You are on page 1of 10

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

net/publication/5464114

Does finger training increase numerical


performance?

ARTICLE in CORTEX · MAY 2008


Impact Factor: 5.13 · DOI: 10.1016/j.cortex.2007.08.020 · Source: PubMed

CITATIONS READS

66 116

2 AUTHORS:

Maria Gracia-Bafalluy Marie-Pascale Noel


University of Zaragoza Catholic University of Louvain
6 PUBLICATIONS 74 CITATIONS 66 PUBLICATIONS 1,664 CITATIONS

SEE PROFILE SEE PROFILE

Available from: Marie-Pascale Noel


Retrieved on: 26 September 2015
This article was published in an Elsevier journal. The attached copy
is furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:

http://www.elsevier.com/copyright
Author's personal copy

cortex 44 (2008) 368–375

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/cortex

Special issue: Original article

Does finger training increase young


children’s numerical performance?

Maria Gracia-Bafalluy* and Marie-Pascale Noël


Université Catholique de Louvain, Louvain-la-Neuve, Belgium

article info abstract

Article history: Butterworth (1999) suggested that fingers are important in representing numerosities. Fur-
Received 5 April 2007 thermore, scores on a finger gnosis test are a better predictor of numerical performance up
Accepted 22 August 2007 to 3 years later than intellectual measures (Marinthe et al., 2001; Noël, 2005). We hypothes-
Published online 8 January 2008 ised that training in finger differentiation would increase finger gnosis and might also im-
prove numerical performance. Accordingly, 47 first-grade children were selected and
Keywords: divided into 3 groups: children with poor finger gnosis who followed the finger-differenti-
Finger training ation training programme (G1), a control-intervention who were trained in story compre-
Numerical performance hension (G2), and a group with high finger gnosis scores who just continued with normal
Magnitude representation school lessons (G3). The finger training consisted of 2 weekly sessions of half an hour
each, for 8 weeks. Before the training period, children in G3 performed better in finger gno-
sis and enumeration than children in the two other groups. After the training period this
pattern remained for the children in G2 and G3, but the children in G1 were significantly
better than those in G2 at finger gnosis, representation of numerosities with fingers, and
quantification tasks; they also tended to be better at the processing of Arabic digits. These
results indicate that improving finger gnosis in young children is possible and that it can
provide a useful support to learning mathematics. Such an approach could be particularly
appropriate for children with a developmental Gerstmann syndrome. Theoretically, these
results are important because they suggest a functional link between finger gnosis and
number skills.
ª 2008 Elsevier Masson Srl. All rights reserved.

1. Introduction fundamental basis of numerical knowledge. Indeed, if a child


can count a set of elements but cannot show the correspond-
Fingers have been related to numerical processing for a long ing quantity with her/his fingers without counting, it is con-
time. In the course of development, fingers are used to repre- sidered that she/he does not have a good representation of
sent numerosities before symbolic representations such as quantities (Brissiaud, 2003). Fingers are a support when recit-
number words or Arabic digits are employed (Fuson, 1988; ing the numerical chain (see Sato and Lalain, 2008, this issue)
Rusconi et al., 2005), implying that finger use might be the and they might also help children to understand the cardinal

* Corresponding author. 10 place Cardinal Mercier, Faculté de Psychologie – Unité CODE, 1348 Louvain la Neuve, Belgium.
E-mail address: maria.gracia-bafalluy@psp.ucl.ac.be (M. Gracia-Bafalluy).
0010-9452/$ – see front matter ª 2008 Elsevier Masson Srl. All rights reserved.
doi:10.1016/j.cortex.2007.08.020
Author's personal copy

