, & Monk (1995),
two strategies are possible: a ‘‘direct translation approach’’ and a ‘‘problem model approach’’. In the
direct translation approach, the comprehension phase consists in the problem solver’s decision as to
whether the statement currently being processed contains a key fact or not (e.g., a quantity of relevant
information or keywords such as ‘‘more’’ or ‘‘less’’). Hegarty et al. (1995) supposed that at this stage,
problem solvers using direct translation delete all information from the text base, except the numbers
and key words. Therefore, the outcome of this stage is a representation that contains less information
than the text base, which is a representation of only the keywords and numbers. On the other hand, in
the problem model approach, problem solvers attempt to construct or update their problem model from
the information in the text. Because a problem model approach is an object-centred representation, the
solver must determine whether the statement currently being processed refers to a new object or one
that is already represented in the problem model. The problem solver using the problem model
approach has a richer representation on which to base his/her solution plan. Not only does this
guarantee an accurate solution plan, but successful problem solvers may also keep the problem model
in working memory to monitor the plan of solution processes and to detect errors in a selected
procedure.

2. Arithmetic word problems and working memory

Arithmetic word problem solving involves several distinct cognitive processes: comprehending the
problem, constructing a representation of it, planning, and supervising single subgoals. In all these
cognitive processes, the role of working memory appears to be important. Several working memory
models have been described in the literature. According to Baddeley’s (1986, 1996) model, working
memory involves two slave systems, the articulatory loop and visuospatial sketch pad, which store
verbal and visuospatial materials, respectively. Their activity is coordinated by the central executive, a
supervisory system with attention functions. Other models, such as Daneman and Carpenter’s (1980),
consider working memory as a global function, in which a general resource maintains and processes
information.
Despite these differences, it is generally agreed that working memory is particularly important to
high-level cognition, such as reading comprehension and mathematics (Daneman & Carpenter, 1980;
Engle, Cantor, & Carullo, 1992; Hitch, Towse, & Hutton, 2001). Several studies have found that children
with specific arithmetic learning difficulties often perform arithmetic computations slowly and
inaccurately—an impairment that has been attributed to limitations in working memory (Hitch &
McAuley, 1991; Siegel & Linder, 1984; Siegel & Ryan, 1989; Swanson, 1993).
 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230 221

Among the typical functions of the central executive is the inhibition of irrelevant information
(Miyake & Shah, 1999) and memory updating (Baddeley, 1996). Several studies have shown that
inhibiting irrelevant information is related to good text comprehension and success in problem solving.
In both text comprehension and problem solving, it is necessary to process a great number of information
units, but some must be rejected to maintain only those which are relevant. In particular, integration of
the relevant information into a coherent structure allows a correct and complete mental representation of
a text or problem.
De Beni, Palladino, Pazzaglia, & Cornoldi (1998) found that poor comprehenders recalled fewer
target items and made more intrusion errors in a revised version of the Listening Span Test (LST;
Daneman & Carpenter, 1980). The LST requires participants to listen to lists of sentences, composed of
2–6 sentences, and to recall only the last word of each sentence in the right order. These authors
affirmed that poor comprehenders’ performance in the LST was impaired because they were unable to
inhibit irrelevant information adequately, with the consequence of overloading their working memory
system (see also Chiappe, Hasher, & Siegel, 2000). Similarly, research on arithmetic word problem
solving shows that poor problem solvers remember fewer target items and make more intrusion errors in
working memory tasks. Passolunghi and Siegel (2001) found that poor problem solvers’ performance
was impaired in both verbal and numerical working memory tasks and that poor problem solvers had a
deficit in their ability to reduce accessibility of nontarget and irrelevant information (see also
Passolunghi & Cornoldi, 2000; Passolunghi, Cornoldi, & De Liberto, 1999). Altogether, these results
demonstrate that working memory and inhibition are related to problem solving, as they are to reading
comprehension.

