Zarnhofer et al.

Behavioral and Brain Functions 2012, 8:13

http://www.behavioralandbrainfunctions.com/content/8/1/13

RESEARCH Open Access

The Influence of verbalization on the pattern of

cortical activation during mental arithmetic
Sabrina Zarnhofer1*, Verena Braunstein1, Franz Ebner2, Karl Koschutnig2,4, Christa Neuper3,4, Gernot Reishofer2 and
Anja Ischebeck1

Abstract
Background: The aim of the present functional magnetic resonance imaging (fMRI) study at 3 T was to investigate
the influence of the verbal-visual cognitive style on cerebral activation patterns during mental arithmetic. In the
domain of arithmetic, a visual style might for example mean to visualize numbers and (intermediate) results, and a
verbal style might mean, that numbers and (intermediate) results are verbally repeated. In this study, we
investigated, first, whether verbalizers show activations in areas for language processing, and whether visualizers
show activations in areas for visual processing during mental arithmetic. Some researchers have proposed that the
left and right intraparietal sulcus (IPS), and the left angular gyrus (AG), two areas involved in number processing,
show some domain or modality specificity. That is, verbal for the left AG, and visual for the left and right IPS. We
investigated, second, whether the activation in these areas implied in number processing depended on an
individual’s cognitive style.
Methods: 42 young healthy adults participated in the fMRI study. The study comprised two functional sessions. In
the first session, subtraction and multiplication problems were presented in an event-related design, and in the
second functional session, multiplications were presented in two formats, as Arabic numerals and as written
number words, in an event-related design. The individual’s habitual use of visualization and verbalization during
mental arithmetic was assessed by a short self-report assessment.
Results: We observed in both functional sessions that the use of verbalization predicts activation in brain areas
associated with language (supramarginal gyrus) and auditory processing (Heschl’s gyrus, Rolandic operculum).
However, we found no modulation of activation in the left AG as a function of verbalization.
Conclusions: Our results confirm that strong verbalizers use mental speech as a form of mental imagination more
strongly than weak verbalizers. Moreover, our results suggest that the left AG has no specific affinity to the verbal
domain and subserves number processing in a modality-general way.
Keywords: fMRI, Cognitive styles, Number processing, Gyrus supramarginalis, Heschl’s gyrus, Rolandic operculum

Background cognitive styles because this distinction has repeatedly

As individuals differ with regard to their potentials, been used to categorize solution alternatives, ability pro-
habits and preferences, they differ in their approach to files and processing differences in mathematics and arith-
complex problems, and the degree to which they employ metic processing [1-3].
different cognitive domains in solving a complex pro-
blem. It is unclear, however, how individual preferences The visual and verbal cognitive style
in problem solving are reflected in the brain. We will A cognitive style is assumed to be a relatively stable char-
focus here on the distinction between visual and verbal acteristic that describes how an individual processes
information, that is, how an individual thinks, perceives,
and remembers [4]. Verbalizers report to repeat informa-
* Correspondence: [email protected]
1
Cognitive Psychology and Neuroscience, Department of Psychology, tion during thinking verbally, whereas visualizers claim
University of Graz, Universitaetsplatz 2/III, 8010 Graz, Austria to represent information during thinking pictorially or
Full list of author information is available at the end of the article

© 2012 Zarnhofer et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 2 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