cortex 44 (2008) 368–375 369

meaning of number words (Butterworth, 1999; Brissiaud, abilities. As mentioned above, this association might just be
2003). Fingers are used to point at each object when counting due to the proximity of the brain areas subserving the pro-
a set, which is important in establishing the one-to-one cor- cessing of these representations. But it is also possible that
respondence principle (Gelman and Gallistel, 1978; Gallistel the relationship is functional. If children have poor finger
and Gelman, 1992). Fingers can help to keep track of the gnosis, they may not be able to use their fingers efficiently
counted items in mental calculations (Geary, 2005) which al- to count, calculate, or show numerosities and this might pre-
leviates the burden on working memory (Alibali and DiRusso, vent them from developing good numerical skills. Such
1999) and increases accuracy (Geary, 2005). Finally, as we a functional hypothesis cannot be properly tested through
have 10 fingers, using them to count and represent numbers correlative studies showing associations of symptoms in
probably also helps to develop an understanding of the 10- groups of brain-damaged patients, related developmental
base numerical system. More specifically, Butterworth (1999, trends in children, or common neural bases to the processes.
pp. 249–250) hypothesised that ‘‘without the ability to attach Rehabilitation is a much more powerful way of addressing
number representations to the neural representations of fin- this issue. If there is a functional link between finger gnosis
gers and hands in their normal locations, the numbers them- and number skills, then increasing finger gnosis should lead
selves will never have a normal representation in the brain’’. to improvement in number skills. Such a functional link
Based on this observation of the role of fingers in the course might only exist at some time-points in development, and if
of numerical learning, Fayol et al. (1998) and Marinthe et al. so it would probably occur in the earlier phases of numerical
(2001) have shown that a neuropsychological test at 5 years learning.
of age, including measures of finger gnosis (i.e., the ability to Accordingly, in this study, we address the question of a pos-
differentiate one’s own fingers when they are touched with- sible functional link between finger gnosis and number skills
out any visual clues), was a powerful predictor of numerical in young children entering primary school. First, we compared
ability up to 3 years later. Indeed it was even stronger than the numerical abilities of children with good or poor finger
a general intellectual measure. A similar finding was pub- gnosis. Second, we trained the children with poor finger gno-
lished by Noël (2005), showing that measures of finger gnosis sis in finger differentiation (a control group also received
predicted scores on numerical ability tests 15 months later, training). The impact of the training on both finger gnosis
but did not predict reading ability. and numerical performance was examined. If good finger gno-
This relationship between fingers and numbers has also sis contributes to the development of numerical skill then this
been reported in adults. Andres et al. (2004, see also Andres training might also have positive consequences on numerical
et al., 2008, this issue; Fischer, 2008, this issue) showed that skills. Some indirect evidence for such a link has been
opening or closing the finger grip is mediated by the per- reported recently by Gracia-Bafalluy et al. (submitted for pub-
ception of a single Arabic digit: grip closure gets faster as lication) who showed that children who play a musical instru-
the presented digit gets smaller, and grip opening gets ment such as the piano or guitar, which requires good finger
faster as the digit gets bigger. differentiation, perform better than children who do not
Research has also pointed out the anatomical proximity of play an instrument, not only on finger gnosis tests but also
the brain areas subserving the processing of numbers and fin- on numerical tasks.
gers (Dehaene et al., 2003; Zago and Pesenti, 2002). Both the
precentral gyrus and the parietal cortex are involved in finger
and number representations. For instance, Simon et al. (2002) 2. Method
have shown that both specific and common regions in the
intraparietal sulcus, which are irrigated by the angular gyrus 2.1. Participants
artery, are activated during calculation and manual tasks
(i.e., pointing and grasping). This could explain the frequent The purpose of this project was to evaluate finger gnosis and
co-occurrence of deficits in number processing and finger numerical abilities in the early phases of learning mathe-
representation. In particular, Gerstmann (1927, 1940) reported matics, so we chose to study children entering primary
that a lesion of the parietal >cortex of the dominant hemi- schooling. First graders (112) from three different normal
sphere led to four symptoms: finger agnosia, agraphia, right- Belgian schools were screened for finger gnosis. From this
left disorientation, and acalculia. This association of four sample, we selected the 33 children with the lowest scores
symptoms is called the Gerstmann syndrome. Its existence and the 14 children with the highest scores; all of them
has been questioned (Benton, 1961, 1977), but there are still were native French speakers and parental consent for par-
reports of adults showing such a syndrome (Strub and ticipating in the study was given. The group of 33 children
Geschwind, 1974; Mayer et al., 1999), and of children exhibit- with poor scores was randomly divided into two subgroups:
ing a developmental Gerstmann syndrome (Grigsby et al., the action group (G1, mean age 6.4 years, SD ¼ 3 months;
1987; Suresh and Sebastian, 2000). range 5.8–6.8 years old; nine girls, total N ¼ 16 children) re-
In summary, research from developmental psychology, ceived training in finger differentiation, while the control
brain imaging, and neuropsychology have all shown evi- group (G2, mean age 6.4 years, SD ¼ 5 months; range
dences of a special link between numbers and fingers. In par- 5.9–7.9 years old; nine girls, total N ¼ 17 children) received
ticular, the use of fingers in childhood seems to be really training in story comprehension. The group with good finger
important in numerical development. gnosis scores (G3, mean age 6.4 years, SD ¼ 4 months; range
Studies with children and with brain-damaged adults have 5.8–6.9 years old; seven girls, total N ¼ 14 children) did not
reported an association between finger gnosis and numerical receive any training. This control group allowed us to
Author's personal copy

