0% found this document useful (0 votes)
17 views16 pages

Abacus Training's Cognitive Transfer Effects

Uploaded by

x2m7ndf99p
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
17 views16 pages

Abacus Training's Cognitive Transfer Effects

Uploaded by

x2m7ndf99p
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Current Psychology

https://doi.org/10.1007/s12144-021-01968-1

Transfer effects of abacus training on cognition


Yujie Lu 1,2,3 & Mengyi Li 1,2,3 & Zhijun Cui 1,2,3 & Li Wang 1,2,3 & Yuwei Hu 1,2,3 & Xinlin Zhou 1,2,3

Accepted: 3 June 2021


# The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2021

Abstract
The Chinese abacus was invented 1800 years ago as a piece of calculation equipment for economic education and support. The
abacus no longer serves as an economic tool but has emerged as a powerful educational tool to promote individual development.
It is not yet known, however, whether abacus training may lead to cognitive transfer in the individual development context. This
transfer effect was investigated in the present study with a randomized controlled design centered on comparison against abacus
training with cognitive training and English training. The final sample consisted of 343 vocational school/college students (55
males, mean age = 17.79 years, range: 14.17 to 24.67 years) whom completed the pre-test, the 52-h training session, and the post-
test. Participants in each school or college were randomly assigned into three training groups based on their pre-test scores.
Abacus training is shown to promote calculation and spatial abilities by comparison with English training, and to promote
calculation ability and processing speed by comparison with cognitive training. Cognitive training shows transfer effect of spatial
ability by comparison with English training. However, abacus training does not show pervasive transfer effect for any general
cognitive abilities that do not typically involve numerical processing or visuospatial processing. These findings suggest a limited
transfer effect of abacus training on cognitive development.

Keywords Mathematical cognition . Abacus training . Transfer effect . Arithmetical cognition . Numerical processing . Spatial
ability

Introduction includes beads strung across rows in a frame (Fig. 1). The
modern Chinese abacus typically has one bead above (the
The abacus was invented in China 1800 years ago as a piece of “top bead”), with each standing for five, and four beads below
calculation equipment for economic education and support. (the “bottom bead”), with each standing for one, allowing for
The abacus has been replaced by computer technology, but computations in the decimal system.
abacus training persists as an educational tool for promoting Abacus training generally includes two stages: “real” and
individual development. The basic structure of the abacus “mental”. The former is the basis for the latter. During the “real”
training stage, leaners perform computation with the use of a real
Yujie Lu, Mengyi Li, Zhijun Cui and Li Wang contributed equally to this
abacus. During the “mental” training stage, leaners have skilled
work. enough with the real one, and perform mental calculations by
operating an imagined mental abacus (Frank & Barner, 2012).
This research was supported by Advanced Innovation Center for Future
Education (No. 27900-110631111), the 111 Project (No. BP0719032), Compared with the conventional mental calculation, mental aba-
and National Natural Science Foundation of China (Nos. 31600896, cus calculation involves the use of visuospatial images (Chen
31671151). et al., 2006). Children can acquire simple abacus skills with
corresponding mental skills after several months of training
* Xinlin Zhou (e.g., Liu & Sun, 2017; Wang et al., 2015).
zhou_xinlin@bnu.edu.cn
Parents, educators, and teachers have reported the positive
1
State Key Laboratory of Cognitive Neuroscience and Learning, IDG/
transfer effects of abacus training. It is yet uncertain, however,
McGovern Institute for Brain Research, Beijing Normal University, whether abacus training yields transfer effects on cognitive
Beijing 100875, China abilities. Some researchers have observed positive transfer
2
Advanced Innovation Center for Future Education, Beijing Normal effects (e.g., Ku et al., 2012; Liu & Sun, 2017; Tanaka et al.,
University, Beijing, China 2002; Wang et al., 2015; Wang et al., 2019; Zhou et al., 2019),
3
Siegler Center for Innovative Learning, Beijing Normal University, whereas others have not (e.g., Barner et al., 2016; Lee et al.,
Beijing, China 2007; Na et al., 2015). Differences in training intensity and
Curr Psychol

Fig. 1 The structure of the


decimal system abacus. There is
one bead above, which stands for
five; and there are four beads
below, each of which standing for
one, allowing for computations in
the decimal system

experimental approach may be the cause of these mixed ef- Wang et al., 2019), and even language processing (Jiang,
fects. For instance, training intensity has varied from 20 days 2009, 2012), but other studies have failed to demonstrate such
(Dong et al., 2016) to more than 17 years (Hanakawa et al., effects.
2003). Experimental approaches also may center on post-hoc Liu and Sun (2017) found that long-term abacus training
analysis (e.g., Ku et al., 2012; Tanaka et al., 2002; Wang et al., promoted children’s spatial attention orientation ability on a
2015) or on randomized control designs (Barner et al., 2016; Posner spatial cueing task. Na et al. (2015) found that children
Barner et al., 2017; Wang et al., 2019; Zhou et al., 2019). with abacus training made fewer errors than controls in tasks
requiring sustained attention inhibition. Again, studies have
Near Transfer Effect also failed to show far-reaching transfer in attention abilities,
for example, in selective attention (e.g., Matías-Guiu et al.,
Transfer effects can be divided into two categories: near trans- 2016; Na et al., 2015).
fer (applying or transferring what is learned to the same or a Studies on memory abilities (excluding numerical memory
closely similar domain) and far-reaching transfer (transferring ability, which can be considered near transfer) have also pro-
what is learned to a quite different domain). Given that abacus duced inconsistent results. Tanaka et al. (2002) found that
training and mental abacus calculation involves the perception adult abacus experts and children showed better letter memory
and processing of numerical values of abacus beads and than controls. Works by Chen et al. repeatedly observed trans-
Arabic numbers, the effect of abacus training on numerical fer effect in the spatial working memory as-measured with the
abilities can be seen as near transfer. Previous studies have n-back paradigm (Dong et al., 2016; Wang et al., 2019; Zhou
produced consistent results regarding the near transfer of aba- et al., 2019). Others have found that the transfer effect for
cus training. Abacus training has been shown to promote cer- memory can only observed in digit material, and not in non-
tain skills associated with basic numerical processing (e.g., Du digit material such as abstract shapes (Kawakami, 1995), spa-
et al., 2014; Wang et al., 2013; Yao et al., 2015), typical tial working memory (Barner et al., 2016; Lee et al., 2007; Na
arithmetic based on Arabic digits (e.g., Amaiwa & Hatano, et al., 2015).
1989; Hatano et al., 1977; Liu et al., 2006; Stigler, 1984; Some studies found the transfer abacus training to induc-
Wang et al., 2015; Wang et al., 2019; Zhou et al., 2019), tive reasoning ability, in either matrix reasoning (e.g., Fan
and numerical memory (i.e., digit span, e.g., Ku et al., 2012; et al., 2005; Jia et al., 2021; Jiang, 2012; Liu et al., 2006;
Tanaka et al., 2002). Wang et al., 2019) or numerical inductive reasoning (Jia
et al., 2021), while others found no differences (e.g. Amaiwa
Far-Reaching Transfer Effect & Hatano, 1989; Wang et al., 2013; Zhou et al., 2019; Zhou
et al., 2020). As mentioned above, these mixed findings may
The far-reaching transfer of abacus training, as mentioned be a result of distinct experimental approaches.
above, refers to the transferring in regard to other non- To our knowledge, most previous studies on abacus train-
trained abilities. Scholars have reached mixed findings on this ing are post-hoc investigations (e.g., Chen et al., 2006; Du
topic. Previous studies have revealed transfer effects in atten- et al., 2013; Tanaka et al., 2002; Wang et al., 2013; Wang
tion (e.g., Geeta & Gavas, 2014; Stigler, 1984), memory (e.g., et al., 2015; Wang et al., 2017). Few studies have been con-
Amaiwa & Hatano, 1989; Bhaskaran et al., 2006; Hatano ducted with randomized control designs (Barner et al., 2016;
et al., 1977; Liu et al., 2006; Liu & Sun, 2017; Stigler, 1984; Barner et al., 2017; Dong et al., 2016; Wang et al., 2017;
Sun et al., 2006; Wang et al., 2015; Zhen et al., 2007), rea- Wang et al., 2019; Yao et al., 2015; Zhou et al., 2019; Zhou
soning (e.g., Fan et al., 2005; Jiang, 2012; Liu et al., 2006; et al., 2020). The control groups in most randomized control
Curr Psychol

studies received no parallel active training, which may have 1) Though both the abacus training and cognitive training
introduced confounds. Only Barner et al. (2016, 2017) used a may improve numerical ability and spatial ability, abacus
standard math curriculum to train children in their control training would have larger pre-to-post improvements.
group, one study showed only transfer effect of abacus in 2) Cognitive training would have larger pre-to-post im-
arithmetic performance (2016), and another found no signifi- provements on abilities that the training targets (i.e., process-
cant transfer in cognitive abilities or arithmetic computation ing speed, numerical ability, spatial ability, attention, and
(2017). An active control group is essential, as it ensures a memory ability) when compared with English training.
comparable experience across participants in the control and 3) The English training group (baseline) would show no
treatment groups, minimizes motivational differences be- advantages relative to abacus or cognitive training in the cog-
tween the groups, and eliminates the placebo effect (Simons nitive abilities that explored here.
et al., 2016).