schematically. In the domain of arithmetic, a visual style a verbal cognitive style could have a tendency to convert
might mean to visualize numbers and (intermediate) pictorially presented information into a verbal mental
results while calculating or to move mentally along the representation. These studies suggest that general differ-
mental number line. A visual style might also mean for ences between verbalizers and visualizers exist with
example imagining a mass that increases or decreases in regard to brain activation patterns, but it is yet unclear
magnitude, or some collection of dots that grows or whether these individual differences also affect brain
shrinks in number. In the domain of arithmetic, a verbal areas specifically involved in arithmetic number
style might mean that numbers and (intermediate) results processing.
are verbally repeated during calculation.
The visual and verbal cognitive style was first conceptua- The neural basis of number processing and arithmetic
lized as a bipolar construct with the preference to visual or In recent years, research with imaging and other neuros-
verbal ways of information processing portrayed as two cientific methods has greatly advanced our understanding
contrasting poles. Although this categorization of indivi- of the neural basis of number processing. Humans as well
duals into visualizers and verbalizers is intuitively convin- as animals were found to have a basic concept of number
cing, the independence of brain areas subserving verbal and order, and they can roughly estimate [9]. Results from
and visual processing makes it implausible to assume that fMRI studies [10,11] as well as single unit recordings in
good visualizers have to be bad verbalizers and vice versa, monkeys [12] have shown that the anterior part of the
as implied in the one-dimensional model of cognitive intraparietal sulcus (IPS), bilaterally, is crucially involved
styles. Accordingly, Blazhenkova and Kozhevnikov [5] sug- in the processing of numbers. Specifically, some research-
gested independence of visualization and verbalization ers have proposed that the IPS hosts an amodal represen-
during information processing. A given individual might tation of quantity, which has been referred to as the
as well be visualizing as verbalizing during information ‘mental number line’ [13,14]. In arithmetic, the IPS plays a
processing but might show a preference for visualization role for example in solving subtraction problems. Whereas
or verbalization. other researchers have proposed that the IPS shows some
To our knowledge, however, there are only two fMRI modality specificity as the IPS also plays a role in visual
studies so far that investigated the impact of visual and attention [15-17], and in visual-spatial short-term memory
verbal cognitive styles on cognitive performance and [18,19]. Dehaene and colleagues [20], however, proposed
brain activation. An fMRI study by Burbaud et al. [6] that the IPS is not domain-specific but rather amodal and
using an arithmetic task found brain activation differ- hence not necessarily a visual processing area. Another
ences between visualizers and verbalizers in areas brain area important for arithmetic problem solving is the
involved in verbal and visual processing, but not in brain left angular gyrus (AG), which is assumed to support the
areas associated with number processing. It should be long term memory retrieval for arithmetic fact knowledge
noted that the authors assessed cognitive styles using a [20]. Arithmetic fact knowledge is required, for example,
one-dimensional 5-point rating scale (1...purely verbal/5... in the skilled solving of multiplication problems by retriev-
purely visual). The authors assumed thus the one-dimen- ing the result from verbal long-term memory, namely,
sional model with two contrasting poles of cognitive from the multiplication tables learned in childhood. This
styles as the basis for their study. As described above, the has lead researchers to believe that the AG might show
one-dimensional model was replaced by a multi-dimen- some affinity to the verbal domain [20]. Some recent find-
sional model, and the results of the study of Burbaud et ings, however, raise some doubts about the assumption
al. should hence be interpreted with great care. A that the left AG mediates verbal fact retrieval during mul-
recently published fMRI study by Kraemer et al. [7] tiplication. No activation above control condition, arith-
observed that individuals tend to mentally convert infor- metic tasks performed with Roman numerals, was found
mation that is presented in a not preferred mode to the during mental arithmetic tasks performed with Arabic
preferred mode of processing. In this study, the cognitive numerals in either AG region [21]. In another fMRI study,
style of an individual was assessed with the Verbalizer- multiplication and subtraction problems differed signifi-
Visualizer-Questionnaire [8]. The authors observed that cantly in right, but not left, IPS and AG activity [22]. A
during reading, visualizers showed activation in the right meta-analyses showed that the ability to process numbers
gyrus fusiformis, an area implicated in visual processing, and perform calculations relies on a large number of brain
whereas during picture presentation, verbalizers showed regions [23]. In addition to the brain areas of the triple-
activation in the left gyrus supramarginalis, an area impli- code model [20], activation during number and calculation
cated in verbal processing. The findings indicate that tasks were also observed in the cingulate gyri, the insula,
individuals with a visual cognitive style could have a ten- and the cerebellum. Furthermore, activation in dorsolat-
dency to convert linguistically presented information into eral and frontopolar areas of the prefrontal cortices was
a visual mental representation. Similarly, individuals with modulated by task difficulty [23].
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 3 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

The present study visualizers compared to weak visualizers, because strong

If an individual’s cognitive style reflects a general prefer- visualizers tend to convert verbally presented information
ence for a stimulus modality, we hypothesized, first, that into a visual mental representation.
strong verbalizers show higher activation in brain areas The present fMRI study aims at expanding our pre-
associated with language (supramarginal gyrus, Broca’s sent general knowledge of brain function in the field of
area) and auditory (Heschl’s gyrus, Rolandic operculum) arithmetic problem solving to better understand and
processing than weak verbalizers, and that strong visuali- manage the consequences of interindividual differences.
zers show higher activation in brain areas associated with
visual processing (fusiform gyrus, visual cortex) than weak Methods
visualizers while solving arithmetic problems. Participants
If the brain areas involved in number processing are 42 healthy young adults (21 female) participated in the
domain-specific (IPS) or modality-specific (AG) as pro- fMRI study. All were right-handed, all had normal or cor-
posed in the triple code model [20], that is, that the left rected to normal vision, and no history of neurological or
AG shows an affinity to the verbal domain, and that the psychiatric illness. All participants were university stu-
left and right IPS shows an affinity to the visual domain, dents. The participants had an average age of 23 years
the activation of these areas should show a dependency on (S.D. = 3.22). All participants were tested during daytime.
an individual’s cognitive style. We hypothesized, second, The study was approved by the ethics committee of the
that strong verbalizers show higher activation within the Medical University Graz.
left AG than weak verbalizers, and that strong visualizers
show higher activation within the IPS, bilaterally, than Self-report measure
weak visualizers. We designed a short self-report measure to assess the use
The present study comprised two functional sessions in of verbalization and visualization in mental arithmetic in a
an event-related design. In the first functional session, preliminary study. From an initial number of ten items
subtraction and multiplication problems were presented, five were excluded because they failed quality criteria (e.g.,
and in the second functional session, multiplications difficulty, selectivity). From the remaining five items,
were presented in two formats, with Arabic numerals or two items addressed the use of visualization in mental
with written number words. The individual’s habitual use arithmetic (e.g., In my mind, I see the numbers and (inter-
of visualization and verbalization during mental arith- mediate) results as written on paper.), and three items
metic was assessed with a short self-report measure. addressed the use of verbalization in mental arithmetic
Additionally, to assess a possible correlation between (e.g., I solve a problem easier if I repeat the problem and
cognitive style and intelligence, the participants com- the (intermediate) results in my mind.). The participants
pleted an intelligence test in a separate session. of the present study had to evaluate themselves for all
In the first functional session, multiplication and sub- items on a five point rating scale. The ticked values for the
traction problems were presented. When these different two visual items were added and divided by two. The
kinds of problem types are compared as a function of cog- ticked values for the verbal items were added and divided
nitive style, we expected, first, higher activation within the by three. As a consequence the self-reported use of verba-
IPS, bilaterally, while solving subtraction problems. This lization and visualization was comparable.
activation might be higher in strong visualizers compared
to weak visualizers. We expected, second, higher activation Intelligence test
within the left AG while solving multiplication problems. The intelligence test was conducted to assess the relation-
This activation might be higher in strong verbalizers com- ship between cognitive style and intelligence to be sure
pared to weak verbalizers. that the assessed use of visualization and verbalization
In the second functional session, multiplication pro- during mental arithmetic of an individual does not depend
blems were presented in two formats. When these differ- on intelligence. We used a standardized intelligence test
ent types of format are compared as a function of (Intelligence Structure Test in German language: I-S-T
cognitive style, we expected, first, higher activation within 2000 R) [24]. The battery contains a basic module measur-
brain areas associated with language processing while sol- ing verbal, numerical, and figural intelligence. Each con-
ving problems that are presented with Arabic numerals tent area of intelligence is assessed through three subtests.
for strong verbalizers compared to weak verbalizers,
because strong verbalizers tend to convert visually pre- Stimuli
sented information into a verbal mental representation. For the first functional session, multiplication and sub-
We expected, second, higher activation within areas that traction problems were designed. 15 multiplications were
are involved in visual processing while solving problems chosen from the multiplication table. All were one-digit
that are presented with written number words for strong times one-digit problems with two-digit solutions. The
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 4 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