3. Relationship between updating ability and arithmetic word problems

Until now, the relationship between updating ability and problem solving has not been specifically
investigated. Updating is a complex activity that requires attributing different levels of activation to the
items presented and maintaining a restricted set of elements activated continuously. A typical measure of
updating ability is Morris and Jones’ (1990) Updating Task, which requires participants to listen to
several lists of letters of varying length (4 to 10). They are then asked to recall only the final four letters
of each list. Inasmuch as the length of each series was unknown, participants are forced to update the
information continuously to remember the final four letters only. Palladino, Cornoldi, De Beni and
Pazzaglia (2002) found a relationship between memory updating and reading comprehension using a
different memory updating test (Semantic Updating Task, SUT), based on that of Morris and Jones
(1990). Updating processes in this new task were more similar to those implied in reading comprehen-
sion. Participants had to listen to lists comprising 12 words and were asked to recall the words denoting
the 3 (or 5) smallest items in each list. Thus, differently from Morris and Jones’ task, a semantic criterion
was given for updating the relevant information. In four experiments, Palladino et al. (2002) compared
the performances in the SUT of poor and good comprehenders matched for age, gender, and logical
abilities. Poor comprehenders performed worse than good comprehenders did. They recalled a fewer
number of target words and made a higher number of intrusions, that is, words contained in the lists but
not among the 3 (or 5) smallest ones. These results were interpreted as supporting the relation between
memory updating and reading comprehension: readers need to construct a mental representation of the
contents of the text, integrating information from the text with previous knowledge (Gernsbacher, 1993).
222 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230

During the reading process, this sort of representation is enriched as soon as new information is
processed. However, when new information is not in continuity or is in contrast with old information,
readers had to update their mental model shifting to a new mental representation (Gernsbacher, 1993).
The mechanisms involved in this process are enhancement of new relevant information and inhibition of
information already processed but no longer relevant.
A similar process could also take place in comprehending arithmetic problems. Problem solvers have
to process all information derived from texts. Some information will be inhibited very early because it
is not relevant to the solution. Other information will be connected in a coherent model which will be
enriched successively by new information. This model will be complete when all the information
relevant to solving the question is integrated. Further information concerning other questions will then
be processed and structured in different models. Shifting from one model to another requires updating
information in working memory, in a fine modulation of the mechanisms of enhancement and
inhibition.
The first aim of the present research was to demonstrate that memory updating is involved in
resolving arithmetic word problems. Our hypotheses were that memory-updating ability may identify
children with good and poor word problem solution competence and computation ability, and that
the relationship between memory updating and math competence might not be due to intellectual
ability.
The second aim of the present research concerns the relation between updating ability and recall of the
relevant information regarding problems. We expected that children with a high updating ability would
also have good memory of text problems, which would foster representation of the mental model of the
problem and therefore individuation of the correct solution procedure.
We tested these hypotheses by comparing two groups of fourth-grade children with high and low
updating ability. The groups were formed on the basis of a screening of a larger group of fourth graders.
Children’s updating ability was examined by the SUT developed by Palladino et al. (2002). The high
updating ability group included children whose correct recall in the semantic updating memory test was
1 standard deviation above the mean score, whereas the low updating ability group had 1 standard
deviation less than the mean score. The two groups were required to solve two sets of arithmetic word
problems, which differed from each other for the quantity of irrelevant information. One set of problems,
classified as demanding high suppression, had five pieces of numerical information irrelevant to the
solution. The other set, classified as demanding low suppression, had three pieces of information
irrelevant to the solution. Our hypothesis was that the low memory-updating ability group may have a
lower problem solving ability, solving significantly fewer problems than the high memory-updating
group. Further, we hypothesised an effect of inhibition request, that is, that a higher quantity of to-be-
inhibited information embedded in the text problems would negatively affect building a mental model of
the problem and therefore the solution.
To test the second hypothesis, the same two groups of children with high and low updating abilities
were asked to recall the information relevant to the solution in a series of arithmetic word problems. We
expected that the group with low performance in the SUT would recall less relevant information than the
group with a high semantic updating ability.
The two groups of children were also tested for computation ability, memory span tasks (forward and
backward digit and word span task), and verbal intelligence. Our hypotheses were that the two groups
would not differ in verbal intelligence or in the working memory tests but that they would differ in the
arithmetic performance.
 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230 223