The Current Investigation


Methods
A randomized control scheme was designed in this study to
Participants
explore the near as well as far-reaching transfer effects of
abacus training. A series of cognitive abilities were examined
A total of 347 participants were recruited from five vocational
with a pre-post randomized control design, enabling us to
schools/colleges in China (56 males, mean age = 17.16 years,
explore both the near (i.e., calculation ability, digit span, num-
range: 14.17 to 24.67 years). All participants are right-handed
ber sense) and far-reaching (i.e., processing speed, attention,
with normal or corrected-to-normal visual acuity. Participants
memory, spatial ability, language ability, inductive reasoning)
in each school were assigned into one of the three training
transfer effects. Basic cognitive ability (i.e., processing
groups (abacus training, cognitive training, or English train-
speed), language ability, and domain general inductive rea-
ing) pseudo-randomly with provision to match them for age,
soning ability were also tested, as they all essential variables
gender, and pre-test scores. One participant from the abacus
for the pseudo-random assignment of students. Next, cogni-
training group and three from the cognitive training group
tive training and English training were treated as parallel ac-
failed to complete the training procedure. At the end of the
tive control groups to remedy drawbacks in previous study
study, 343 participants (55 males, mean age = 17.79 years,
designs.
range: 14.17 to 24.67 years) had completed all procedures
Cognitive training was selected as it has been shown to
include pre-test, training, and post-test portions. There were
directly enhance cognitive abilities (e.g., Anguera et al.,
117 participants in abacus training, 112 in cognitive training,
2013; Cheng et al., 2019; Jaeggi, Buschkuehl et al., 2011;
and 114 in English training (Fig. 2). Statistical analysis
see a review by Simons et al., 2016). It was selected for the
showed that the three groups did not differ in terms of gender
current study to explore whether abacus training is the only
distribution, age, or pre-tests’ scores (all p > .05), suggesting
approach to enhancing certain cognitive abilities. A compari-
that the randomization was effective (Table 1).
son between cognitive training and abacus training may reveal
The investigation received consent from the Institutional
the advantages of abacus training over other cognitive training
Review Board of the authors’ affiliated institution. Written
tools. For example, one meta-analysis showed that spatial vid-
consent forms were also provided by each participant before
eo game training, course training and spatial task training also
the study.
could significantly improve spatial skills (Uttal et al., 2013).
There has been no previous study comparing the cognitive
transfer effect between abacus training and direct cognitive Training Procedure
training. English training was also used in this study as a
baseline, as it typically centers on the improvement of lan- The current study implemented a computerized training pro-
guage ability (e.g., English listening, speaking) outside of gram along with a pre- and post-test design. Participants in
general cognitive abilities or numerical abilities, making it each training group were trained extensively for two months,
an ideal parallel active control. with 1 h in each lesson covering 52 lessons in total. Training
Based on the studies mentioned above, and because abacus sessions were performed during regular class hours. All par-
training inherently involves numerical processing, and active ticipants from each school were regular classroom peers.
manipulation of the imagined beads (visuospatial processing), Students participated in training on a voluntary basis. If they
we hypothesize that abacus training has transfer effects limited did not want to participate in training, they were given learn-
to numerical processing (near transfer) and spatial-related pro- ing tasks following their regular curriculum. Three to eight
cessing (far-reaching transfer) capabilities. We make the fol- lessons were given each week, depending on the schedule
lowing specific predictions. availability of schools/colleges.
Curr Psychol

Fig. 2 Experimental procedure flow diagram

Each training group was administered by an instructor, all computer to allow the tasks to be completed online. There
of whom were teachers from experiment school or educational were three types of training task administered: basic knowl-
organizations. We have trained these instructors regarding the edge of the abacus, abacus decoding (associating the abacus
study principles before the training. Each lesson was moni- beads images with certain numbers in both visual and auditory
tored through the teaching guidelines proposed by the instruc- form), and calculation training (mental addition and subtrac-
tor. The trainee students were asked not to communicate the tion via the visual or auditory materials). Details regarding the
contents of training to others to prevent any interference. All abacus training are given in Supplementary Material.
training was completed in the on-campus computer room. All
training tasks were web-based and are available at the “Online
Psychological Experiment System (OPES)” (www.dweipsy. Cognitive Training
com/lattice) and “Web-based Learning System” (www.
dweipsy.com/exam). Cognitive training was designed to train participants on cog-
nitive ability. There were 28 cognitive training tasks in total
Abacus Training encompassing five categories: processing speed, numerical
ability, spatial ability, attention, and memory ability.
The goal of the training was to give the participants basic and Participants completed 5 to 8 tasks per lesson, ultimately re-
mental abacus skills. This portion of the training process was ceiving training on each task for 15 times. All these tasks were
administered on an abacus, and accompanied by a personal combined with games to stimulate the participants’ interest.
Curr Psychol

Table 1 Means, standard errors, and main effect of group based one-way ANOVA analysis for all variables in pre-test

Variables/ability Name Index Abacus Cognitive English F p Split-half reliability

Demographic Male/Female Number 20/97 16/96 19/95 – – –


Age Year 17.7 (0.2) 17.7 (0.2) 17.9 (0.2) 0.50 .609 –
Processing speed Choice reaction time Reaction time (ms) 355 (4) 356 (7) 370 (6) 2.31 .100 .92
Number sense Numerosity comparison Proportion correct (%) 80.4 (0.7) 81.1 (0.7) 81.4 (0.7) 0.50 .607 .77
Calculation Simple subtraction Adj. No. of correct trials 49.3 (0.6) 49.3 (0.7) 48.4 (0.8) 0.57 .567 .92
Complex subtraction Adj. No. of correct trials 25.3 (0.6) 25.3 (0.6) 25.1 (0.6) 0.04 .959 .92
Multi-step computation Adj. No. of correct trials 45.4 (1.6) 43.7 (1.5) 42.9 (1.6) 0.65 .525 .98
Spatial 2D mental rotation Adj. No. of correct trials 27.2 (1.6) 26.8 (1.7) 27.1 (1.5) 1.72 .180 .99
3D mental rotation Adj. No. of correct trials 20.7 (1.0) 20.9 (0.9) 20.6 (1.0) 0.02 .982 .89
Paper folding task No. of correct trials 7.6 (0.3) 7.2 (0.3) 7.4 (0.2) 0.59 .557 .83
Attention Visual tracking No. of correct trials 19.5 (0.4) 19.3 (0.5) 19.8 (0.5) 0.21 .812 .92
UFOV attention Accuracy (%) 92.5 (0.4) 92.6 (0.4) 92.6 (0.5) 0.01 .993 .98
Focused attention No. of correct trials 30.3 (1.2) 31.4 (1.2) 29.9 (1.2) 0.42 .659 .96
Distributed attention No. of correct trials 27.5 (0.8) 28.4 (0.7) 27.1 (0.8) 0.69 .502 .88
Geometric form searching No. of correct trials 75.7 (1.6) 76.8 (2.1) 74.4 (2.1) 0.41 .664 .96
Memory Digit span (Forward) No. of recalled digits 8.3 (0.2) 8.8 (0.1) 8.7 (0.1) 2.47 .086 –
Digit span (Backward) No. of recalled digits 6.7 (0.2) 6.8 (0.2) 6.6 (0.2) 0.16 .853 –
Spatial 2-back Hit minus false alarm 23.6 (0.7) 25.2 (0.6) 24.4 (0.6) 1.72 .180 .99
Spatial 4-back Hit minus false alarm 18.8 (0.5) 19.2 (0.6) 19.4 (0.5) 0.43 .654 .99
Spatial short-term memory Accuracy (%) 85.1 (0.4) 85.4 (0.5) 84.8 (0.6) 0.43 .648 .92
Visual short-term memory Proportion correct (%) 75.0 (0.6) 73.8 (0.7) 74.0 (0.7) 0.98 .375 .67
Language Sentence completion Adj. No. of correct trials 36.5 (0.7) 36.9 (0.7) 36.3 (0.7) 0.19 .829 .85
Inductive reasoning Nonverbal matrix reasoning No. of correct trials 10.5 (0.6) 11 (0.6) 10.7 (0.7) 0.19 .824 .96

Note. Adjust no. of correct trials = Total correct trials minus total incorrect trials