one-digit operands were 3, 4, 6, 7, 8, and 9. Problems presented in white letters against a grey background. The
with other operands (1, 2 or 5) and ties (two equal oper- first functional session lasted approximately 20 minutes.
ands, e.g. 4 × 4) were not presented [25]. The order of During the second functional session the participant had
the factors of the 15 multiplication problems was chan- to solve 30 multiplication problems in two presentation
ged (e.g., 3 × 4 and 4 × 3) yielding 30 multiplication pro- formats, namely with Arabic numerals or with written
blems in total. The 30 multiplication problems were number words. Each trial started with the presentation of
presented twice during the first functional scan. Distrac- a fixation cross for 1 to 7 s (average presentation time of
tors for the multiplication problems were operand- 4 s), followed by the presentation of the problem for 4 s.
related (i.e. solutions of related problems; e.g., the dis- The result was shown for 2 s, giving a total trial duration
tractors of the problem 3 × 4 would be 8 (2 × 4), 16 (4 × of 9 to 13 s. The scheme of the task is shown in Figure 1B.
4), 9 (3 × 3), and 15 (3 × 5)). Subtractions were chosen Response times were measured from the onset of the pre-
such that their mean problem size was equal to the mean sentation of the result that could be correct or incorrect.
problem size of the multiplication problems. The min- The participants were instructed to use their right index
uend, the subtrahend (13, 14, 16, 17, 18 or 19) and the finger to press on one of the two buttons depending on
difference were two-digit numbers (e.g., 36-19). The whether the answer is correct or not. All stimuli were
15 selected subtraction problems were each presented presented in white letters against a grey background. The
four times during the first functional session. To prevent second functional session lasted approximately ten
the use of shortcut strategies, distractors for the subtrac- minutes.
tion problems were chosen such that the differences to Stimulus presentation and response recording were
the correct solution (± 1 or ± 10) were balanced over all programmed with the software Presentation (Neurobe-
problems and participants. Multiplication and subtrac- havioral Systems).
tion problems were intermixed and presented in random On a different day the participants took the intelligence
order. The first functional session consisted of 120 trials, test (Intelligence Structure Test in German language, I-S-
60 multiplication and 60 subtraction problems. The T 2000 R) [24]. This test took approximately 90 minutes
mean problem size (i.e. solution) of those multiplication to complete.
and subtraction problems was equal.
For the second functional session, the same 30 multipli- fMRI data acquisition
cation problems as in the first functional session were Imaging was preformed with a 3 T Siemens Magnetom
used. Each problem was presented twice in different for- Tim Trio scanner (Siemens Medical Solutions, Erlan-
mats, namely, either as Arabic numerals (e.g., 3 × 4) or as gen, Germany) and a 32-channel head coil. For the
written number words (e.g., three times four). The number anatomical images, an isotropic MPRAGE sequence
word problems were presented in German. The distractors was used (TR = 1900 ms, TE = 2.19 ms, TI = 900 ms,
were operand-related. The distractor and the solution flip angel = 9° inplane acqisition matrix = 256 × 256,
were presented in the same format as the problem. A total FoV = 256 (saggital view), 176 partitions, slice thick-
of 60 trials were presented in random order. ness 1 mm). For the functional images, a T2*-weighted
EPI-sequence was employed (TR/TE = 2000 ms/24 ms,
Procedure matrix = 64 × 64, FoV = 192, 31 axial slices, inplane
Before entering the scanner, the participant had to evalu- resolution: 3 × 3 mm, slice thickness 3 mm and 0.90
ate all items of the self-report measure to assess his use of mm gap, acquired descendingly and parallel to the AC-
verbalization and visualization in mental arithmetic. PC line) sensitive to brain oxygen-level dependent
During the first functional session, the participant had to (BOLD) contrast. 31 axial contiguous slices parallel to
solve 60 multiplication and 60 subtraction problems pre- the bicommissural plane were acquired, covering the
sented in random order. Each trial started with the presen- whole brain. Participants wore ear plugs as protection
tation of a fixation cross for 1 to 7 seconds (s; average against the scanner noise.
presentation time of 4 s), followed by the presentation of
the problem for 4 s. Then the two alternatives - the solu- fMRI data analysis
tion and one of the distractors - were presented for 2 s, Data analyses and pre-processing were performed with
giving a total trial duration of 9 to 13 s. The scheme of the the SPM software (SPM 8, Wellcome Department of
task is shown in Figure 1A. Response times were measured Cognitive Neurology, London, U.K.). The first two func-
from the onset of the presentation of the two alternatives. tional images of each participant were discarded to allow
The participant had to indicate on which side of the for magnetic saturation. The remaining functional images
screen the correct response was presented by pressing the were motion-corrected, unwarped, and corrected for slice
corresponding button with his right hand. All stimuli were acquisition time. The functional images were then
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 5 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Figure 1 Schematic of the task. A) First functional session: Subtraction and multiplication problems were presented in randomized order (120
trials). Participants had to choose the alternative that corresponds the solution. Response times were measured from the onset of the
presentation of the alternatives. B) Second functional session: 60 multiplication problems were presented as Arabic numerals or as written
number words in German language. Participants had to indicate if the solution is correct or not. Response times were measured from the onset
of the presentation of the solution.