4. Method

4.1. Participants

Participants were 35 fourth-grade children, 21 with a high memory-updating ability (10 males and 11
females) and 14 with a low memory-updating ability (8 males and 6 females). These groups were formed
on the basis of a screening with the SUT (Palladino et al., 2002, described in detail in the Materials and
procedure section) administered to 89 children in a primary school in northeast Italy, during the fall
semester of the school year. The participants had no evidence of brain injury, sociocultural disadvantage,
or behavioural problems as documented by a review of student records. Children were included in the
high memory-updating ability group if their score in correct recall was at or above 1 standard deviation
from the mean of the entire sample in the SUT. The low memory-updating ability group included children
whose correct recall score was less than 1 standard deviation in the same updating memory test. The mean
scores in correct recall of the SUT were 45.29 (S.D. = 2.17) for the high memory-updating ability group
and 29.93 (S.D. = 2.76) for the low memory-updating ability group [t(33) = 18.39, P < .001]. Further
analyses concerning the number of intrusion errors show that the high memory-updating ability group
made fewer intrusion errors than the low memory-updating ability group. The mean scores of intrusion
errors in the SUT were 10.93 (S.D. = 5.53) for the good problem solver group and 4.48 (S.D. = 2.44) for
the poor problem solver group. The difference was significant [t(33) = 4.73, P < .001].
The two groups were tested individually in November and December in six sessions administered on
separate days, in the order described. In the first and second sessions, they had to solve a set of 12
arithmetic problems, and in the third and fourth sessions, they had to recall the relevant information of a
set of 12 other arithmetic word problems. In the fifth session, the children performed the computation
test and digit span tasks, while in the sixth session, they performed the IQ test (vocabulary subtest of
PMA battery, Thurstone & Thurstone, 1941). Each session lasted about 20 min.

4.2. Materials and procedure

4.2.1. Memory updating task

Palladino et al.’s (2002) SUT included 16 lists of 12 words, which were divided into four conditions:
high suppression/high load, high suppression/low load, low suppression/high load, and low suppression/
low load. Half the lists in each condition concerned animals, and the other half regarded objects (e.g.,
key, desk). High- and low-load conditions differed for the number of words to be recalled at the end of
each list: 5 and 3 of the smallest objects (or animals), respectively. High- and low-suppression conditions
differed for the number of words necessary to suppress: 5 and 2, respectively, of the concrete words
representing animals or objects larger than the target words. Below is a detailed description of each
condition.

1. High suppression/high load, with lists comprising 5 concrete words to recall, 5 concrete words to
inhibit, and 2 abstract words as fillers. Participants had to recall the 5 smaller items. Example: liberty,
panther, gorilla, owl, squirrel, mouse, whale, goose, buffalo, dog, strength, and lion.
2. High suppression/low load, with lists comprising 3 concrete words to recall, 5 concrete words to
inhibit, and 4 abstract words as fillers. Participants had to recall the three smaller items. Example:
pencil, bus, door, period, computer, sock, palace, situation, solitude, earring, month, and blackboard.
224 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230

3. Low suppression/high load, with lists comprising 5 concrete words to recall, 2 concrete words to
inhibit, and 5 abstract words as fillers. Participants had to recall the 5 smaller items. Example: hare,
tiger, game, type, mole, fly, fox, return, mosquito, idea, answer, and spider.
4. Low suppression/low load, with lists comprising 3 concrete words to recall, 2 concrete words to
inhibit, and 7 abstract words as fillers. Participants had to recall the three smaller items. Example: key,
shadow, action, desk, die, value, trunk, duty, satisfaction, lamp, excuse, and car.