Details regarding the cognitive training are given in study. First, each group received the same English
Supplementary Material. coursework. Second, our training content for the English
group focused on listening and simple oral expression rather
English Training than reading and writing. Previous studies have indicated that
listening, oral expression, reading, and written expression are
English training was designed to enhance the participants’ related but distinct linguistic systems (Berninger & Abbott,
English listening and speaking skills. The materials were de- 2010).
signed by the researchers under the supervision of the school’s
full-time English teacher, to ensure their suitability to the par- Cognitive Tests
ticipants. Each training lesson included three phases. First was
a listening phase, where participants listened to the material Within one week before and one week after the training, par-
and then answered questions presented on the computer ticipants completed 20 computerized cognitive tests from the
screen to demonstrate their comprehension. Next came a “Online Psychological Experiment System (OPES)” (www.
learning phase, where participants read texts and received re- dweipsy.com/lattice; Wei et al., 2012; Zhou et al., 2015).
lated vocabulary and grammar instruction from the English These tests evaluated the participants’ processing speed
group instructor. Finally, participants completed an oral com- (choice reaction time), number sense (numerosity
munication phase, where they asked each other questions comparison), calculation ability (simple subtraction,
about the material in English. Details regarding English train- complex subtraction, multi-step computation), spatial ability
ing are given in Supplementary Material. (two-dimensional mental rotation, three-dimensional mental
Participants in all three groups have regular English classes rotation, paper folding task), attention (visual tracking,
at their schools or colleges, however, their regular English UFOV attention, focused attention, distributed attention, geo-
classes typically focus on reading and writing – these partic- metric form searching), memory (digit span, spatial 2-back,
ular skills would not have influenced the results of the current spatial 4-back, spatial short-term memory, visual short-term
Curr Psychol

memory), language ability (sentence completion), and reason- Results


ing ability (nonverbal matrix reasoning). Figure 3 shows ex-
amples of all cognitive tests; one screen of stimulus is repre- The means and standard errors of the pre-test scores for all
sented as a black box. For example, in the test for focused tests are shown in Table 1. Post-test and training effect (post-
attention, each trial is consisted of six successively presented test score minus pre-test score) scores for all tests are shown in
stimuli. All cognitive tests were confirmed to have acceptable Table 2. The repeated-measure MANOVA analysis results
reliability (Table 1). Detailed descriptions of the cognitive showed that at multivariate level, the main effect of test time
tests are given in the Appendix. was significant, F (21, 320) = 30.36, p < .001, η2 = .666, and
The pre-test and post-test were administered by trained the interaction between group and test time was significant, F
experimenters who were unaware of the participants’ group (42, 642) = 2.53, p < .001, η2 = .142. There was no multivari-
assignments. One experimenter tested four participants at the ate main effect by group, F (42, 642) = 1.34, p = .078,
same time in a quiet room. All tests were completed in two η2 = .081. Follow-up univariate results for all tests are shown
hours and spaced between a 30-mintue resting periods. The in Table 3.
experimenters gave instructions, allowed the participants to There was significant interaction between group and test
complete a practice test, and received confirmation that the time on the choice reaction time test, F (2, 340) = 3.17,
participants understood the task before the formal test began. p = .043, η2 = .018. Post-hoc analysis showed that in the
Participants used desktop computers to finish all the tests. post-test, the abacus training group had significantly lower
reaction times than the cognitive training group (p = .006).
Data Analyses There was no significant difference between the abacus and
the English training groups (p = .098) or between the cogni-
To examine the training effects between the abacus training, tive and the English training groups (p = .275). Additionally,
cognitive training, and English training, a repeated-measure for the cognitive training group, the reaction times in the post-
MANOVA was performed with group (abacus training, cog- test was significantly longer than those in the pre-test
nitive training, and English training) as the between-subject (p = .012), and there was no significant pre-to-post difference
variable, test time (pre-test and post-test) as the repeated mea- for the abacus (p = .705) or the English (p = .490) training
sures variable, and all test scores as dependent variables. group.

Fig. 3 Examples for each test at pre-test and post-test. Note. 2D mental rotation: Two-dimensional mental rotation, 3D mental rotation: Three-
dimensional mental rotation
Curr Psychol

Table 2 Means and standard errors of post-test scores and training effects (post-test score minus pre-test score) for all tests

Ability Test name Index Post-test score Training effect

Abacus Cognitive English Abacus Cognitive English

Processing speed Choice reaction time Reaction time (ms) 352 (4) 374 (8) 365 (4) -3 (4) 17 (10) -5 (6)
Number sense Numerosity comparison Proportion correct (%) 81.4 (0.6) 81.5 (0.8) 81.0 (0.7) 0.9 (0.8) 0.5 (0.9) −0.4 (0.8)
Calculation Simple subtraction Adj. No. of correct trials 49.8 (0.9) 48.7 (0.7) 47.3 (0.9) 0.4 (0.9) −0.7 (0.6) −1.1 (0.9)
Complex subtraction Adj. No. of correct trials 27.6 (0.7) 25.5 (0.7) 25.5 (0.6) 2.3 (0.6) 0.3 (0.6) 0.4 (0.5)
Multi-step computation Adj. No. of correct trials 54.8 (1.7) 43.4 (1.7) 43.1 (1.8) 9.5 (1.4) −0.3 (1.1) 0.2 (1.0)
Spatial 2D mental rotation Adj. No. of correct trials 36.4 (1.6) 37.0 (1.6) 31.1 (1.6) 9.1 (1.2) 10.1 (1.3) 4.0 (1.3)
3D mental rotation Adj. No. of correct trials 25.0 (1.1) 23.4 (1.1) 21.5 (1.2) 4.3 (1.0) 2.5 (0.9) 0.8 (0.9)
Paper folding task No. of correct trials 10.1 (0.4) 10.1 (0.4) 9.9 (0.4) 2.5 (0.3) 2.9 (0.4) 2.5 (0.4)
Attention Visual tracking No. of correct trials 21.5 (0.5) 21.3 (0.6) 21.3 (0.6) 2.0 (0.4) 2.0 (0.5) 1.5 (0.5)
UFOV attention Accuracy (%) 94.2 (0.2) 94.1 (0.3) 93.4 (0.4) 1.7 (0.3) 1.5 (0.4) 0.8 (0.4)
Focused attention No. of correct trials 38.5 (1.1) 38.7 (1.1) 34.8 (1.2) 8.1 (1.0) 7.3 (1.1) 5.0 (1.1)
Distributed attention No. of correct trials 30.1 (0.8) 32.1 (0.8) 29.1 (0.9) 2.6 (0.6) 3.7 (0.8) 2.0 (0.7)
Geometric form searching No. of correct trials 84.3 (1.8) 87.9 (1.8) 86.0 (1.7) 8.6 (1.8) 11.1 (1.7) 11.6 (1.7)
Memory Digit span (Forward) No. of recalled digits 8.7 (0.2) 8.7 (0.2) 8.9 (0.1) 0.4 (0.3) −0.1 (0.2) 0.2 (0.2)
Digit span (Backward) No. of recalled digits 7.1 (0.2) 7.1 (0.2) 7.0 (0.1) 0.4 (0.2) 0.4 (0.2) 0.3 (0.2)
Spatial 2-back Hit minus false alarm 24.4 (0.6) 25.7 (0.6) 25.2 (0.5) 0.9 (0.7) 0.5 (0.6) 0.8 (0.6)
Spatial 4-back Hit minus false alarm 18.1 (0.6) 20.5 (0.6) 18.3 (0.7) −0.6 (0.6) 1.3 (0.6) −1.2 (1.7)
Spatial short-term memory Accuracy (%) 86.7 (0.3) 86.0 (0.5) 86.0 (0.4) 1.6 (0.4) 0.6 (0.5) 1.2 (0.5)
Visual short-term memory Proportion correct (%) 74.1 (0.8) 73.4 (0.7) 71.7 (1.0) −0.9 (0.8) −0.4 (0.7) −2.3 (1.0)
Language Sentence completion Adj. No. of correct trials 36.7 (0.7) 36.3 (0.8) 37.1 (0.7) 0.2 (0.8) −0.7 (0.7) 0.8 (0.7)
Inductive reasoning Nonverbal matrix reasoning No. of correct trials 13.3 (0.6) 13.2 (0.6) 12.7 (0.7) 2.8 (0.8) 2.1 (0.8) 2.0 (0.8)

Note. Adjust no. of correct trials = Total correct trials minus total incorrect trials