normalized to correspond more closely to the MNI ana- comparisons at p < .05 and an uncorrected voxelwise p <
tomical template. Images were finally smoothed with a .001 level.
Gaussian kernel of 8 mm FWHM.
Statistical analyses were performed on the basis of the Region of interest analyses
general linear model implemented in SPM8. A model with For our ROI analysis, we focused on brain areas impli-
two conditions for each functional scan was analyzed (first cated in language, auditory, visual, and number proces-
functional session: multiplication/subtraction; second sing. The ROIs selected for analysis were the right and
functional session: Arabic format/word format). The left fusiform gyrus (BA 37; visual processing), supra-
experiment was analyzed on the basis of single events marginal gyrus (BA 40; language processing), Heschl’s
(event-related). The trial onsets of the single events were gyrus (BA 41; auditory processing), Rolandic opercu-
calculated from the logfiles saved after presenting the two lum (BA 43; auditory processing), visual cortex (BA 17/
experimental conditions with the software Presentation 18/19; visual processing), and intraparietal sulcus
(Neurobehavioral Systems) for each participant separately. (number processing) as well as the left Broca’s area
The delta-function of the trial onsets for each condition (BA 44/45; language processing), and the left angular
was convolved with the canonical form of the hemody- gyrus (BA 39; number processing). The ROIs were
namic response function and its first and second deriva- defined on the basis of the AAL atlas [26]. For the ROI
tive. A high-pass filter of 1/200 Hz and an autocorrelation analysis, the Marsbar toolbox was used (M. Brett,
model (AR(1)) were employed, but no low-pass filter and http://marsbar.sourceforge.net). Effect sizes were aver-
no global normalization. For the statistical group analyses, aged over all voxels of the ROI on the individual parti-
one sample t-tests were calculated to realize a random cipant level, for each condition against baseline. Next,
effects analysis. Significant activation clusters were deter- we tested for correlations between these effect sizes
mined using a height threshold of p < .001 uncorrected, and values of the self-reported use of verbalization and
with family-wise error (FWE) correction for multiple visualization.
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 6 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Behavioural data analysis In the second functional session, multiplication pro-