Children were tested individually and performed all four conditions in a counterbalanced order. They
were instructed to listen to the words in the lists and to verbally recall the three (or five) smallest objects
(or animals) immediately after each list. Words were presented at the rate of one word per second.
Practice trials preceded the experimental task. For all conditions, responses were scored for the number
of correctly recalled words (independently of order) and for number of intrusions.

4.2.2. Arithmetic word problems to be solved

A set of 12 arithmetic word problems was developed. Each had 4 pieces of information relevant to the
solution and differed for the amount of irrelevant information. Half the problems contained 3, and half
contained 5 pieces of irrelevant information. Each problem could therefore belong to one of two
experimental conditions: low-suppression condition for problems with 3 irrelevant pieces of information
or high-suppression condition for problems with 5 pieces of irrelevant information. All problems
required two arithmetic operations. There were two problems with two multiplications, two problems
with a multiplication and a subtraction, and two problems with a multiplication and an addition for each
condition. Four teachers, asked to give independent judgements, agreed in considering the problems
appropriate for the children’s grade.
Each child resolved the problems individually on two consecutive days, six problems per day. During
each session, children were given a booklet with six problems and were invited to solve them. One of
four possible presentation orders was randomly assigned to each child. There were no time restrictions,
and one point was assigned for each operation correctly identified for arriving at the solution.
An example of a problem is shown in Fig. 1.

4.2.3. Arithmetic word problems to be recalled

A set of 12 arithmetic word problems was developed, identical for structure and requisites to the to-
be-solved problems, with the only procedural difference being that the experimenter required the child to
listen carefully to each problem to memorise relevant information. Instructions included a short example.
One of four possible presentation orders was assigned randomly to each child. Each problem was read by

Fig. 1. Example of problem to be solved in the low-suppression condition (three irrelevant information items).
 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230 225

the experimenter once at a standard rate, using normal intonation (about 35 s for problem). The child
then had to recall aloud all the relevant information. Recall was tape-recorded, and one point was
assigned for each relevant unit of information correctly recalled.

4.2.4. Span tasks

4.2.4.1. Digit span tasks
These tasks (forward and backward) were tested using the WISC-R Digit Span subtest (Wechsler,
1974). In the backward task, digits were recalled in the reverse order of presentation. Each task was
scored separately. The digits were presented at a 1-s-per-digit rate until a child made a mistake in both
trials of the same span length. A span was considered correct if all digits were recalled in the correct
order. The experimenter started with a span length of two, and, when repetition was correct, moved to a
higher level. When the repetition was incorrect, the child had the opportunity to try again with another
trial of the same span length. If the child failed on the second trial, the test was discontinued.

4.2.4.2. Word span task

The word span tasks used an increasing number of familiar two-syllable words, from two to eight
words per span length (two trials for each span length). The same procedure and scoring as the digit span
task (forward and backward) was used for the word span tasks.

4.2.5. Wide Range Achievement (WRAT3) test

This test is a standardised test (Jastack and Wilkinson, 1993). The arithmetic subtest consists of 40
written items measuring skills such as reading number symbols and performing written computations
(i.e., 34  21=, 724 597=, factor: r2 10r + 25). There are two forms, Blue and Tan, that can be used
with individuals aged from 5 to 74 years. Each item correctly solved counts as one point. In this
research, the Blue form was used in a single examination setting.

4.2.6. PMA test

The PMA test (Thurstone & Thurstone, 1941) is a standardised test. The vocabulary subtest consists
of 60 items, 30 of which are words. The child is asked to indicate the word which has the same meaning
as another. The other 30 items are pictures, where the subject is asked to indicate the picture
corresponding to the spoken word. The form for ages 9–11 years was used. Each item correctly solved
was scored as one point.