There was significant interaction between group and test There was significant interaction between group and test
time on the complex subtraction test, F (2, 340) = 4.09, time on the 2D mental rotation test, F (2, 340) = 6.60,
p = .018, η2 = .023. Post-hoc analysis showed significantly p = .002, η2 = .037. Post-hoc analysis showed that on the
higher post-test scores in the abacus training group than the post-test, scores in the abacus (p = .019) and cognitive
cognitive (p = .025) and English (p = .024) training groups, (p = .010) training groups were significantly higher than those
and no significant difference between the cognitive and in the English training group; there was no significant differ-
English training groups (p = .992). Additionally, post-test ence between the abacus and cognitive training groups
scores in the abacus training group were also significantly (p = .788). Post-test scores were also significantly higher than
higher than that on the pre-test (p < .001), and there was no pre-test scores across all three training groups (all p < .01).
significant pre-to-post difference for the cognitive (p = .659) There was significant interaction between group and test
or English (p = .435) training group. time on the 3D mental rotation test, F (2, 340) = 3.50,
There was significant interaction between group and test p = .031, η2 = .020. Post-hoc analysis showed that post-test
time on the multi-step computation test, F (2, 340) = 22.19, scores in the abacus training group were significantly higher
p < .001, η2 = .115. Post-hoc analysis showed that the aba- than in the English training group (p = .029), and there was no
cus training group’s post-test scores were significantly significant difference between abacus and cognitive training
higher than those of the cognitive (p < .001) or English groups (p = .330) or between cognitive and English training
(p < .001) training groups, and there was no significant groups (p = .232). Additionally, post-test scores were signifi-
difference between the cognitive and English training cantly higher than pre-test scores for the abacus (p < .001) and
groups (p = .887). Additionally, for the abacus training cognitive (p = .007) training groups, with no such significant
group, post-test scores were significantly higher in pre- difference in the English training groups (p = .373). The train-
test scores (p < .001). There was no significant pre-to- ing effects among each group from the five schools on com-
post difference for the cognitive (p = .804) or English plex subtraction, multi-step computation, 2D mental rotation,
(p = .864) training group. and 3D mental rotation tests are plotted in Fig. 4.
Curr Psychol

Table 3 Univariate test results for all tests

Test name Main effect of group Main effect of test time Interaction of group × test time

F p η2 F p η2 F p η2

Choice reaction time 3.07 .048* .018 0.74 .391 .002 3.17 .043* .018
Numerosity comparison 0.14 .873 .001 0.52 .473 .002 0.65 .524 .004
Simple subtraction 1.66 .192 .010 0.93 .335 .003 1.00 .367 .006
Complex subtraction 1.27 .282 .007 9.41 .002** .027 4.09 .018* .023
Multi-step computation 6.54 .002** .037 21.26 <.001*** .059 22.19 < .001*** .115
2D mental rotation 1.20 .304 .007 111.62 <.001*** .247 6.60 .002** .037
3D mental rotation 0.93 .397 .005 22.62 <.001*** .062 3.50 .031* .020
Paper folding task 0.17 .842 .001 158.70 <.001*** .318 0.41 .665 .002
Visual tracking 0.04 .960 .001 48.44 <.001*** .125 0.45 .640 .003
UFOV attention 0.52 .595 .003 38.63 <.001*** .102 1.68 .187 .010
Focused attention 1.86 .157 .011 122.65 <.001*** .265 2.36 .096+ .014
Distributed attention 2.07 .127 .012 48.03 <.001*** .124 1.46 .233 .009
Geometric form searching 0.64 .526 .004 107.57 <.001*** .240 0.89 .412 .005
Digit span (Forward) 1.10 .333 .006 2.17 .142 .006 1.38 .254 .008
Digit span (Backward) 0.26 .770 .002 14.37 <.001*** .041 0.10 .906 .001
Spatial 2-back 2.12 .122 .012 3.65 .057+ .011 0.07 .929 .000
Spatial 4-back 2.16 .117 .013 0.19 .663 .001 3.37 .036* .019
Spatial short-term memory 0.53 .590 .003 18.69 <.001*** .052 1.10 .335 .006
Visual short-term memory 1.84 .161 .011 5.71 .017* .017 1.17 .313 .007
Sentence completion 0.02 .983 .001 0.05 .830 .001 0.98 .376 .006
Nonverbal matrix reasoning 0.19 .826 .001 26.97 <.001*** .074 0.33 .722 .002

Note. + p < .1 (borderline significant), * p < .05, ** p < .01, *** p < .001

There was significant interaction between group and test attention, geometric form searching, digit span (backward),
time on the spatial 4-back test, F (2, 340) =3.37, p = .036, spatial short-term memory, and nonverbal matrix reasoning
η2 = .019. Post-hoc analysis showed that the cognitive training (all p < .001) tests. In all cases, post-test scores were higher
groups’ pose-test scores were significantly higher than those than pre-test scores.
of abacus (p = .012) or English (p = .019) training group, and
there was no significant difference between the abacus and the
English training group (p = .875). Additionally, the cognitive
training groups’ post-test scores were borderline-significantly Discussion
higher than their pre-test scores (p = .070), and there was no
significant pre-to-post difference for the abacus (p = .362) and A randomized controlled design with a large battery of cog-
English (p = .093) training groups. nitive tests was adopted in this study to explore the transfer
The interaction between group and test time on the focused effect of abacus training. A cognitive training group and
attention test was borderline-significant, F (2, 340) =2.36, English training group served as active control groups, en-
p = .096, η2 = .014. Post-hoc analysis showed that the abacus abling us to comparatively investigate abacus-specific transfer
(p = .020) and cognitive (p = .015) training groups’ post-test effects. Abacus training had near transfer effect on calculation
scores were significantly higher than those of the English ability (only for complex subtraction and multi-step computa-
training group, and there was no significant difference be- tion) compared with English and cognitive training. Besides,
tween abacus and cognitive training group (p = .885). abacus training also showed far-reaching transfer effect on
Additionally, post-test scores were also significantly higher spatial ability (for both 2D and 3D mental rotation) when
than pre-test scores across all three training groups (all compared against English training. These results supported
p < .001). our hypothesis that abacus training has transfer effects limited
The main effect of test time was significant on the paper to numerical- and spatial-related processing capabilities. It is
folding, visual tracking, UFOV attention, distributed worth noting that our focus is on the relative transfer effect of
Curr Psychol

Fig. 4 Training effect (post-test score minus pre-test score) of five mental rotation: Two-dimensional mental rotation, 3D mental rotation:
schools for each training group in complex subtraction, multi-step com- Three-dimensional mental rotation
putation, 2D mental rotation and 3D mental rotation tests. Note. 2D

abacus training by comparison with cognitive training and people’s spatial ability (Cherney, 2008; Cherney et al.,
English training. 2014), working memory (Fellman et al., 2020; Forsberg
Cognitive training group showed significant larger pre-to- et al., 2020), and attention (Ruiz-Marquez et al., 2019;
post improvements in spatial ability (2D mental rotation), Ziegler et al., 2019) can be improved by short-term training.
working memory (spatial 4-back), and attention (focused at- However, we observed a decrease in processing speed (choice
tention) than English training group. The improvements in reaction time) in the cognitive training group, which
spatial ability and working memory were consistent with our contradicted our hypothesis. Previous studies on processing
hypothesis. Previous studies have also shown that young speed training often center on older adults or clinical cohorts
Curr Psychol