For the behavioural data analysis, only the data of those blems presented with Arabic numerals were solved faster
participants who were entered into the functional data (863 ms vs. 1126 ms) than multiplication problems pre-
analysis were used. For the first functional session, five sented with written number words (response times: t36 =
participants (three female) had to be excluded from the 1.0 p < 0.001). No difference was observed for the error
analysis because of excessive motion. The response times rates (error rates: 4.25% vs. 4.56%, n.s.).
data of one subject had to be excluded because of a tech-
nical malfunction. For the first functional session, the Whole brain analyses
behavioural data were analyzed for the remaining 37 par- In the first functional session, we compared two opera-
ticipants (18 female; age: M = 23.11, S.D. = 3.23). For the tions, multiplication and subtraction problems. Compared
second functional session, six participants (four female) to multiplication problems, subtraction problems activated
had to be excluded from the analysis because of excessive a range of areas in the frontal and parietal lobe, including
motion. For the second functional session, the beha- the IPS, as well as the basal ganglia. In the reverse con-
vioural data were analyzed for the remaining 36 partici- trast, more activation was observed within the left and
pants (17 female; age: M = 23.03, S.D. = 3.09). For the right AG, and in frontal brain areas (see Table 2 and
analysis of the response times, response times below 200 Figure 2).
ms were considered outliers. In total 386 data points for In the second functional session, we compared multipli-
the first functional session (from 4440 data points in cation problems presented with Arabic numerals with the
total), and 201 data points for the second functional ses- same problems presented with written number words.
sion (from 2160 data points in total) were removed from Compared to problems presented with Arabic numerals,
the analysis. Error rates were arcsin √p transformed to problems presented with written number words activated
achieve approximate variance equality [27]. Response more strongly inferior frontal brain areas including Broca’s
times and the arcsin √p transformed error rates were area, temporal brain areas including the left superior tem-
entered into two separate paired t-tests with the factors poral gyrus, parietal brain areas including the left and
OPERATION (multiplication, subtraction) for the first right AG, IPS, the fusiform gyrus, and supramarginal
functional session, and FORMAT (Arabic, word) for the gyrus, and occipital brain areas. In the reverse contrast the
second functional session. left and right AG, the left posterior cingulum, the right
superior frontal gyrus, and the left insula were activated
Results (see Table 3 and Figure 3).
With regard to intelligence, we observed no significant No significant correlations were observed in the whole
correlation between the self-reported use of verbalization brain analysis between brain activation and the values of
or visualization in mental arithmetic and verbal, numerical the self-reported use of visualization or verbalization.
or figural intelligence. This indicates that an self-reported
habitual use of visualization and verbalization during men- ROI analyses
tal arithmetic does not depend on intelligence. Statistics Correlations were calculated between averaged effect sizes
for the correlations between the cognitive style and intelli- within each ROI and the values of the self-reported use of
gence are presented in Table 1. verbalization as well as visualization in mental arithmetic.
We used a Bonferroni correction for the number of ROIs
Response times and error rates for each modality. Thus, the critical p value for the verbal
In the first functional session, multiplication problems cognitive style hypotheses is .006 (.05/8; 8 independent
were solved faster (766 ms vs. 904 ms) and with fewer ROIs: left angular gyrus, left supramarginal gyrus, right
errors (4.57% vs. 9.08%) than subtraction problems supramarginal gyrus, left Heschl’s gyrus, right Heschl’s
(response times: t35 = -6.46, p < 0.001; error rates: t36 = gyrus, left Rolandic Operculum, right Rolandic Opercu-
-4.71, p < 0.001). lum, and left Broca’s area), and the critical p value for the
visual cognitive style hypotheses is .005 (.05/10; 10 inde-
pendent ROIs: left inferior visual cortex, left middle visual
Table 1 Correlations between verbalization/visualization cortex, left superior visual cortex, right inferior visual cor-
and intelligence tex, right middle visual cortex, right superior visual cortex,
Verbal Numerical Figural left fusiform gyrus, right fusiform gyrus, left IPS, and right
intelligence intelligence intelligence IPS).
Visualization -.066 (p = .678) -.090 (p = .572) -.125 (p = .429) In the first functional session, correlations were calcu-
Verbalization .114 (p = .472) .223 (p = .156) .141 (p = .374) lated between averaged effect sizes within each ROI for
The self-reported use of verbalization (M = 3.46, S.D. = 1.07) and visualization
multiplication and subtraction problems and the values of
(M = 3.91, S.D. = .93) is correlated negatively (r = .-456, p = .002; n = 42). the self-reported use of verbalization as well as visualization
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 7 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Table 2 First functional session: Group analysis of the contrasts subtraction versus multiplication and multiplication
versus subtraction
Side Sub > Mult x y z k Z Side Mult > Sub x y z k Z
right Parietal_Sup 15 -64 52 1465* Inf left Angular -54 -64 34 369 6.25
left Parietal_Sup -12 -67 52 866* Inf right Angular 57 -58 34 502 6.05
right Caudate 21 14 4 247* Inf left Frontal_Sup -12 41 49 364 5.57
left Frontal_Mid -24 -1 55 114* 7.30 left Frontal_Inf_Orb -42 32 -17 85 5.04
left Putamen -18 14 1 160* 7.27 left Temporal_Inf -57 -22 -26 142 4.92
right Frontal_Sup 27 2 52 84* 7.10 right Cingulum 9 -49 31 194 4.88
left Occipital_Inf -42 -79 -2 204* 7.00 right Frontal_Sup 15 35 55 199 4.67
left Thalamus -12 -19 10 19* 6.49 left Insula -39 -13 7 158 4.36
left Supp_Motor_Area 3 14 52 48* 6.49
left Precentral -45 5 31 12* 6.40
right Lingual 24 -55 -2 2* 6.23
right Thalamus 15 -19 10 1* 6.10
left Occipital_Mid -24 -97 -2 3* 6.07
(Abbreviations: k = cluster size, Z = Z-value; activation significant at p < 0.001 uncorrected, p < 0.05 FWE corrected on cluster level)
Coordinates are reported as given by SPM8 (MNI space) and correspond only approximately to Talairach and Tournoux space (Talairach and Tournoux, 1988,
Brett et al., 2001). The label denotes the location of the maximum. Abbrevations: Parietal_Sup - superior parietal lobe, Caudate - caudate nucleus, Frontal_Mid -
middle frontal gyrus, Frontal_Sup - superior frontal gyrus, Occipital_Inf - inferior occipital gyrus, Supp_Motor_Area - supplementary motor area, Precentral -
precentral gyrus, Lingual - lingual gyrus, Occipital_Mid - middle occipital gyrus, Angular - angular gyrus, Frontal_Inf_Orb - inferior orbital frontal gyrus,
Temporal_Inf - inferior temporal gyrus, Cingulum - cingulate gyrus, * this activation is part of a bigger cluster

Figure 2 Whole Brain Analysis: First functional session. A) One-sample t-test of the contrast multiplications versus subtractions (p < 0.001).
B) One-sample t-test of the contrast subtrations versus multiplications (p < 0.001).
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 8 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Table 3 Second functional session: Group analysis of the contrasts word versus Arabic format and Arabic versus word
format
Side Word > Ara x y z k Z Side Ara > Word x y z k Z
left Calcarine 9 -79 1 7028 Inf right Angular 48 -58 28 209 5.06
left Frontal_Inf_Oper -39 8 25 2597 Inf right Cingulum_Post 9 -49 28 192 5.05
right Frontal_Inf_Oper 45 11 28 693 5.09 right Frontal_Sup 33 26 55 84 5.03
right Temporal_Mid 48 -34 1 94 5.07 left Angular -51 -64 31 141 4.04
right Frontal_Inf_Orb 39 23 -8 294 4.99 left Insula -33 16 4 101 3.98
right Thalamus 12 -4 -5 77 4.22
(Abbreviations: k = cluster size, Z = Z-value; activation significant at p < 0.001 uncorrected, p < 0.05 FWE corrected on cluster level)
Coordinates are reported as given by SPM8 (MNI space) and correspond only approximately to Talairach and Tournoux space (Talairach and Tournoux, 1988,
Brett et al., 2001). The label denotes the location of the maximum. Abbrevations: Calcarine - calcarine gyrus, Frontal_Inf_Oper - inferior frontal gyrus,
Temporal_Mid - middle temporal gyrus, Frontal_Inf_Orb - inferior orbital frontal gyrus, Angular - angular gyrus, Cingulum_Post - posterior cingulum, Frontal_Sup -
superior frontal gyrus

in mental arithmetic. Significant correlations were observed and one area implicated in auditory processing, namely the
only for the subtraction problems, and the scores of the left Rolandic operculum (r = .361, p = .028). These results
self-reported use of verbalization. We found significantly suggest that the higher the self reported tendency to verba-
positive correlations between scores on the self-reported lize, the higher the activation in an area implicated in lan-
use of verbalization and one area implicated in verbal pro- guage processing, namely the left and right supramarginal
cessing, the right and left supramarginal gyrus (right: r = gyrus, and an area implicated in auditory processing,
.364, p = .027; left: r = .462, p = .004, Bonferroni-corrected), namely the left Rolandic operculum, while solving