370 cortex 44 (2008) 368–375

measure the development of numerical performance in comparison of the magnitude of Arabic digits, ordinality
a group with good finger gnosis. judgements, quantification tasks (subitizing and enumera-
tion), and additions.
2.2. Testing material In the ‘‘how many fingers’’ task (similar to Noël, 2005;
DiLuca et al., 2006), photographs of hands were presented,
The tests were administered individually to each of the 47 par- showing the palms as a ‘‘subjective’’ perspective on hands
ticipants. It included measures of finger gnosis and number would do, on a computer screen placed in front of the child
processing. (50 cm distance). The child was asked to state the number
of fingers raised as fast as possible. Pictures include canoni-
2.2.1. Finger gnosis assessment cal and non-canonical finger shapes. The canonical finger
Finger gnosis was measured by the task used by Noël (2005). shapes are those a Belgian child would usually use when
The child was asked to place her/his dominant hand, palm counting on his/her fingers: raising the right thumb for
down on a table with the fingers spread out. A white sheet one, the right thumb and the index finger for two and con-
of paper was placed between the child and her/his hand, so tinuing with the natural order of fingers for larger numbers.
that the hand could not be seen. Stimulation consisted of For quantities above five, all the fingers of the right hand are
a light press on the nail. First, isolated touches were per- raised and the counting continues using the fingers of the
formed: the examiner pressed the nail of a finger, and the left hand in the same manner. Ten canonical finger shapes
sheet was removed. The child was then asked to point with and nine non-canonical ones (i.e., any other configuration)
the index finger of the non-dominant hand to the nail she/ were each shown twice each, making a total of 38 items.
he thought had been touched. Each finger was touched twice, One point was given for each correct answer. Six items
giving a total of 10 touches, randomly organised. Then, five were presented before the task, to help the child understand
simultaneous touches of two fingers were administered. Fi- the instructions.
nally, five trials consisted of touching two fingers successively The verbal numerical chain was assessed for both its ex-
and the child had to point, in the correct order, to the two fin- tension and elaboration. For the extension, the child was
gers she/he thought had been touched (a correct response cor- asked to count as far as possible and was stopped if and
responds to the correct identification of the two fingers when she/he arrived at 100. The score was the largest number
touched in the correct order). All correct trials were scored the child reached, allowing one local error (i.e., the omission of
as 1, giving a maximum of 20. one number word or the inversion of two consecutive number
Two drawing tasks were presented to the child: the ‘‘draw words). The level of elaboration of the verbal numerical chain
a man’’ test (Goodenough and Pasquasy, 1967) and the ‘‘draw was tested with three tasks: (1) counting on from a number
a hand’’ task. We reasoned that if children’s finger gnosis im- (two items: from 7 and from 8); (2) counting with a lower
proves, this might be reflected in drawings of the hand (e.g., and upper limit (three items: from 8 to 14, 9 to 16 and 7 to
in reproducing the correct number of fingers) but should 15); and (3) counting ‘‘x’’ steps from a starting point (four
not change their ability to draw a whole body. The ‘‘draw items: ‘‘starting at 3, you must count 2 steps’’; from 8, 2 steps;
a man’’ test was scored on 35 criteria: 17 concerned the pres- from 9, 3 steps; and from 7, 8 steps).
ence of diverse elements (head, legs, arms, etc.), six concerned The processing of Arabic digits was evaluated in two tasks
the relations between the elements (e.g., arms and legs at- differing only in the instruction: a magnitude comparison task
tached to the body), six concerned the proportions of different and an ordinality judgement task. In both tasks, pairs of Ara-
parts of the body (e.g., the parts of the face), and six were bic digits between 1 and 9 differing by a small (1) or a large nu-
a qualitative score for the whole drawing. One point was merical distance (3 or 4) (e.g., 6 and 7, or 5 and 8) were
awarded for each correct item, with 0 if the item was absent presented on a computer screen facing the child (at about
or unrecognisable. 50 cm). The pairs appeared after a fixation point (for
In the ‘‘draw a hand’’ task, the child’s non-dominant hand 500 msec) and remained on the screen until a response was
was placed under the table so as not to be visible, and the child given by the child. They were followed by a blank screen
was asked to draw a hand, but was not allowed to put the pen (500 msec). In the magnitude comparison task, children were
down or to use his/her drawing hand as a model. The scoring asked to press a key on the side of the larger digit. In the ordi-
was based on seven criteria: the presence of five fingers, an op- nality judgement, they had to press the key on the side of the
position between the thumb and the other fingers, a differen- digit that comes first when counting. Turconi et al. (2006)
tiation between the palm and the fingers, the fingers being showed that this simple change in instruction guided the par-
attached to the palm, the fingers being the correct relative ticipant to process two different dimensions of the number: its
sizes, the palm and the fingers being in proportion, and the cardinal and its ordinal value. Response times (RTs) and errors
quality of the drawing (e.g., a two-dimensional shape for were encoded.
each finger, not just a simple line). For each of these criteria, Two quantification tasks were administered. In the subi-
a score of 2 points was given if the criterion was met, 1 point tizing task, 25 presentations of sets of 2–6 items (five sets
if the criteria was partially but not fully met, and 0 point if it for each number of items) were presented in a random order
was not met at all. The maximum score is thus 14. on a computer screen for 300 msec, after a fixation point
(500 msec), and followed by a blank screen (500 msec). The
2.2.2. Numerical skills assessment child was asked to say how many items were present. In
The numerical testing included assessments of the represen- the counting task, a set of items was randomly distributed
tation of numerosities by fingers, the verbal numerical chain, on a sheet of paper. There were two sets of 12 and 16 items
Author's personal copy