(e.g. Takeuchi & Kawashima, 2012). It is possible that it here over time. The result suggested that cognitive training and
less capacity for the training effect in our young adult cohort English training have comparable transfer to digit span com-
compared to older participants. It is also possible that non- pared with abacus training.
adaptive processing speed training led to fatigue over day-
by-day cognitive training in our participants. Number Sense
Consistent with our hypothesis, the English training group
showed no significantly greater improvement in any ability No significant group differences were found among the three
when compared against the other two groups. We further ex- groups in post-test scores on number sense. Only one previous
plored abacus training transfer effects by separating them into study with children abacus experts showed significant im-
two categories, near and far-reaching, as discussed in greater provement in number sense (Cui et al., 2020). It is possible
detail below. that for adult learners, abacus training has a limited effect on
improving the adult trainee’s number sense skills.
Near Transfer
Far-Reaching Transfer
We tested numerical-related processing capabilities via tasks
centered on calculation, digit span, and number sense, as they Spatial Ability
directly involve numbers or numerosity manipulating. Abacus
training led to greater improvement in complex computation Spatial ability encompasses various mental operations in-
skills (multi-digit subtraction and multi-step computation) volved in visualizing, manipulating, and reasoning in regard
than cognitive or English training. No significant greater to visual-spatial objects. This study centered on two classic
pre-to-post improvements were observed for abacus training spatial skills measures: mental rotation (Shepard & Metzler,
in simple computation, digit span, or number sense compared 1971) and paper folding (Shepard & Feng, 1972), which are
to the other two types of training. commonly used to measure an individual’s spatial ability
(e.g., Wai et al., 2009; Wei et al., 2012). The three groups
Calculation Ability showed same level of improvement on paper folding test.
Compared with English training, both abacus training and
The near transfer effect of complex computations (Fig. 4) cognitive training demonstrated significantly higher levels of
observed in this study was consistent with previous studies improvement on 2D and 3D mental rotation tests, both of
(e.g., Barner et al., 2016; Hatano et al., 1977; Zhou et al., which involve active spatial manipulation.
2019). The complex subtraction test is a one-step mental ar- Among these tasks, we only observed abacus training
ithmetic test comprised of double-digits, and is generally used transfer effect for mental rotation. The possible explanation
to measure individual’s arithmetic fluency, i.e., his or her abil- is that mental rotation task involves elements that more closely
ity to solve simple symbolic arithmetic problems fluently and mimic the abacus training process. Mental rotation involves
accurately (Carr & Alexeev, 2011; Zhang et al., 2019). The only rigid transformation, while paper folding involves local
multi-step calculation test is a three-step mental arithmetic test piece-wise rigid transformations (Atit et al., 2013). During the
that involves maintaining and updating intermediate calcula- abacus training, the transformation of the visual images was a
tion results (Ding et al., 2019). Near transfer in calculation bead-based rigid transformation, which was similar to the
may be attributable to the familiarity with regular arithmetic mental rotation task. Learners in the abacus training group
facts rather than the direct utilization of the “mental abacus”. It were also trained to use visuospatial strategies and actively
is possible that no transfer effect for simple arithmetic (i.e., the manipulate visuospatial images for numerical calculation
inverse operation of single-digit addition) can be attributed to (Hatano et al., 1977; Hatano & Osawa, 1983), which may
a ceiling response for the simple arithmetic. have enhanced their spatial as well as visual strategies in com-
pleting tasks that involved active manipulation of objects (i.e.,
Digit Span mental rotation), but had minor or no transfer effect for spatial
tasks that require a very high level of passive spatial memori-
No greater improvements were found for abacus training ver- zation (i.e., paper folding).
sus other training types on the digit span test Previous studies
on abacus have repeatedly shown that individuals with abacus Attention
training have higher levels of digit span ability, because as
they can rely on abacus-based representations in addition to Several attention indicators were evaluated in the pre- and
phonological representations (Hatano & Osawa, 1983; Hu post-test assessments, including visual tracking, UFOV atten-
et al., 2011; Tanaka et al., 2002). The current results showed tion, focused attention, distributed attention, and geometric
that all three groups improved on backward digit span test form searching. Post-hoc analysis showed that in the post-test,
Curr Psychol

the focused attention scores in the abacus and cognitive train- on inductive reasoning was observed when compared with
ing groups were significantly higher than those in the English cognitive or English training in this study. A possible expla-
training group. The abacus training did not involve attention nation for this is that the mental abacus calculation emphasizes
training directly, but they have similar transfer effects for fo- mental manipulation and imaginary rather than abstract or
cused attention when compared with the English training. inductive reasoning.
Previous studies have also shown that abacus training requires
participants to maintain high levels of attention and involves Limitations
the allocation of visual resources, which can enhance chil-
dren’s attention orientation ability (Liu & Sun, 2017) and The present study also had some limitations. First, we only
sustained attention (Na et al., 2015) abilities. We conclude utilized two active control groups, and did not include a blank
that the two-month, short-term abacus training exerted certain control group (i.e., one with no training intervention). This
effects on attention ability, especially in comparison with made it difficult to assess the real effects of the three types
English training. This result, which was not included in our of interventions. Indeed, previous abacus intervention studies
hypothesis, complements our work regarding far-reaching have used a blank control group and observed significant en-
transfer effects of attention for abacus training. hancement due to abacus training, including on memory (e.g.,
Bhaskaran et al., 2006), matrix reasoning (e.g., Jiang, 2012),
Working Memory Ability and even language processing (Jiang, 2009, 2012) tests. The
effects may be from placebo effect or inflated estimates of the
No larger transfer effects of abacus training were observed in intervention effect (Cunningham et al., 2013; Simons et al.,
spatial N-back and short-term memory tests. This is inconsis- 2016). This is why we focused on active groups in the current
tent with previous studies where abacus training effects were study. Second, the post-test was completed within one week
observed for working memory (e.g., Amaiwa & Hatano, after the completion of the training, and the persistence of the
1989; Bhaskaran et al., 2006; Hatano et al., 1977; Liu et al., transfer effects was not further explored. Future studies may
2006; Liu & Sun, 2017; Sun et al., 2006; Tanaka et al., 2002; include a delayed post-test design.
Wang et al., 2015; Zhen et al., 2007). Some researchers have
proposed that abacus training improves the working memory
for letters or words (e.g., Hu et al., 2011) and visuospatial
materials (e.g., Wang et al., 2015; Zhen et al., 2007; Zhou Conclusion
et al., 2019). However, post-analysis studies have focused
on digital spans in alphabets or fruit names (Hatano & The results of this work revealed a limited transfer effects of
Osawa, 1983); studies with pre-post-tests focused on spatial abacus training for numerical abilities (near transfer) and spa-
working memory (Barner et al., 2016) showed no significant tial abilities (far-reaching transfer). Our results also suggested
working memory improvements, arguing that it is the baseline that mental abacus calculation is a spatial computation skill
working memory capacity that plays the most essential role in that strongly relies on visual imagery, thus promoting the
mental abacus training effects. trainees’ calculation skills (on Arabic number), memory for
In this study, we balanced the baseline working memory digits, and spatial abilities that involve active manipulation
abilities in the three training groups to find that the abacus rather than passive spatial memorization.
training group demonstrated no significant improvements on
any visual and spatial working memory indicators. Therefore,
it is highly likely that previous abacus training effects are
attributable to baseline working memory capacity rather than Appendix
the abacus training itself. Other studies with pre-post designs
should account for baseline working memory ability apart Descriptions of the Cognitive Tests
from nonverbal intelligence as a key factor influencing train-
ing effects. Choice Reaction Time

Inductive Reasoning This test was used to measure processing speed (Butterworth,
2003). In each trial, a white dot was presented on a black
No larger transfer effects of abacus training were observed in screen to the left or right of a white “+” without time limita-
the inductive reasoning indicators, which is consistent with a tion. Participants were asked to determine whether white dots
previous study that termed reasoning ability as, simply, “in- appeared on the left or right side of the “+” on their computer
telligence” (Wang et al., 2019). It is worth noting that neither screen. The interval between the response and the onset of the
significant promotion nor degradation of the abacus training next trial varied randomly between 1500 and 3000 ms. The
Curr Psychol

test contained 30 random presentation trials, and reaction 360 random presentation trials and was limited to 4 min. The
times were recoded. adjusted number of correct trials was recorded.

Numerosity Comparison Two-dimensional Mental Rotation

This test was used to measure number sense ability (Rodic This test was designed to evaluate two-dimensional (2D) spa-
et al., 2018; Zhang et al., 2019). In each trial, two dot arrays tial ability (Collins & Kimura, 1997). In each trial, one irreg-
were simultaneously presented on the screen, lasting for ular shape was presented at the top of screen and two irregular
200 ms. Participants were asked to judge which dot array shapes were presented at the bottom. Participants were asked
contained more dots while ignoring the sizes of individual to choose the figure from the alternatives that would create a
dots. The number of dots in each set varied from 5 to 32. complete rectangle with the figure at the top after mentally
The interval between the response and the onset of the next rotating it. This test had 120 random presentation trials and
trial was 1000 ms. There were three blocks administered with was limited to 4 min. The adjusted number of correct trials
40 trials in each block. The proportion of correct trials was was recorded.
calculated.
Three-dimensional Mental Rotation
Simple Subtraction
This test was adapted from Shepard and Metzler’s mental
This test was used to measure simple calculation ability
rotation task (Shepard & Metzler, 1971), and was used to
(Wang et al., 2016). In each trial, a subtraction equation was
assess three-dimensional (3D) spatial ability. In each trial,
presented at the top of screen and two answers were presented
one 3D image was presented at the top of screen, with two
at the bottom of screen with no time limitation. Participants
analogues at the bottom. Participants were asked to choose the
were asked to select the correct answer from the two alterna-
figure from the alternatives that would match the target figure
tives. All minuends were no more than 18, and the answers
after mentally rotating it. The angles of rotation ranged from
were single-digit numbers, (e.g., 7–2, 15–8). The differences
15° to 345° with a step of 15°. This test had 180 random
between false and true answers were 1 to 3. The test had 92
presentation trials and was limited to 3 min. The adjusted
random presentation trials and was limited to 2 min. The ad-
number of correct trials was recorded.
justed number of correct trials was calculated as the difference
between the numbers of correct responses and incorrect re-
sponses to control for guessing effect (Cirino, 2011). Paper Folding Task