Figure 3 Whole Brain Analysis: Second functional session. A) One-sample t-test of the contrast Arabic versus word format (p < 0.001). B)
One-sample t-test of the contrast word versus Arabic format (p < 0.001).
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 9 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

subtraction problems. It should be noted, however, that Rolandic operculum, and the right and left Heschl’s
only the correlation with the left supramarginal gyrus gyrus. The supramarginal gyrus is a region in the inferior
passes the Bonferroni correction. parietal lobe, and has been found to be involved in pho-
In the second functional session, correlations were cal- nological processing [28], reading both in regards to
culated between averaged effect sizes within each ROI for meaning [29] and phonology [30], word production [31],
problems presented with Arabic numerals and problems and grammar learning [32]. The Rolandic operculum has
presented with written number words and the values of been found to be a somatosensory region [33,34],
the self-reported use of verbalization as well as visualiza- involved in auditory processing, activated by listening to
tion in mental arithmetic. Significant correlations were the sound of one’s own voice [35], and the processing of
observed for both presentation formats. For problems pre- prosody [36,37]. The left Rolandic operculum is assumed
sented with Arabic numerals we found significantly posi- to be involved in syntactic encoding during speaking
tive correlations between the self-reported use of [38], and phonological rehearsal [39]. The right Rolandic
verbalization and activation in ROIs implicated in auditory operculum has been associated with the processing of
processing, the right and left Heschl’s gyrus (right: r = sentence intonation [40], and of slow prosodic modula-
.367, p = .028; left: r = .357, p = .033), and the left Rolandic tions [41]. The activation of the right Rolandic opercu-
Operculum (r = .374, p = .025). For problems presented lum increased with the degree an individual relied on
with written number words we found significant positive verbalization during mental arithmetic when the pro-
correlations between the self-reported use of verbalization blems were presented with written number words. This
and areas implicated in language and auditory processing, result suggests that verbalizers imagine hearing the
the left supramarginal gyrus (r = .346, p = .039), the right sound of their voices mentally while reading numbers
and left Heschl’s gyrus (right: r = .413, p = .012; left: r = presented with number words, confirming their own sub-
.463, p = .005, Bonferroni-corrected), and the right and jective self-report. The right and left Heschl’s gyrus is
left Rolandic operculum (right: r = .360, p = .031; left: r = found in the area of the primary auditory cortex in the
.486, p = .003, Bonferroni-corrected). We observed no sig- superior temporal gyrus of the human brain, the first cor-
nificant correlations for the self-reported use of visualiza- tical structure to process incoming auditory information
tion. These results suggest that the higher the self [42]. Naming numbers has been shown to be dependent
reported tendency to verbalize while solving multiplication on linguistic properties. For example, naming latencies
problems the higher the activation in areas implicated in for two-digit numbers increase with syllable length
verbal and auditory processing, namely in the left supra- [43,44]. The authors suggested that digits are translated
marginal gyrus, the right and left Heschl’s gyrus, and the into a verbal code before being processed [43,44].
right and left Rolandic operculum. It should be noted, that It is unclear, why we found significant correlations only
only the correlations between problems presented with for the verbalizer dimension but not for the visualizer
written number words and the left Heschl’s gyrus as well dimension. One reason might be that participants are not
as the left Rolandic Operculum passed the Bonferroni cor- very good at self-reporting their habits in information pro-
rection. Figure 4 shows the significant results for the cessing. We think it would be a major contribution for
supramarginal gyrus, Figure 5 for the Heschl’s gyrus, and further studies in this field to design a more valid ques-
Figure 6 for the Rolandic operculum. tionnaire about cognitive styles in mental arithmetic. A
limitation of the present study might be that the self-
Discussion reported use of visualization in mental arithmetic was lim-
The aim of the present fMRI study was to investigate the ited to the visualization of numbers and (intermediate)
influence of the verbal-visual cognitive style on cerebral results while calculating. Other types of visualizations,
activation patterns during mental calculation. The habitual such as to move mentally along the number line, imagin-
use of visualization and verbalization during mental arith- ing a mass that increases or decrease in magnitude, or
metic of each of the 42 right-handed participants was some collection of dots that grows or shrinks in number
assessed with a short self-report measure. In the first func- and others were not included. Another reason may be that
tional session, subtraction and multiplication problems the statistical spread of the self-reported use was greater
were presented, and in the second functional session, mul- for the verbalizer dimension than for the visualizer dimen-
tiplications were presented in two formats, either as sion (Var(score verb = 1.15) > Var(score vis = .87) ).
Arabic numerals or as written number words. It is surprising that the correlations for the first func-
With regard to verbalization, we found that the higher tional session between individual differences in verbaliza-
the self reported tendency to verbalize during mental tion and brain activation during calculation were only
arithmetic the higher the activation in brain areas related found for the subtraction but not for the multiplication
to language or auditory processing, namely, within the problems (see Figure 7). It has frequently been argued
right and left supramarginal gyrus, the right and left that multiplication relies more on verbal processing than
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 10 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Figure 4 ROI analysis results: Supramarginal Gyrus. Significant correlation between peak activations in the supramaginal gyrus and the self-
reported use of verbalization: In the first functional session significant correlations appeared for the subtraction problems and the right and left
[* correlation passes the Bonferroni correction (critical p value is .006)] supramarginal gyrus. In the second functional session a significant
correlation appeared for multiplications presented with written number words and the left supramarginal gyrus.