5. Results

5.1. Arithmetic word problem solutions

Table 1 (first and second row) presents the mean number of operations correctly identified for the
solution in the two groups of children.
A 2 (group)  2 (suppression) ANOVA revealed the following: a significant main effect for group
[ F(1,33) = 16.20, MSE = 23.86, P < .001] due to the higher number of operations correctly identified by
the high memory-updating ability group; and a significant main effect for suppression [ F(1,33) = 4.51,
MSE = 3.63, P < .05] due to the lower scores in the high-suppression condition. This result is consistent
226 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230

Table 1
Mean scores and standard deviations of correct procedures in arithmetic word problems with low and high memory updating
requests and recall of relevant information in arithmetic word problems in children with low and high memory-updating abilities
Variables Children with high updating ability Children with low updating ability
Mean S.D. Mean S.D.
Correct procedures in arithmetic word 8.57 3.41 2.71 3.81
problems with low inhibitory request
Correct procedures in arithmetic word 6.52 3.68 2.78 4.06
problems with high inhibitory request
Recall of relevant information in 28.47 8.59 16.85 7.62
arithmetic word problems

with our idea that inhibiting irrelevant information and updating relevant information are crucial to
problem solving.
Finally, a significant group by suppression interaction [ F(1,33) = 5.19, MSE = 3.63, P < .05] showed
that the previous result was limited to the high ability group. In fact, while the high memory-updating
ability group performed better in the low-suppression than in the high-suppression condition, no
differences between the two conditions emerged for the low memory-updating ability group, which
obtained a very poor performance overall with a higher dispersion of scores.

5.2. Arithmetic word problems recall

Table 1 (third row) presents the mean scores of correct recall for the two groups. A comparison of the
two groups’ recall shows that the group with high memory-updating ability had a higher recall than the
other group [t(1,33) = 4.09, P < .001].

5.3. Span tasks

The results of the span tasks (forward and backward) are presented in Table 2. There were no
differences between the memory updating groups in the measures of forward word span ( P=.46),
backward word spans ( P=.17), and forward digit span ( P=.12). The differences between the groups on
backward digit span were not statistically significant [t(1,33) = 2.90, P < .01] when a Bonferroni
approach was used by dividing the Type I error rate by the number of comparisons: .05/9=.006.

Table 2
Mean numbers and standard deviations of correct recall of span tasks in children with low and high memory-updating abilities
Variables Children with high updating ability Children with low updating ability
Mean S.D. Mean S.D.
Digit span forward 5.76 0.83 5.29 0.91
Digit span backward 4.00 0.71 3.29 0.73
Word span forward 4.29 0.46 4.14 0.66
Word span backward 3.57 0.98 3.00 1.41
 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230 227

5.4. Calculation ability (WRAT3 test) and verbal intelligence (PMA test)

The two groups with high and low memory-updating abilities showed a significant difference in
calculation ability. The average scores in the arithmetic subtest of WRAT3 test (Jastack & Wilkinson,
1993) were 42.74 (S.D. = 2.54) for the high memory-updating ability group and 37.46 (S.D. = 4.82) for
the low memory-updating ability group. The difference between the groups was significant [t(33) = 4.04,
P < .0001].
However, there was no significant difference between the two groups in the scores of the vocabulary
subtest of PMA battery (Thurstone & Thurstone, 1941). Mean scores in this test were 56 (S.D. = 6.78)
for the high memory-updating ability group and 52 (S.D. = 8.86) for the low memory-updating ability
group ( P=.15). Converted to verbal IQ, the mean scores were 122 (S.D. = 10.60) for the high memory-
updating ability group and 116 (S.D. = 13.78) for the low memory-updating ability group.