cortex 44 (2008) 368–375 371

and one set of 14 and 18 items. The child was asked to count 2.4. Intervention material
and say how many items there were (the maximum score
being 6). Then, to assess the non-pertinence of order princi- 2.4.1. G1: finger-differentiation intervention
ple (i.e., the fact that the numerosity is independent of the The training aimed at helping children differentiate their fin-
order in which the items are counted, as defined by Gelman gers, in terms of both sensitivity and motility. To that end,
and Gallistel, 1978), the child was presented with a 12-item children learned an association between each finger and a col-
set to count. When she/he had finished, the examiner our. Each session started by putting a coloured sticker on each
showed her/him a different pointing sequence than the one finger nail of every child in the group, keeping the same col-
already used – starting with a different element and follow- ours for all the activities: thumb-white, index-green, middle-
ing a different order – and asked ‘‘if I count like this, how blue, ring-yellow, and little-red. The activities were designed
many items will there be?’’ The child’s answer was consid- so that the number of fingers involved was progressively in-
ered correct if he/she said ‘‘12’’ without counting the items. creased; thus, the level of difficulty of the exercises increased
This question was asked a second time with yet another progressively throughout the intervention programme. We in-
pointing sequence. cluded a fixed number of trials for each level. All the activities
Finally, 19 single-digit additions were presented with Ara- were first performed with the dominant hand and then with
bic digits on paper cards: 10 with a sum less than or equal to the non-dominant one. Four different games were used.
10, and nine with a sum between 11 and 17. There were no The labyrinth game consisted of coloured interlinked path-
ties and no additions with 0 or 1 as an operand. The accuracy ways, including from 2 to 5 different colours. A total of 20 lab-
of the responses was recorded. yrinths were proposed, including four exercises for each
number of fingers involved. The children’s task was to follow
2.3. Testing procedure each colour pathway with the corresponding finger, so that
their attention was concentrated on the relevant finger for
The tasks were administered individually before starting the some seconds. Fig. 1a gives an example of a three-colour laby-
intervention programme (pre-training testing), and 2 weeks rinth (although in the graphic they appear as different grey
after the end of that intervention (post-training testing), i.e., levels). So, children had to follow the green pathway first
about 2 months later. with the index finger of the dominant hand and then with

Fig. 1 – (a) Labyrinth game. (b) Pointing game. (c) Piano game, natural sequence. (d) Piano game, random stimuli.
Author's personal copy

372 cortex 44 (2008) 368–375

the index finger of the non-dominant hand. Then, they had to testing and a technical problem prevented the data for one
follow the yellow pathway with the dominant ring finger, fol- child on two computerised tasks being recorded.1
lowed by the non-dominant one. And finally, the blue pathway
was followed by the dominant and then the non-dominant
3.1. Pre-training assessment
middle finger.
The pointing game consisted of coloured circles of about
Results from the pre-training baseline were analysed by
2  2 cm aligned in sets of 4–5 rows and 5–8 columns, with
means of one-way analyses of variances (ANOVAs) with group
a progressive difficulty level from 2 to 5 colours. The child-
(G1: finger training, G2: story comprehension training, and G3:
ren’s task was to point to each circle, with the finger having
control group) as the between-subjects factor. These analyses
the corresponding colour. For instance, in Fig. 1b, four colours
had two aims: (1) to check that the two intervention groups, G1
are involved: white (which corresponded to the thumb), red
and G2, were comparable before the intervention; and (2), to
(little finger), yellow (ring finger) and blue (middle finger).
examine whether the children with poor finger gnosis had
Children had to start by pointing to the first (white) circle
lower numerical skills than those with good finger gnosis.
with the thumb, and then to the second (red) circle with the
The ANOVAs were followed by post hoc comparisons between
little finger, continuing with the ring finger for the yellow
G1 and G2, and between G1–G2 together and G3.
circle and the middle finger for the blue circle and so on.
The ANOVA on the finger gnosis score was significant
The whole exercise was then repeated with the fingers of
(F(2,43) ¼ 34.53; p < .0001). The post hoc t-test confirmed that
the non-dominant hand.
G1 and G2 performed similarly (t(30) < 1), and that both these
In the piano game, children were shown a sequence of col-
groups performed less well than G3 (t(44) ¼ 8.39; p < .0001;
oured rectangles, arranged as if they were piano keys. First, pia-
see Table 1 for the means of all the groups).
nos with colours following the natural consecutive sequence of
The three groups performed similarly both on drawing
the 10 fingers were used (the thumb first, then the index, mid-
a man (F(2,44) < 1) and drawing a hand (F(2,44) < 1).
dle, ring and little finger), so that the children could understand
The ANOVA on the ‘‘how many fingers?’’ score was mar-
and develop the skill of identifying each square and following
ginally significant (F(2,42) ¼ 2.94; p ¼ .064). Post hoc compari-
the order of the squares (see Fig. 1c). Later the task included
sons indicated that G1 scored slightly better than G2
from 2 to 5 colours randomly distributed, and an increasing
(t(29) ¼ 1.87; p ¼ .071), and both groups were somewhat more
quantity of stimuli (see Fig. 1d). The child’s task was to follow
accurate than G3 (t(43) ¼ 1.75; p ¼ .086). Median reaction times
the colour sequence with the corresponding finger.
were computed for each participant for pictures showing
Finally, children were trained in performing a grip, with the
small (1–5) and large (6–10) numbers of fingers in either a ca-
thumb contacting each of the other fingers in turn. This task
nonical or a non-canonical hand shape. A repeated-measures
aimed to work each possible combination, closing the grip
ANOVA was calculated with group as a between-subjects fac-
as strongly as possible, so that children had to concentrate
tor, and size (small or large) and type of hand shape as
not only on the two fingers involved but also on the strength
within-subjects factors. This analysis showed highly signifi-
of the contact, for about 5 sec for each grip closure. This exer-
cant main effects of size (F(1,40) ¼ 253.53; p < .0001) and of
cise was performed with both hands at the same time.
hand shape (F(1,40) ¼ 77.71; p < .0001) as well as a significant