Complex Subtraction This test was used to measure spatial visualization ability
(Harris et al., 2013; Shepard & Feng, 1972). In each trial,
This test was used to measure complex calculation ability two parallel rows of figures were presented on the screen with
(Zhou et al., 2015). In each trial, one equation was presented no time limitation. The first row showed the folding and
at the top of screen and two answers were presented at the punching process of the paper, and the second row showed
bottom of screen with no time limitation. Participants were five alternative items. Participants were asked to imagine how
asked to choose correct answer from the two alternatives. the piece of paper would appear once unfolded and select the
All minuends were larger than 20 and smaller than 98, with correct figure from the alternatives given. This test had 18
the subtractors ranging from 10 to 88 (e.g., 26–14, 91–33). random presentation trials and was limited to 4 min. The num-
The differences between false and true answers were 1 to 10. ber of correct trials was recorded.
This test had 96 random presentation trials and was limited to
2 min. The adjusted number of correct trials was recorded. Visual Tracking

Multi-step Computation This test was adapted from the task designed by Groffman
(1966), and was used to assess visual attention ability. In each
This test was used to measure multi-step calculation ability. trial, several curved lines were interwoven within a square,
Again, in each trial, one equation was presented at the top of starting from the left side of the square and ending on the right
screen and two answers were presented at the bottom of screen side. Participants were asked to track three particular lines
without time limitation. Participants asked to select the correct from the beginning to the end of the square using only their
answer from the two alternatives. All problems included four vision and then to mark the correct end point. The total num-
one-digit addition or subtraction (e.g., 7–3 + 5–4). The differ- ber of lines increased from 8 to 12. The test was limited to
ences between false and true answers were 1 to 2. This test had 4 min. The number of correct trials was recorded.
Curr Psychol

UFOV Attention reached the top. And new row of graphics appeared every
3000 ms. Participants were asked to use the mouse to click
This test was adapted from the useful field of view (UFOV) the graphic that contained both a circle and a square as quickly
task, and was used to measure spatial attention ability (Ball as possible. This test had 252 random presentation rows and
et al., 1988). In each trial, a bright outlined box was presented was limited to 4 min (or terminated after 50 missed targets had
in the center of the screen (1000 ms); then a series of other accumulated). The number of correct trials was recorded.
boxes, one of which showed a circle containing a triangle (the
target box) presented for 50 ms; then a mask was presented for Digit Span
750 ms. Finally, a radial pattern appeared with eight equally
spaced spokes with no time limit. Participants clicked on the Short-term memory ability for digits (Wechsler, 1974) was
target position on the line segment as accurately as possible. tested by presenting a series of digits aurally through
This test had 48 random presentation trials. Accuracy was earphones. Participants were asked to remember the order of
calculated as follows: the digits and type them into the computer at the end of each
series in the forward condition, or enter them in reverse order
jresponse−standard answerj
Accuracy ¼ 100−  100; in the backward condition. The test began with three digits
standard answer þ jresponse−standard answerj
and increased gradually until the children failed to key them
where “response” refers to the location where participants properly three consecutive times. The number of correctly
clicked, and “standard answer” refers to the correct location recalled digits was recorded.
of the target box.
Spatial 2-back
Focused Attention
This test was used to measure spatial working memory ability
This test was adapted from Franceschini et al. (2013), and was (Dong et al., 2016). A 3 × 3 matrix with nine black squares
used to measure focused attention ability. In each trial, a fix- was presented on the center of screen. Each trial began with a
ation point (1000 ms) appeared before the onset of the six fixation point (3000 ms) followed by 11 consecutively pre-
successive stimuli. A red dot focused participants’ attention sented white squares randomly filled into one of the nine
on the target location for 34 ms, then a row of six meaningless squares in the matrix. Each of them appeared for 500 ms with
symbols was presented (150 ms) and participants were asked an inter-stimulus interval of 2500 ms. Participants were asked
to remember the symbol that appeared at the target location. A to memorize the location of each white square, and to succes-
blank screen with fixation point was presented then for sively judge whether the location of the current white square
100 ms, followed by a post-mask for 50 ms and another blank was same as that of the white square presented two steps back
screen with fixation point was then presented for 1000 ms. (e.g., whether the location of the third white square was same
Finally, participants were asked to identify the target as quick- as that of the first). There were a total of 10 trials. The hit
ly as possible with no stated time limit. This test had 48 ran- minus false alarm was recorded.
dom presentation trials. The number of correct trials was
recorded. Spatial 4-back

Spatial working memory (Dong et al., 2016) was tested via


Distributed Attention spatial 4-back test. Unlike the spatial 2-back test, participants
were asked to successively judge whether the location of a
This test was also adapted from Franceschini et al. (2013), and current white square was same as that of the white square
was used to evaluate distributed attention ability. The stimuli presented four steps back (e.g., whether the location of the
and procedure were same as that in focused attention task, fifth black square was same with as of the first). There were
excepted for the presentation order of the red dot indicating a total of 10 trials with 13 consecutively presented white
the target location. In this test, the red dot appeared after a row squares in each trial. The hit minus false alarm was recorded.
of six meaningless symbols. This test had 48 random presen-
tation trials. The number of correct trials was recorded. Spatial Short-term Memory

Geometric Form Searching This test was adapted from Corsi block task (Corsi, 1972), and
was used to measure spatial short-term memory. In each trial,
This test was used to evaluate attention ability (Lu et al., a series of dots was sequentially presented on the computer
2021). Ten graphics were presented in each row, moving from screen each for 1000 ms with a 1000 ms blank-screen inter-
the bottom upward on the screen and disappearing once they stimulus interval. Participants were asked to sequentially click
Curr Psychol

the position of each dot as they it in the same order as the dot of the institutional and/or national research committee and with the 1964
Helsinki declaration and its later amendments or comparable ethical
had appeared. This test had 10 trials, and the number of dots in
standards.
each trial increased from 3 to 7 (two times for each number of
dots). The accuracy was calculated using the same formula as Informed Consent Informed consent was obtained from all individual
the UFOV attention task. participants included in the current study.

Conflict of Interest On behalf of all authors, the corresponding author


Visual Short-term Memory
states that there is no conflict of interest.

This test was used to evaluate short-term memory (Zhang


et al., 2019). In each trial, four irregular shapes consecutively
References
presented on the screen. Each of them presented for 500 ms,
with an inter-stimulus interval of 300 ms. Participants were Amaiwa, S., & Hatano, G. (1989). Effects of abacus learning on 3rd-
asked to judge whether the fourth shape had been presented in graders’ performance in paper-and-pencil tests of calculation.
the previous series without time limit. There were 2 blocks Japanese Psychological Research, 31(4), 161–168. https://doi.org/
10.4992/psycholres1954.31.161.
with 40 trials in each block. The proportion of correct trial was
Anguera, J. A., Boccanfuso, J., Rintoul, J. L., Al-Hashimi, O., Faraji, F.,
recorded. Janowich, J., Kong, E., Larraburo, Y., Rolle, C., Johnston, E., &
Gazzaley, A. (2013). Video game training enhances cognitive con-
Sentence Completion trol in older adults. Nature, 501(7465), 97–101. https://doi.org/10.
1038/nature12486.
Atit, K., Shipley, T. F., & Tikoff, B. (2013). Twisting space: Are rigid and
This test assessed language ability (Coltheart et al., 2001; Cui non-rigid mental transformations separate spatial skills? Cognitive
et al., 2019). In each trial, a sentence with one word missing Processing, 14(2), 163–173. https://doi.org/10.1007/s10339-013-
was presented. Participants were asked to select the missing 0550-8.
Ball, K. K., Beard, B. L., Roenker, D. L., Miller, R. L., & Griggs, D. S.
word to complete the sentence from two alternatives. This test
(1988). Age and visual search: Expanding the useful field of view.
had 120 trials, and was limited to 5 min. The adjusted number Journal of the Optical Society of America A, 5(12), 2210–2219.
of correct trials was recorded. https://doi.org/10.1364/josaa.5.002210.
Barner, D., Alvarez, G., Sullivan, J., Brooks, N., Srinivasan, M., & Frank,
M. C. (2016). Learning mathematics in a visuospatial format: A
Nonverbal Matrix Reasoning randomized, controlled trial of mental abacus instruction. Child
Development, 87(4), 1146–1158. https://doi.org/10.1111/cdev.
This test was a short version of nonverbal intelligence tasks 12515.
that adapted from Raven’s Progressive Matrices (Raven, Barner, D., Athanasopoulou, A., Chu, J., Lewis, M., Marchand, E.,
Schneider, R., & Frank, M. (2017). A one-year classroom-random-
1998), and was used to measure reasoning ability. ized trial of mental abacus instruction for first- and second-grade
Participants were asked to identify the missing segment of a students. Journal of Numerical Cognition, 3(3), 540–558. https://
figure according to its inherent regularity from 6 or 8 candi- doi.org/10.5964/jnc.v3i3.106.
date answers. This test had 60 random presentation trials and Berninger, V. W., & Abbott, R. D. (2010). Listening comprehension, oral
expression, reading comprehension, and written expression: Related
was limited to 10 min. The number of correct trials was
yet unique language systems in grades 1, 3, 5, and 7. Journal of
recorded. Educational Psychology, 102(3), 635–651. https://doi.org/10.1037/
a0019319.
Bhaskaran, M., Sengottaiyan, A., Madhu, S., & Ranganathan, V. (2006).
Supplementary Information The online version contains supplementary Evaluation of memory in abacus learners. Indian Journal of
material available at https://doi.org/10.1007/s12144-021-01968-1. Physiology and Pharmacology, 50(3), 225–233.
Jaeggi, S. M, Buschkuehl, M., Jonides, J., & Shah, P. (2011). Short- and
Availability of Data and Material The datasets generated during and long-term benefits of cognitive training. Proceedings of the National
analysed during the current study are available from the corresponding Academy of Sciences, 108(25), 10081–10086. https://doi.org/10.
author on reasonable request. 1073/pnas.1103228108.
Code Availability Not applicable. Butterworth, B. (2003). Dyscalculia screener. NFER-Nelson (Software
and Manual).
Funding This research was supported by Advanced Innovation Center Carr, M., & Alexeev, N. (2011). Fluency, accuracy, and gender predict
for Future Education (No. 27900–110631111), the 111 Project (No. developmental trajectories of arithmetic strategies. Journal of
BP0719032), and National Natural Science Foundation of China (Nos. Educational Psychology, 103(3), 617–631. https://doi.org/10.1037/
31600896, 31671151). a0023864.
Chen, F., Hu, Z., Zhao, X., Wang, R., Yang, Z., Wang, X., & Tang, X.
(2006). Neural correlates of serial abacus mental calculation in chil-
Declarations dren: A functional MRI study. Neuroscience Letters, 403(1–2), 46–
51. https://doi.org/10.1016/j.neulet.2006.04.041.
Ethical Approval All procedures performed in the current studies in- Cheng, D., Xiao, Q., Cui, J., Chen, C., Zeng, J., Chen, Q., & Zhou, X.
volving human participants were in accordance with the ethical standards (2019). Short-term numerosity training promotes symbolic
Curr Psychol