6. Discussion

The first aim of this study was to verify whether memory updating is related to arithmetic word
problem solution and computation. Two groups with high and low memory-updating abilities were
selected from a sample of fourth-grade children by means of a working memory updating test (Palladino
et al., 2002). The two groups had to recall the relevant information of a set of arithmetic word problems
and to solve another set of word problems, with high- and low-suppression request. They were also
tested for verbal intelligence, computation ability, and digit and word spans.
We expected that children with different memory-updating abilities would perform differently in
solving and recalling arithmetic problems and in computation, but, by contrast, they would have the
same level of verbal intelligence. Further, we intended to test if the quantity of irrelevant information in
the text of a problem affects its solution, that is, whether it results in a worse performance when the text
contains more irrelevant information.
Our results support the hypothesis of a relationship between memory updating and arithmetic problem
solving, and between memory updating and recalling problem texts. The high-performance group in the
SUT performed better in solving the arithmetic problems and recalling relevant information from the
text. These results were further reinforced by the outcome that the two groups did not differ, as expected,
in verbal intelligence (PMA vocabulary subtest) and span tasks. Our sample had an overall verbal
competence, as measured by the PMA, that was higher than average. This IQ score may be an
overestimation inasmuch as the PMA battery was normed in 1968 and a general cross-generational
increase in mean IQ score has since been reported (Flynn, 1987). However, in terms of the current study,
the good intellectual level of the sample strengthens our results on the specificity of the relationship
between problem solving and updating, which cannot be explained simply in terms of low intellectual
resources at the base of both poor performance in problem solving and mental ability tasks. We believe
that our findings can be generalized to children with average or close to average mental ability.
Differently, in the presence of an intellectual deficit, the relationship between updating and problem
solving is not so specific, due to the cognitive deficits which impair both updating and problem solving.
As a consequence and in accordance with previous research, we can affirm that specific difficulties in
problem solving are due to neither a general intelligence deficit nor an impairment in span capacity, but
rather to a difficulty in processing and updating information. However, several results from other studies
228 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230

show that children with mathematical disabilities or who are at risk of a mathematical disability, exhibit
impairment in memory span when the task requires recall of numerical information but not when the task
requires recall of verbal information (Geary, 1993; Geary, Hoard, & Hamson, 1999; Hitch & McAuley,
1991; Passolunghi & Siegel, 2001; Passolunghi et al., 1999).
Another important result is the relation between the number of irrelevant units of information
contained in the text of a problem and the difficulty of the solution. We would expect this result to be
more evident in the poor updating group because of their difficulty in inhibiting irrelevant information.
Conversely, the effect was obtained only in the high memory-updating ability group, while the
performance of the low memory-updating ability group was very low in both conditions. Children
with a low memory-updating ability have a memory overload even with a small amount of information,
which is relevant or irrelevant to the solution. This explains why their performance, close to a floor
effect, was not affected by suppression.
Taken together, these results support the hypothesis that memory updating is involved in problem
solving and that it is connected to the ability to inhibit irrelevant information. This is consistent with
theoretical models and empirical results from studies of reading comprehension (Gernsbacher, 1993;
Palladino et al., 2002; Passolunghi & Siegel, 2001). The relation between memory updating and problem
solving suggests that constructing and updating good mental representations of problems have a
beneficial effect on the solution. Indeed, problem solving does not simply involve passive maintenance
of information but requires controlling that information (Passolunghi et al., 1999; Turner & Engle,
1989). From the point of view of Baddeley’s (1986, 1996) three-component model, the central executive
could be more specifically and strongly involved than the articulatory loop in this process. Moreover, our
findings that children with low memory-updating ability have impaired performance in arithmetic
problem solving are in agreement with other experimental evidence that show that problem solving
requires components such as working memory and inhibition (Bull & Sherif, 2001; Nigg, 2000).
The two groups also differed in computation. To solve multidigit problems presented auditorily, such
as 34 + 48, children break the problem down into elementary stages, in which it is necessary to hold
partial information (e.g., partial results and the amount to be carried) until reaching the final result
(Fuerst & Hitch, 2000; Hitch, 1978; Logie, Gilhooly, & Wynn, 1994). This requires temporary storage of
information in working memory. In this research, the computation ability of primary school children was
assessed by a written calculation test. Nevertheless, working memory and updating processes were still
important, even if the arithmetic operations were presented in written format. Solving a written
computation may require less involvement of the temporary information storage. However, in both oral
and written multidigit computations, it is necessary to process a large body of information (e.g., the
numbers and type of operation) but hold only part of it (e.g., the partial results), which must be
continuously updated until the final result is reached.
We cannot exclude the possibility that children with low updating ability and who are poor in problem
solving are also poor in comprehension, inasmuch as it was found that updating processes are related to
comprehension (Palladino et al., 2002). However, previous experimental evidence has shown a
significant difference between good and poor problem solvers in working memory tasks even using
comprehension as a covariate or matching the group for reading ability (Passolunghi & Siegel, 2001;
Passolunghi et al., 1999). Moreover, Zentall (1990) found that intellectual level and reading compre-
hension were not predictive of math performance in children with attention deficits and hyperactivity
disorders. These results suggest that difficulties in problem solving cannot be entirely explained by a
deficit in reading comprehension.
 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230 229