2.4.2. G2: story-comprehension intervention 1


During the data trimming, we removed, from the pre-training
The second group, G2, were trained in story comprehension. A measures, the data of one child in G1 for the drawing task, two
researcher read a story to the children and then asked them children of G1 and of G2 for the counting task and one child of
questions about its content. Finally they were asked to pro- G1 for the magnitude comparison task. For the post-training mea-
duce possible endings to the story. sures, there were two outliers in G1 on the finger gnosis task; one
in G3 on the draw a man test, two in each of G1 and G2 on the
‘‘how many fingers?’’ task, two in G2 and one in G3 on the simple
2.5. Intervention procedure
counting task, one in G1 and two in G3 on counting with limits,
one in G1 and one in G3 on magnitude comparison, two in G1
Both interventions were run in small groups of about five chil- on subitizing and two in G1 on additions. There were also some
dren, and conducted by the same student researcher, who had children whose data could not be recorded because they were ab-
also participated in the baseline evaluation of the children. sent at the time of testing. For the pre-training testing there were
The second evaluation was undertaken by a different student, two absences in G1 for finger gnosis, two in G2 and two in G3 for
who recorded the data without knowing the children. the counting tasks, two in G1, three in G2 and one in G3 for mag-
nitude comparison and three in G1, three in G2 and one in G3 for
There were two half-hourly intervention sessions each
the ordinality judgement. In the post-training evaluation there
week for 8 weeks. Both types of training (finger intervention was one absence in G1 for the finger gnosis, two in G1 and one
and story comprehension) were delivered in a dedicated in G2 for the ‘‘how many fingers?’’ task, two in G1, two in G2
room in each school, during class time. and one in G3 for the magnitude comparison, and two in G1,
one in G2 and one in G3 for the ordinality judgement. Finally,
there were some data which could not be encoded in the
pre-training assessment because there were technical problems
3. Results
with their registration in the computer. This affected two partic-
ipants in G2 for the finger gnosis task and the ‘‘how many fin-
All the analyses were run after excluding results that differed gers?’’ task, and two in G1 on the magnitude comparison. The
from the mean of the group by more than two standard devi- resulting empty cells account for the different degrees in freedom
ations. Some children were absent for some parts of the in Section 3.
Author's personal copy

cortex 44 (2008) 368–375 373

Table 1 – Means and standard deviations for the three groups before and after training
Data before training Data after training Maximum
possible score
G1 G2 G3 G1 G2 G3

Total finger gnosis 13.36 (2.09) 13.12 (1.86) 18.29 (1.13) 16.57 (1.01) 13.06 (2.94) 14.79 (2.25) 20
Draw a hand score 9.00 (2.44) 8.94 (2.16) 8.29 (3.81) 10.81 (1.87) 9.35 (2.09) 9.46 (2.14) 14
Draw a man score 15.33 (4.83) 16.17 (2.60) 17.00 (5.58) 16.81 (3.67) 19.41 (3.77) 18.57 (4.29) 35
‘‘How many fingers?’’ RTs 3426 (957) 3615 (1089) 3176 (1068) 1974 (299) 2374 (430) 2342 (738) –
‘‘How many fingers?’’ score 36.50 (1.96) 34.77 (2.97) 34.07 (3.47) 36.62 (1.12) 34.87 (3.18) 35.57 (1.95) 38
Counting score 34.64 (11.86) 40.36 (11.77) 39.23 (12.47) 63.31 (22.67) 68.40 (19.71) 71.23 (21.70) 100
Counting with limits score 5.25 (2.43) 4.88 (2.05) 6.50 (1.22) 6.47 (1.35) 6.53 (1.58) 5.57 (1.91) 9
Magnitude comparison score 45.00 (2.03) 43.08 (2.69) 41.85 (5.39) 35.21 (11.59) 31.80 (10.42) 32.23 (10.99) 48
Magnitude comparison RTs 1810 (379) 1830 (266) 1713 (475) 1251 (236) 1351 (313) 1271 (260) –
Judgement of ordinality score 43.23 (5.93) 42.50 (5.03) 42.83 (3.90) 45.50 (2.17) 43.36 (3.67) 45.31 (2.39) 48
Judgement of ordinality RTs 2121 (648) 2283 (507) 2015 (695) 1460 (365) 1488 (244) 1434 (331) –
Enumeration score 4.07 (1.53) 4.00 (1.36) 5.17 (1.33) 4.25 (1.52) 4.12 (1.65) 4.93 (1.07) 6
Subitizing score 14.13 (1.36) 14.13 (1.08) 21.43 (2.73) 22.64 (1.44) 20.00 (2.57) 22.07 (2.23) 25
Calculation score 9.06 (5.31) 9.12 (6.47) 9.36 (6.10) 15.27 (3.99) 12.82 (4.81) 13.86 (3.97) 19

N.B.: standard deviations are given in brackets. Reaction times are measured in msec.