subtraction, which is assumed to rely more on visuo-spa- language processing, we observed no modulation of acti-
tial processing [2,45]. Therefore, we would have expected vation in the left AG as a function of the self-reported
verbalizers to show greater language-related activation tendency to use verbalization while performing mental
for multiplications. One possibility why we did not find arithmetic. The left AG is assumed to support the long
significant activation patterns for the multiplication pro- term memory retrieval for arithmetic fact knowledge
blems in the first functional session could be that the [20]. Arithmetic fact knowledge is required, for example,
multiplication problems used here were very easy to in the skilled solving of multiplication problems by
solve. They were part of the multiplication tables and the retrieving the result from the verbal long-term memory,
results could be retrieved from long-term memory. The namely, from the multiplication tables learned in child-
behavioral results showed that the presented subtraction hood. The left AG shows stronger activation for solving
problems were more difficult than the presented multi- arithmetic problems for which participants report fact
plication problems. It is possible that differences with retrieval whereas the application of procedural strategies
regard to verbalization as cognitive style emerge only for is accompanied by widespread activation in a fronto-
more difficult arithmetic problems that have to be solved parietal network [46]. These findings link the left AG to
in several processing steps leading to intermediate results arithmetic fact retrieval. In the present study, the activa-
that have to be kept in memory. tion of the left AG was not modulated by verbalization,
With regard to a possible domain-specificity (IPS) or and it could be concluded that the left AG shows no
modality-specificity (AG) of brain areas involved in specific affinity to the verbal domain and subserves
number processing we did not observe any modulation. number processing in a modality-general way. This
Although we found significant brain activation as a interpretation corresponds to some recent findings
function of verbalization in areas related to sound and [21,22] which also raise some doubts about the
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 11 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Figure 5 ROI analysis results: Heschl’s Gyrus. Significant correlations between peak activations in the Heschl’s gyrus and the self-reported use
of verbalization: In the second functional session significant correlations appeared for multiplications presented with written number words and
the right and left [* correlation passes the Bonferroni correction (critical p value is .006)] Heschl’s Gyrus, and for multiplications presented with
Arabic numerals and the right and left Heschl’s Gyrus.

assumption that the left AG mediates verbal fact retrie- multiplications presented with Arabic numerals. It is
val during multiplication. highly likely that the greater activation in visual areas for
In the first functional session, we found that subtrac- multiplications presented with written number words is
tion problems activated the right and left IPS more due to the larger number of characters. The multiplica-
strongly than multiplication problems, whereas multipli- tions in written number words were presented using
cation problems activated the left and right angular three words with several letters, whereas the multiplica-
gyrus more strongly than subtraction problems. This tions presented with Arabic numerals consisted of two
finding corresponds well to results reported in previous Arabic numerals and a multiplication sign. Other studies
studies [47,48] as well as to the model of Dehaene and also reported format effects in arithmetic. It has been
collegues [20]. It should be noted, however, that multi- suggested that numbers presented in different surface-
plication problems were solved faster and more accu- formats have differential access to number representa-
rately than subtraction problems, although problem tions [49]. Format differences were also observed in an
sizes were identical. It can therefore not be excluded EEG-study, with more negative event-related potentials
that some of the observed activation differences are also for written number words than for Arabic numerals and
due to a difference in task difficulty. auditorily presented number words [50]. The modality-
In the second functional session, we found that multi- dependent access to numerical information may be a
plications presented with written number words activated consequence of modality-dependent access to the num-
areas implicated in visual processing more strongly than ber representation in parietal cortex. A transcranical
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 12 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Figure 6 ROI analysis results: Rolandic Operculum. Significant correlations between peak activations in the Rolandic operculum and the self-
reported use of verbalization: In the first functional session a significant correlation appeared for the subtraction problems and the left Rolandic
Operculum. In the second functional session significant correlations appeared for multiplications presented with written number words and the
right and left [* correlation passes the Bonferroni correction (critical p value is .006)] Rolandic Operculum, and for multiplications presented with
Arabic numerals and the left Rolandic Operculum.

magnetic stimulation experiment showed a dissociation A cognitive style is assumed to be a relatively stable
between digits and number words in the right parietal characteristic that describes an individual’s way to process
lobe, whereas the left parietal lobe showed a double dis- information [4]. Consequently a questionnaire that
sociation between the different numerical formats [51]. assesses the habitual use of visualization and verbalization
Typically, problems presented with number words take has to measure an independent construct and not person-
longer to solve than problems presented with Arabic ality or intelligence [54,55]. Previous studies already
numerals [52,53]. In our study, multiplications presented showed independence of the visual-verbal cognitive style
with Arabic numerals were also solved faster than multi- from intelligence and personality [56,57]. Regarding intelli-
plication problems presented with written number words. gence, we found no significant correlations between the
The visual-verbal cognitive style is assumed to be a verbal or visual cognitive style and verbal, numerical or
relatively stable characteristic although it might depend figural intelligence. This indicates that the correlations
on the task [5]. An individual might, for example, prefer between cognitive style and brain activation are not due to
visualization for solving arithmetic problems and verba- differences in intelligence.
lization for memorizing a poem. It is therefore prefer- The aim and results of the present study might be
able to assess the visual-verbal cognitive style specifically understood by some researchers as advocating a view
for the tested task domain. We tried to assess the visual- that different learning styles have real consequences for
verbal cognitive style during mental arithmetic with a the brain and that education should be adapted accord-
short self-report measure. Our results indicate that peo- ingly. It is important to distinguish between cognitive
ple who say they verbalize more show more activity in styles and learning styles. Cognitive styles are assumed to
brain areas related to language and auditory processing. be an individual’s way to process information, whereas
It is unclear, however, in how far these results are speci- learning styles are concerned with the learning environ-
fic to mental arithmetic. ment [58]. Some studies showed correlations between
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 13 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