It may be questioned whether difficulty in updating processes is due to a general factor such as the
quantity of cognitive resources available. Conway, Tuholski, Shisler, and Engle (1999), and Engle et al.
(1992) posit that success in remembering relevant and suppressing irrelevant information in working
memory tasks is related to the quantity of cognitive resources available to the working memory system.
Subjects with a low span do not have the attention and resource capacity necessary to inhibit irrelevant
information. This perspective is compatible with the view that the updating task requires complex
management of activation processes, continuously modulating the status of all items. The study of the
relation between problem solving and memory updating is still in its early stages, and further research is
clearly necessary to fully understand its mechanisms. However, we think that the results provided in the
present study might serve as an avenue for future work.

References

Baddeley, A. D. (1986). Working memory. Oxford: Clarendon Press.

Baddeley, A. D. (1996). Exploring the central executive. Quarterly Journal of Experimental Psychology, 49a, 5 – 28.
Bull, R., & Sherif, G. (2001). Executive functioning as a predictor of children’s mathematics ability: Inhibition, switching, and
working memory. Developmental Neuropsychology, 19, 273 – 293.
Chi, M. T. H., Glaser, R., & Ress, E. (1982). Problem solving ability. In R. J. Stemberg (Ed.), Human abilities. New York:
Freeman.
Chiappe, P., Hasher, L., & Siegel, L. S. (2000). Working memory, inhibitory control and reading disability. Memory &
Cognition, 28, 8 – 17.
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 investigation. Memory & Cognition, 27, 1042 – 1050.
Daneman, M., & Carpenter, P. A. (1980). Individual differences in working memory and reading. Journal of Verbal Learning
and Verbal Behavior, 19, 450 – 466.
De Beni, R., Palladino, P., Pazzaglia, F., & Cornoldi, C. (1998). Increases in intrusion errors and working memory deficit of
poor comprehenders. Quarterly Journal of Experimental Psychology, 51A, 305 – 320.
Engle, R. W., Cantor, J., & Carullo, J. J. (1992). Individual differences in working memory and comprehension: A test of four
hypotheses. Journal of Experimental Psychology. Learning, Memory, and Cognition, 18, 972 – 992.
Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ test really measure. Psychological Bulletin, 101, 171 – 191.
Fuerst, A. J., & Hitch, G. J. (2000). Separate roles for executive and phonological components of working memory in mental
arithmetic. Memory & Cognition, 28, 774 – 782.
Geary, D. C. (1993). Mathematical disabilities: Cognitive, neuropsychological, and genetic components. Psychological Bulle-
tin, 114, 345 – 362.
Geary, D. C., Hoard, M. K., & Hamson, C. O. (1999). Numerical and arithmetical cognition: Patterns of functions and deficits
in children at risk for mathematical disability. Journal of Experimental Child Psychology, 74, 213 – 239.
Gernsbacher, M. A. (1993). Less skilled readers have less efficient suppression mechanisms. Psychological Science, 4,
294 – 298.
Hayes, J. R., & Simon, H. A. (1974). Understanding written instruction. In L. W. Gregg (Ed.), Knowledge and cognition.
Hillsdale: Erlbaum.
Hegarty, M., Mayer, R. E., & Monk, C. A. (1995). Comprehension of arithmetic word problems: A comparison of successful
and unsuccessful problem solvers. Journal of Educational Psychology, 87, 18 – 32.
Hitch, G. J. (1978). The role of short-term working memory in mental arithmetic. Cognitive Psychology, 10, 302 – 323.
Hitch, G. J., & McAuley, E. (1991). Working memory in children with specific arithmetical learning difficulties. British Journal
of Psychology, 82, 375 – 386.
Hitch, G. J., Towse, J. N., & Hutton, U. (2001). What limits children’s working memory span? Theoretical accounts and
applications for scholastic development. Journal of Experimental Psychology. General, 130, 184 – 198.
Jastack, S., & Wilkinson, G. S. (1993). Wide range achievement test—revised. Wilmington, DE: Jastak Associates.
230 M.C. Passolunghi, F. Pazzaglia / Learning and Individual Differences 14 (2004) 219–230