interaction between these factors (F(1,40) ¼ 20.40; p < .0001). Scores on the subitizing task did not differ significantly be-
Children answered faster when they were presented with small tween groups (F(2,44) < 1). The enumeration task showed
numbers of fingers than with large numbers, and they a marginally significant effect of group (F(2,41) ¼ 3.01;
responded faster to canonical hand shapes than to non- p ¼ .060). Post hoc t-tests showed that G1 and G2 were similar
canonical ones. The size effect was larger for non-canonical (t(30) < 1), but both groups performed worse than G3
hand shapes (difference of 2914 msec) than for canonical (t(42) ¼ 2.45; p ¼ .018). No main effect of the group was found
hand shapes (1737 msec). The group had no effect (F(2,40) < 1) for the additions (F(2,44) < 1).
and did not enter into any significant interaction (all Fs < 1). In summary, the main differences between the groups are
In the numerical testing, there were no significant group that G1 and G2 performed less well than G3 on the finger gno-
effects for the counting task involving the extension of the sis test as well as on the two numerical tasks: counting with
verbal numerical chain (F(2,41) ¼ 1.79; p ¼ n.s.). However, limits and enumeration.
a marginally significant effect of the group was obtained
(F(2,44) ¼ 2.68; p ¼ .080) on the counting with limits task. Post 3.2. Post-training assessment
hoc t-tests showed no differences between G1 and G2
(t(31) < 1) but these two groups performed less well than G3 Post-training outcomes were analysed by means of one-way
(t(41.64) ¼ 2.84; p ¼ .007).2 ANOVAs with group as the between-subjects factor. The
There was no significant effect of group (F(2,38) < 1) on accu- main goal of this analysis was to examine whether the fin-
racy on the magnitude comparison task. The median RTs were ger-differentiation intervention led to better finger gnosis
entered into a mixed ANOVA with group as a between-subjects and better numerical performance in G1 than the story-com-
factor, and size of the digits and distance between them as prehension intervention in G2. Therefore, post hoc t-tests
within-subjects factors. This analysis revealed the classical comparing G1 and G2 were also calculated.
effects of distance (F(1,35) ¼ 45.79; p < .0001) and size The ANOVA for the finger gnosis score revealed a signifi-
(F(1,35) ¼ 29.63; p < .0001) with no significant interaction be- cant group effect (F(2,43) ¼ 7.99; p ¼ .001). t-Tests indicated
tween them (F(1,35) < 1). The group effect was not significant better performance for G1 than for G2 (t(22.98) ¼ 4.12;
(F(2,35) < 1) and did not interact with the other factors (all Fs < 1). p < .0001). G1 also performed significantly better than G3
The accuracy of the ordinality judgement did not differ (t(20.90) ¼ 2.22; p ¼ .037), although G2 still performed worse
among the three groups (F(2,37) < 1). Analysis of the RTs than G3 (t(29) ¼ 1.79; p ¼ .083). The means of all the groups
with group as a between-subjects factor, and size of the digits are presented in Table 1.
and distance between them as within-subjects factors No main effects were found in the one-way ANOVA for the
revealed a distance (F(1,37) ¼25.90; p < .0001) and a size effect ‘‘draw a man’’ test (F(2,44) ¼ 1.88; p ¼ n.s.). However the
(F(1,37) ¼ 5.87; p ¼ .002), as well as a significant interaction be- ANOVA on the ‘‘draw a hand’’ task was marginally significant
tween these two factors (F(1,37) ¼ 6.10; p ¼ .018) which indi- (F(2,43) ¼ 2.53; p ¼ .091). Post hoc t-test measures revealed that
cated a smaller distance effect for small numbers (difference G1 had a significantly better score than G2 (t(31) ¼ 2.11;
of 216 msec) than for large ones (difference of 463 msec). The p ¼ .043). G1 also performed marginally better than G3
group effect was not significant (F(2,37) < 1) and did not enter (t(27) ¼ 1.81; p ¼ .081), while the scores of G2 and G3 were sta-
into any significant interactions (all Fs < 1). tistically similar (t(29) < 1).
The one-way ANOVA for the score of the ‘‘how many fin-
2
These degrees of freedom were corrected because Levene’s gers?’’ task with group as between-subjects factor revealed
test indicated that the variances were not equal. no significant group differences (F(2,39) ¼ 2.04; n.s.). A mixed
Author's personal copy