arithmetic in children with developmental dyscalculia: The mediat- Frank, M. C., & Barner, D. (2012). Representing exact number visually
ing role of visual form perception. Developmental Science, 23, using mental abacus. Journal of Experipemtal Psychology: General,
e12910. https://doi.org/10.1111/desc.12910. 141(1), 134–149. https://doi.org/10.1037/a0024427.
Cherney, I. D. (2008). Mom, let me play more computer games: They Geeta, N., & Gavas, R. D. (2014). Enhanced learning with abacus and its
improve my mental rotation skills. Sex Roles, 59(11), 776–786. analysis using BCI technology. International Journal of Modern
https://doi.org/10.1007/s11199-008-9498-z. Education and Computer Science, 6(9), 22–27. https://doi.org/10.
Cherney, I. D., Bersted, K., & Smetter, J. (2014). Training spatial skills in 5815/ijmecs.2014.09.04.
men and women. Perceptual and Motor Skills, 119(1), 82–99. Groffman, S. (1966). Visual tracing. Journal of the American Optometric
https://doi.org/10.2466/23.25.PMS.119c12z0. Association, 37, 139–141.
Cirino, P. T. (2011). The interrelationships of mathematical precursors in Hanakawa, T., Honda, M., Okada, T., Fukuyama, H., & Shibasaki, H.
kindergarten. Journal of Experimental Child Psychology, 108(4), (2003). Neural correlates underlying mental calculation in abacus
713–733. https://doi.org/10.1016/j.jecp.2010.11.004. experts: A functional magnetic resonance imaging study.
Collins, D. W., & Kimura, D. (1997). A large sex difference on a two- Neuroimage, 19(2), 296–307. https://doi.org/10.1016/S1053-
dimensional mental rotation task. Behavioral Neuroscience, 111(4), 8119(03)00050-8.
845–849. https://doi.org/10.1037/0735-7044.111.4.845. Harris, J., Newcombe, N. S., & Hirsh-Pasek, K. (2013). A new twist on
studying the development of dynamic spatial transformations:
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001).
Mental paper folding in young children. Mind, Brain, and
DRC: A dual route cascaded model of visual word recognition and
Education, 7(1), 49–55. https://doi.org/10.1111/mbe.12007.
reading aloud. Psychological Review, 108(1), 204–256. https://doi.
Hatano, G., Miyake, Y., & Binks, M. G. (1977). Performance of expert
org/10.1037//0033-295x.108.1.204.
abacus operators. Cognition, 5(1), 47–55. https://doi.org/10.1016/
Corsi, P. M. (1972). Human memory and the medial temporal region of 0010-0277(77)90016-6.
the brain. McGill University. Hatano, G., & Osawa, K. (1983). Digit memory of grand experts in
Cui, J., Xiao, R., Ma, M., Yuan, L., Cohen Kodash, R., & Zhou, X. abacus-derived mental calculation. Cognition, 15(1–3), 95–110.
(2020). Children skilled in mental abacus show enhanced non- https://doi.org/10.1016/0010-0277(83)90035-5.
symbolic number sense. Current Psychology. https://doi.org/10. Hu, Y., Geng, F., Tao, L., Hu, N., Du, F., Fu, K., & Chen, F. (2011).
1007/s12144-020-00717-0. Enhanced white matter tracts integrity in children with abacus train-
Cui, J., Zhang, Y., Wan, S., Chen, C., Zeng, J., & Zhou, X. (2019). Visual ing. Human Brain Mapping, 32(1), 10–21. https://doi.org/10.1002/
form perception is fundamental for both reading comprehension and hbm.20996.
arithmetic computation. Cognition, 189, 141–154. https://doi.org/ Jia, X., Zhang, Y., Yao, Y., Chen, F., & Liang, P. (2021). Neural corre-
10.1016/j.cognition.2019.03.014. lates of improved inductive reasoning ability in abacus-trained chil-
Cunningham, J. A., Kypri, K., & McCambridge, J. (2013). Exploratory dren: A resting state fMRI study. PsyCh Journal. https://doi.org/10.
randomized controlled trial evaluating the impact of a waiting list 1002/pchj.439.
control design. BMC Medical Research Methodology, 13(150), 1–7. Jiang, Z. (2009). Research results and policy recommendations of the
https://doi.org/10.1186/1471-2288-13-150. project “research on the effect of abacus mental arithmetic education
Ding, Y., Liu, R.-D., Liu, H., Wang, J., Zhen, R., & Jiang, R.-H. (2019). on the development of children’s intellectual potential”. Abacus and
Effects of working memory, strategy use, and single-step mental Mental Abacus, 2, 15–24.
addition on multi-step mental addition in Chinese elementary stu- Jiang, Z. (2012). Case study on the influence of abacus and mental arith-
dents. Frontiers in Psychology, 10(148). https://doi.org/10.3389/ metic education intervention to intellectual development of mono-
fpsyg.2019.00148. zygotic twins of grades 1-3 in primary school. Abacus and Mental
Dong, S., Wang, C., Xie, Y., Hu, Y., Weng, J., & Chen, F. (2016). The Abacus, 6, 6–14.
impact of abacus training on working memory and underlying neu- Kawakami, A. (1995). Digit and shape memory function in soroban
ral correlates in young adults. Neuroscience, 332, 181–190. https:// learners: Are Soroban learners superior in general? Perceptual and
doi.org/10.1016/j.neuroscience.2016.06.051. Motor Skills, 81(1), 75–80. https://doi.org/10.2466/pms.1995.81.1.
Du, F., Chen, F., Li, Y., Hu, Y., Tian, M., & Zhang, H. (2013). Abacus 75.
training modulates the neural correlates of exact and approximate Ku, Y., Hong, B., Zhou, W., Bodner, M., & Zhou, Y.-D. (2012).
calculations in Chinese children: An fMRI study. BioMed Research Sequential neural processes in abacus mental addition: An EEG
International, 2013, 1–12. https://doi.org/10.1155/2013/694075. and fMRI case study. PLoS One, 7(5), e36410. https://doi.org/10.
Du, F., Yao, Y., Zhang, Q., & Chen, F. (2014). Long-term abacus training 1371/journal.pone.0036410.
induces automatic processing of abacus numbers in children. Lee, Y., Lu, M., & Ko, H. (2007). Effects of skill training on working
Perception, 43(7), 694–704. https://doi.org/10.1068/p7625. memory capacity. Learning and Instruction, 17(3), 336–344. https://
Fan, K., Dai, Y., & Liu, L. (2005). Effect of abacus and mental arithmetic doi.org/10.1016/j.learninstruc.2007.02.010.
on children’s intelligent development. China Journal of Health Liu, X., Li, X., Li, C., Gao, W., Yang, R., & Yong, A. (2006). Effect of
Psychology, 3, 186–188. mental abacus calculation on cognitive capacity for pupils. Chinese
Journal of Behavioral Medical Science, 11, 1031–1032.
Fellman, D., Jylkkä, J., Waris, O., Soveri, A., Ritakallio, L., Haga, S.,
Liu, X., & Sun, Y. (2017). An event-related potential investigation of
Salmi, J., Nyman, T. J., & Laine, M. (2020). The role of strategy use
spatial attention orientation in children trained with mental abacus
in working memory training outcomes. Journal of Memory and
calculation. Neuroreport, 28(1), 35–41. https://doi.org/10.1097/
Language, 110, 104064. https://doi.org/10.1016/j.jml.2019.104064.
WNR.0000000000000705.
Forsberg, A., Fellman, D., Laine, M., Johnson, W., & Logie, R. H.
Lu, Y., Ma, M., Chen, G., & Zhou, X. (2021). Can abacus course erad-
(2020). Strategy mediation in working memory training in younger
icate developmental dyscalculia. Psychology in the Schools, 58(2),
and older adults. Quarterly Journal of Experimental Psychology,
235–251. https://doi.org/10.1002/pits.22441.
73(8), 1206–1226. https://doi.org/10.1177/1747021820915107.
Matías-Guiu, J. A., Pérez-Martínez, D. A., & Matías-Guiu, J. (2016). A
Franceschini, S., Gori, S., Ruffino, M., Viola, S., Molteni, M., & Facoetti, pilot study of a new method of cognitive stimulation using abacus
A. (2013). Action video games make dyslexic children read better. arithmetic in healthy and cognitively impaired elderly subjects.
Current Biology, 23(6), 462–466. https://doi.org/10.1016/j.cub. Neurología (English Edition), 31(5), 326–331. https://doi.org/10.
2013.01.044. 1016/j.nrleng.2015.02.003.
Curr Psychol