Logie, R. H., Gilhooly, K. J., & Wynn, V. (1994). Counting on working memory in arithmetic problem solving. Memory &
Cognition, 22, 395 – 410.
Mayer, R. E. (1992). Teaching of thinking skills in the sciences and mathematics. In D. F. Halpern (Ed.), Enhancing thinking
skills in the sciences and mathematics ( pp. 95 – 115). Hillsdale, NJ: Lawrence Erlbaum Associates.
Mayer, R. E. (1998). Does the brain have a place in educational psychology? Educational Psychology Review, 10, 389 – 396.
Miyake, A., & Shah, P. (Eds.). (1999). Models of working memory: Mechanisms of active maintenance and executive control.
New York: Cambridge University Press.
Morris, N., & Jones, D. M. (1990). Memory updating in working memory: The role of central executive. British Journal of
Psychology, 81, 111 – 121.
Nigg, J. T. (2000). On inhibition/disinhibition in developmental psychopathology: Views from cognitive and personality
psychology and working inhibition taxonomy. Psychological Bulletin, 126, 220 – 246.
Palladino, P., Cornoldi, C., De Beni, R., & Pazzaglia, F. (2002). Working memory and updating processes in reading
comprehension. Memory & Cognition, 29, 344 – 354.
Passolunghi, M. C., & Cornoldi, C. (2000). Working memory and cognitive abilities in children with specific difficulties in
arithmetic word problem solving. Advances in Learning and Behavioral Disabilities, 14, 155 – 178.
Passolunghi, M. C., Cornoldi, C., & De Liberto, S. (1999). Working memory and inhibition of irrelevant information in poor
problem solvers. Memory & Cognition, 27, 779 – 790.
Passolunghi, M. C., & Siegel, L. S. (2001). Short term memory, working memory, and inhibitory control in children with
specific arithmetic learning disabilities. Journal of Experimental Child Psychology, 80, 44 – 57.
Siegel, L. S., & Linder, B. A. (1984). Short-term memory processes in children with reading and arithmetic learning disabilities.
Developmental Psychology, 20(2), 200 – 207.
Siegel, L. S., & Ryan, E. B. (1989). The development of working memory in normally achieving and subtypes of learning
disabled children. Child Development, 60, 973 – 980.
Swanson, H. L. (1993). Working memory in learning disability subgroups. Journal of Experimental Child Psychology, 56,
87 – 114.
Thurstone, L. L., & Thurstone, T. G. (1941). Factorial studies of intelligence. Psychometric Monographs, 2 (Italian translation
PMA-Batteria Primaria di Abilità, Firenze, Organizzazioni Speciali, 1968).
Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28,
127 – 154.
Wechsler, D. (1974). Manual for the Wechsler Intelligence Scale for children—Revised. New York: Psychological Corporation
(Scala d’intelligenza Wechsler per bambini, 1987, Firenze, O.S.).
Zentall, S. S. (1990). Fact-retrieval automatization and math problem solving by learning disabled, attention-disordered, and
normal adolescents. Journal of Educational Psychology, 82, 856 – 865.

View publication stats