374 cortex 44 (2008) 368–375

ANOVA (between-subjects factor: group; within-subjects fac- The group effect (F(2,37) < 1) was not significant and did not
tors: canonical or non-canonical hand shape, and size (small interact with the other factors (all Fs < 1).
or large) of the quantities) was calculated on the median The one-way ANOVA for the subitizing task was significant
RTs. Again, there were significant effects of size (F(2,43) ¼ 3.35; p ¼ .044). Post hoc t-tests indicated that G1
(F(1,39) ¼ 326.73; p < .0001) and hand shape (F(1,39) ¼ 147.65; scored significantly better than G2 (t(30) ¼ 3.59; p ¼ .046) but
p < .0001), and a significant interaction between size and similarly to G3 (t(27) < 1), while G2 performed worse than G3
hand shape (F(1,39) ¼ 112.52; p < .0001). However the group ef- (t(29) ¼ 2.36; p ¼ .025).
fect was also now significant (F(2,39) ¼ 3.75; p ¼ .032), and it The one-way ANOVA showed no significant differences be-
interacted significantly with size (F(2,39) ¼ 4.27; p ¼ .021). tween the groups on the enumeration task (F(2,44) ¼ 1.32;
Fig. 2 shows that the children in G1 were faster than those p ¼ .277).
in G2, and the difference between the two groups was greater Finally, there were no significant post-training group ef-
when a large number of fingers were presented (difference of fects (F(2,43) ¼ 1.28; p ¼ n.s.) on the additions.
786 msec; t(27) ¼ 3.163, p ¼ .004) than when a small number
of fingers were shown (difference of 271 msec; t(27) ¼ 2.328,
p ¼ .028). The other interactions with group were not signifi- 4. Discussion and conclusions
cant (hand shape  group: F(2,39) < 1, hand shape  si-
ze  group, F(2,39) ¼ 2.14; p ¼ n.s.). In this study, we opted for an intervention method and rea-
There was no group effect for the simple counting task soned as follows: if finger gnosis is functionally related to nu-
(F(2,41) < 1) or for the counting with limits (F(2,44) ¼ 1.01; merical skills development, then improving it should lead to
p ¼ n.s.). positive consequences for number skills as well. With that
The one-way ANOVA on the accuracy of the magnitude aim, we selected two groups of children with poor finger gno-
comparison with group as the between-subjects factor was sis and provided a finger-differentiation intervention to the
not significant (F(2,39) < 1). A mixed ANOVA was performed first one and a story-comprehension intervention to the other
on the RTs with group as the between-subjects factor and one. If good finger gnosis contributes to the development of
the size of the digits and distance between them as numerical skills then training in that skill might also have
within-subjects. As in the pre-training measure, there were positive consequences for numerical skills.
significant distance (F(1,37) ¼ 16.96; p < .0001) and size As expected, our results showed a link between finger gno-
(F(1,37) ¼ 23.20; p < .0001) effects, but no interaction between sis and numerical skills. Before any intervention, children
these two factors (F(1,37) < 1). There was no evidence of with good finger gnosis outperformed the two groups with
a group effect (F(2,37) < 1), and the group did not interact poor finger gnosis in two numerical tasks, one tapping the
with either distance or size (all Fs < 1). level of elaboration of the verbal numerical chain and the
The groups’ scores on the ordinality judgement differed other involving counting collections. More importantly, mea-
significantly (F(2,39) ¼ 3.30; p ¼ .047). Post hoc comparisons sures taken after the intervention indicated that our finger-
showed that children in G2 tended to score worse than those differentiation training was effective. The children who had
in G1 (t(27) ¼ 2.02; p ¼ .053). G2 also scored marginally less received that intervention not only had better finger gnosis
well than G3 (t(26) ¼ 1.82; p ¼ .080), but there were no signif- than the children who had received the control-intervention,
icant differences between G1 and G3 (t(25) < 1). A mixed but also better scores than the group who had good finger gno-
ANOVA was calculated for the RTs of the judgement of ordi- sis to start with. This improvement of the body schema for
nality, with group as the between-subjects factor and the hands was also observed in the ‘‘draw a hand’’ task where,
digits and distance between them as within-subjects factors. once again, children in the finger-intervention group outper-
Analyses revealed significant distance (F(1,37) ¼ 19.25; formed those in the control-intervention group. This could in-
p < .0001) and size effects (F(1,37) ¼ 28.85; p < .0001), which dicate that the finger intervention has improved the internal
did not interact with one another (F(1,37) ¼ 1.42; p ¼ n.s.). representation of the fingers and hands of the participants,
while other representations, such as that of the whole body,
remained stable.
The finger intervention also had an impact on three numer-
ical tasks. First, children in the finger-intervention group were
faster than children in the control-intervention group in saying
how many fingers were raised in pictures, and this effect was
stronger the more fingers were raised. Second, children who fol-
lowed the story-comprehension training scored worse not only
than the control group, but also than the finger-training group.
Finally, children in the finger-intervention group scored better
than those in the control-training group in the subitizing task.
In summary, this study showed that an 8-week training in
finger differentiation led to improved finger gnosis in young
children as well as to better performance in three basic nu-
Fig. 2 – Interaction between the size (small of large) of the merical skills: subitizing, counting raised fingers, and ordinal-
quantities shown and the groups in the ‘‘how many ity judgement. The difference between the two intervention
fingers?’’ task. groups is not obvious in all the numerical tasks, probably
Author's personal copy