Na, K. S., Lee, S., Park, J. H., Jung, H. Y., & Ryu, J. H. (2015). underlying neural correlates in Chinese children. Human Brain
Association between abacus training and improvement in response Mapping, 38(10), 5234–5249. https://doi.org/10.1002/hbm.23728.
inhibition: A case-control study. Clinical Psychopharmacology and Wang, C., Xu, T., Geng, F., Hu, Y., Wang, Y., Liu, H., & Chen, F.
Neuroscience, 13(2), 163–167. https://doi.org/10.9758/cpn.2015. (2019). Training on abacus-based mental calculation enhances vi-
13.2.163. suospatial working memory in children. The Journal of
Raven, J. (1998). Manual for Raven’s progressive matrices and vocabu- Neuroscience, 39(33), 6439–6448. https://doi.org/10.1523/
lary scales. Oxford Psychologists Press. jneurosci.3195-18.2019.
Rodic, M., Cui, J. X., Malykh, S., Zhou, X. L., Gynku, E. I., Bogdanova, Wang, Y., Geng, F., Hu, Y., Du, F., & Chen, F. (2013). Numerical
E. L., et al. (2018). Cognition, emotion, and arithmetic in primary processing efficiency improved in experienced mental abacus chil-
school: A cross-cultural investigation. British Journal of dren. Cognition, 127(2), 149–158. https://doi.org/10.1016/j.
Developmental Psychology, 36(2), 255–276. https://doi.org/10. cognition.2012.12.004.
1111/bjdp.12248. Wang, L., Sun, Y., & Zhou, X. (2016). Relation between Approximate
Ruiz-Marquez, E., Prieto, A., Mayas, J., Toril, P., Reales, J. M., & Number System Acuity and Mathematical Achievement: The
Ballesteros, S. (2019). Effects of nonaction videogames on attention Influence of Fluency. Frontiers in Psychology, 7(1966). https://
and memory in young adults. Games for Health Journal, 8(6), 414– doi.org/10.3389/fpsyg.2016.01966.
422. https://doi.org/10.1089/g4h.2019.0004. Wechsler, D. (1974). Manual for the Wechsler intelligence scale for
Shepard, R. N., & Feng, C. (1972). Chronometric study of mental paper children-Revised. San Antonio, TX: The Psychological
folding. Cognitive Psychology, 3(2), 228–243. https://doi.org/10. Corporation.
1016/0010-0285(72)90005-9. Wei, W., Yuan, H., Chen, C., & Zhou, X. (2012). Cognitive correlates of
Shepard, R. N., & Metzler, J. (1971). Mental rotation of 3-dimensional performance in advanced mathematics. British Journal of
objects. Science, 171(3972), 701–703. https://doi.org/10.1126/ Educational Psychology, 82(1), 157–181. https://doi.org/10.1111/j.
science.171.3972.701. 2044-8279.2011.02049.x.
Simons, D. J., Boot, W. R., Charness, N., Gathercole, S. E., Chabris, C. Yao, Y., Du, F., Wang, C., Liu, Y., Weng, J., & Chen, F. (2015).
F., Hambrick, D. Z., & Stine-Morrow, E. A. L. (2016). Do “brain- Numerical processing efficiency improved in children using mental
training” programs work? Psychological Science in the Public abacus: ERP evidence utilizing a numerical Stroop task. Frontiers in
Interest, 17(3), 103–186. https://doi.org/10.1177/ Human Neuroscience, 9, 245. https://doi.org/10.3389/fnhum.2015.
1529100616661983. 00245.
Stigler, J. W. (1984). “Mental abacus”: The effect of abacus training on Zhang, Y., Liu, T., Chen, C., & Zhou, X. (2019). Visual form perception
Chinese children’s mental calculation. Cognitive Psychology, 16(2), supports approximate number system acuity and arithmetic fluency.
145–176. https://doi.org/10.1016/0010-0285(84)90006-9. Learning and Individual Differences, 71, 1–12. https://doi.org/10.
Sun, Y., Gao, W., Li, X., Liu, X., & Li, C. (2006). Effect of abacus mental 1016/j.lindif.2019.02.008.
calculation training on children cognition ability. China Journal of Zhen, Y., Li, X., & Li, P. (2007). The effects of practices for the abacus
Health Psychology, 6, 714–715. mental calculation on children ‘s digital memory. China Journal of
Takeuchi, H., & Kawashima, R. (2012). Effects of processing speed Health Psychology, 11, 1024–1025.
training on cognitive functions and neural systems. Reviews in the Zhou, H., Geng, F., Wang, Y., Wang, C., Hu, Y., & Chen, F. (2019).
Neurosciences, 23(3), 289–301. https://doi.org/10.1515/revneuro- Transfer effects of abacus training on transient and sustained brain
2012-0035. activation in the frontal-parietal network. Neuroscience, 408, 135–
Tanaka, S., Michimata, C., Kaminaga, T., Honda, M., & Sadato, N. 146. https://doi.org/10.1016/j.neuroscience.2019.04.001.
(2002). Superior digit memory of abacus experts: An event-related Zhou, X., Wei, W., Zhang, Y., Cui, J., & Chen, C. (2015). Visual per-
functional MRI study. Neuroreport, 13(17), 2187–2191. https://doi. ception can account for the close relation between numerosity pro-
org/10.1097/00001756-200212030-00005. cessing and computational fluency. Frontiers in Psychology, 6.
Uttal, D. H., Meadow, N. G., Tipton, E., Hand, L. L., Alden, A. R., https://doi.org/10.3389/fpsyg.2015.01364.
Warren, C., & Newcombe, N. S. (2013). The malleability of spatial Ziegler, D. A., Simon, A. J., Gallen, C. L., Skinner, S., Janowich, J. R.,
skills: A meta-analysis of training studies. Psychological Bulletin, Volponi, J. J., Rolle, C. E., Mishra, J., Kornfield, J., Anguera, J. A.,
139(2), 352–402. https://doi.org/10.1037/a0028446. & Gazzaley, A. (2019). Closed-loop digital meditation improves
Wai, J., Lubinski, D., & Benbow, C. P. (2009). Spatial ability for STEM sustained attention in young adults. Nature Human Behaviour,
domains: Aligning over 50 years of cumulative psychological 3(7), 746–757. https://doi.org/10.1038/s41562-019-0611-9.
knowledge solidifies its importance. Journal of Educational Zhou, H., Geng, F., Wang, T., Wang, C., Xie, Y., Hu, Y., & Chen, F.
Psychology, 101(4), 817–835. https://doi.org/10.1037/a0016127. (2020). Training on abacus-based mental calculation enhances rest-
Wang, C., Geng, F., Yao, Y., Weng, J., Hu, Y., & Chen, F. (2015). ing state functional connectivity of bilateral superior parietal lobules.
Abacus training affects math and task switching abilities and mod- Neuroscience, 432, 115–125. https://doi.org/10.1016/j.
ulates their relationships in Chinese children. PLoS One, 10(10), neuroscience.2020.02.033.
e0139930. https://doi.org/10.1371/journal.pone.0139930.
Wang, C., Weng, J., Yao, Y., Dong, S., Liu, Y., & Chen, F. (2017). Effect Publisher’s Note Springer Nature remains neutral with regard to jurisdic-
of abacus training on executive function development and tional claims in published maps and institutional affiliations.

You might also like