cortex 44 (2008) 368–375 375

partly due to the small size of the sample or to the relatively Dehaene S, Piazza M, Pinel P, and Cohen L. Three parietal circuits for
short duration of the intervention. Yet, these results are the number processing. Cognitive Neuropsychology, 20: 487–506, 2003.
first to clearly support the hypothesis of a functional link be- DiLuca S, Grana A, Semenza C, Seron X, and Pesenti M. Finger-
digit compatibility in Arabic numeral processing. The Quarterly
tween finger gnosis and numerical skill development.
Journal of Experimental Psychology, 59: 1648–1663, 2006.
In another study, Gracia-Bafalluy et al. (submitted for pub- Fayol M, Barrouillet P, and Marinthe C. Predicting arithmetical
lication) have shown that children who have been playing achievement from neuropsychological performance:
a musical instrument which requires good finger differentia- a longitudinal study. Cognition, 68: 63–70, 1998.
tion (Costa-Giomi, 2005; Meister et al., 2005), have equivalent Fischer MH. Finger counting habits modulate spatial-numerical
numerical skills to their non-musician peers after 1 year of associations. Cortex, 44: 386–392, 2008.
musical training. Yet, if the comparison is made after 3 years Fuson KC. Children’s Counting and Concepts of Number. New York:
Springer-Verlag, 1988.
of musical practice, the groups differ both in terms of finger
Gallistel CR and Gelman R. Preverbal and verbal counting and
gnosis and numerical skills. From this, we might conclude computation. Cognition, 44: 43–74, 1992.
that very intensive finger-differentiation training is needed to Geary DC. Les troubles d’apprentissage en arithmétique: Rôle de
achieve major and significant differences in numerical skills. la mémoire de travail et des connaissances conceptuelles. In
In that sense, the results we have obtained here after only 8 Noël MP (Ed), La Dyscalculie: Trouble du Développement
weeks of training are very encouraging. A longitudinal follow- Numérique de l’Enfant. Marseille, France: SOLAL, 2005: 169–191.
Gelman R and Gallistel CR. The Child’s Understanding of Number.
up of these children (or of other children who have experienced
Cambridge: Harvard University Press, 1978.
a longer finger-training period) would certainly be valuable.
Gerstmann J. Fingeragnosie und isolerte Agraphie: ein neues
This work could lead to finger training being included in Syndrom. Zeitschrist für die Gesamte Neurologie und Psychiatrie,
the rehabilitation of numerical skills or in work to improve nu- 108: 152–177, 1927.
merical development in children. Thus, new possibilities of Gerstmann J. Syndrome of finger agnosia, disorientation for right
finger-intervention strategies are opened up by the possibili- and left, agraphia and dyscalculia. Archives of Neurology and
ties suggested by the encouraging results of this intervention Psychiatry, 44: 398–408, 1940.
Goodenough F and Pasquasy R. Le test du dessin d’un bonhomme.
programme.
Bruxelles: Editest, 1967.
Grigsby JP, Kemper MB, and Hagerman RJ. Developmental
Gerstmann syndrome without aphasia in fragile  syndrome.
Acknowledgments Neuropsychologia, 25: 881–891, 1987.
Marinthe C, Fayol M, and Barrouillet P. Gnosies digitales et
The authors wish to thank the professionals and the children développement des performances arithmétiques. In Van
from the Martin V (Louvain-la-Neuve), Saint Etienne (Court Hout A, Meljac C, and Fischer JP (Eds), Troubles du Calcul et
Dyscalculies chez l’Enfant. Paris: Masson, 2001: 239–254.
St Etienne) et Saint Pie X (Ottignies) schools for their participation.
Mayer E, Martory MD, Pegna AJ, Landis T, Delavelle J, and
This project was supported by the grant CEE/Marie Curie Annoni JM. A pure case of Gerstmann syndrome with
No. MRTN-CT-2003-504927, an FRSM fund (fund for scientific a subangular lesion. Brain, 122: 1107–1120, 1999.
and medical research, convention 3.4607.04 F) and a concerted Meister I, Krings T, Foltys H, Boroojerdi B, Müller M, Töpper R, and
research action from the French community of Belgium (ARC Thron A. Effects of long-term practice and task complexity in
05/10-327). musicians and non-musicians performing simple and
Marie-Pascale Noël was also supported by the National complex motor tasks: implications for cortical motor
organization. Human Brain Mapping, 25: 345–352, 2005.
Research Fund of Belgium.
Noël MP. Finger gnosia: a predictor of numerical abilities in
children? Child Neuropsychology, 11: 413–430, 2005.
Rusconi E, Walsh V, and Butterworth B. Dexterity with
references numbers: rTMS over left angular gyrus disrupts finger gnosis
and number processing. Neuropsychologia, 43: 1609–1624,
2005.
Alibali MW and DiRusso AA. The function of gesture in learning to Simon O, Mangin JF, Cohen L, LeBihan D, and Dehaene S.
count: more than keeping track. Cognitive Development, 14: Topographical layout of hand, eye, calculation, and language-
37–55, 1999. related areas in the human parietal lobe. Neuron, 33: 475–487,
Andres M, Davare M, Pesenti M, Olivier E, and Seron X. Number 2002.
magnitude and grip aperture interaction. Neuroreport, 22: Sato M and Lalain M. On the relationship between handedness
2773–2777, 2004. and hand-digit mapping in finger counting. Cortex, 44: 393–399,
Andres M, Ostry DJ, Nicol F, and Paus T. Time course of number 2008.
magnitude interference during grasping. Cortex, 44: 414–419, Strub R and Geschwind N. Gerstmann syndrome without aphasia.
2008. Cortex, 10: 378–387, 1974.
Benton AL. The fiction of the ‘‘Gerstmann syndrome’’. Journal of Suresh PA and Sebastian S. Developmental Gerstmann
Neurology, Neurosurgery and Psychiatry, 24: 176–181, 1961. syndrome: a distinct clinical entity of learning disabilities.
Benton AL. Reflections on the Gerstmann syndrome. Brain and Pediatric Neurology, 22: 267–278, 2000.
Language, 4: 45–62, 1977. Turconi E, Campbell JID, and Seron X. Numerical order and quantity
Brissiaud R. Comment les Enfants Apprennent à Calculer. Paris: Retz, processing in number comparison. Cognition, 98: 273–285, 2006.
2003. Zago L and Pesenti M. Neuroimagerie cognitive des activités
Butterworth B. The Mathematical Brain. London: Macmillan, 1999. numériques. In Houdé O, Mazoyer B, and Tzourio-Mazoyer N
Costa-Giomi E. Does music instruction improve fine motor (Eds), Cerveau et Psychologie: Introduction à l’Imagerie Cérébrale
abilities? Annals of the New York Academy of Sciences, 1060: Anatomique et Fonctionnelle. Paris: Presses Universitaires de
262–264, 2005. France, 2002.

You might also like