Figure 7 ROI analysis results: Multiplication problems in the first functional session. Correlations between peak activations in the left supramaginal
gyrus, the left Rolandic Operculum, and the left Heschl’s gyrus and the self-reported use of verbalization: In the first functional session no significant
correlations appeared for the multiplication problems and the left supramarginal gyrus, the left Rolandic Operculum, and the left Heschl’s gyrus.

cognitive styles and some environmental conditions, correlates positively with activation in brain areas impli-
especially the preferred learning mode. There is a ten- cated in verbal and auditory processing. However, the
dency for visualizers to use pictures and for verbalizers to absence of a similar modulation of activation within areas
prefer writing as working mode or learning mode [59,60]. specifically involved in number processing could be taken
In our study, the self-reported cognitive style influenced to indicate that these areas are less modality-specific (AG)
activity in brain areas directly related to the preferred or domain-specific (IPS) than currently proposed. Our
modality of information processing. results indicate that different cognitive styles do not lead
to a difference in arithmetic processing in the brains of
Conclusions skilled adults. It is possible, however, that cognitive styles
In the present study, we observed that the self-reported play a role in the acquisition of mathematic skills. Further
habitual use of verbalization in mental arithmetic studies could show that identifying and catering to a
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 14 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

student’s preferred cognitive style in education leads to a 16. Siegel M, Donner TH, Oostenveld R, Fries P, Engel AK: Neural
synchronization along the dorsal visual pathway reflects the focus of
more efficient acquisition of arithmetic skills. spatial attention. Neuron 2008, 60:709-719.
17. Materna S, Dicke PW, Their P: Dissociable Roles of the superior temporal
sulcus and the intraparietal sulcus in joint attention: A functional
Acknowledgements magnetic resonance imaging study. J Cogn Neurosci 2008, 20:108-119.
This work was supported by a grant from the Austrian Science Fund 18. Todd JJ, Marois R: Capacity limit of visual short-term memory in human
(FWF): [P22189]. We thank Manuel Ninaus for assisting us with all posterior parietal cortex. Nature 2004, 428:751-754.
measurements. 19. Harrison A, Jolicoeur P, Marois R: “What” and “where” in the intraparietal
sulcus: An fMRI study of object identity and location in visual short-term
Author details memory. Cereb Cortex 2010, 20:2478-2485.
1
Cognitive Psychology and Neuroscience, Department of Psychology, 20. Dehaene S, Piazza M, Pinel P, Cohen L: Three parietal circuits for number
University of Graz, Universitaetsplatz 2/III, 8010 Graz, Austria. 2Department of processing. Cogn Neuropsychol 2003, 20:487-506.
Radiology, Medical University of Graz, Auenbruggerplatz 9, 8036 Graz, 21. Wu SS, Chang TT, Majid A, Caspers S, Eickhoff SB, Menon V: Functional
Austria. 3Department of Knowledge Discovery, University of Technology of heterogeneity of inferior parietal cortex during mathematical cognition
Graz, Krenngasse 37/IV, 8010 Graz, Austria. 4Department of Psychology, assessed with cytoarchitectonic probability maps. Cereb Cortex 2009,
University of Graz, Universitaetsplatz 2, 8010 Graz, Austria. 19:2930-2945.
22. Rosenberg-Lee M, Chang TT, Young CB, Wu S, Menon V: Functional
Authors’ contributions dissociations between four basic arithmetic operations in the human
The presented work was carried out in collaboration between all authors. All posterior parietal cortex: A cytoarchitectonic mapping study.
authors read and approved the final manuscript. SZ designed the Neuropsychologia 2011, doi:10.1016/j.neuopsychologie.2011.04.035.
experiment, collected and analyzed the data, interpreted the results, and 23. Arsalidou M, Taylor MJ: Is 2 + 2 = 4? Meta-analyses of brain areas needed
wrote the paper. VB helped at collecting the data and gave writing for numbers and calculations. Neuroimage 2011, 54:2382-2393.
assistance. AI and CN made substantial contributions to the design, the 24. Liepmann D, Beauducel A, Brocke B, Amthauer R: Intelligenz-Struktur-Test
acquisition of data, and the analysis and interpretation of data. FE, KK and 2000 R, erweiterte Auflage Göttingen: Hogrefe; 2007.
GR provided technical help, and gave writing assistance at the method 25. De Brauwer J, Verguts T, Fias W: The representation of multiplication
section. facts: developmental changes in the problem size, five and tie effects. J
Exp Child Psychol 2006, 94:43-56.
Competing interests 26. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O,
All authors declare not to have any conflict of interest including any Delcroix N, Mazoyer B, Joliot M: Automated anatomical labelling of
financial, personal or other relationships with other people or organizations activations in SPM using a macroscopic anatomical parcellation of the
that could inappropriately influence, or be perceived to influence, their work. MNI MRI single participant brain. Neuroimage 2002, 15:273-289.
27. Bishop YMM, Fienberg SE, Holland PW: Discrete Multivariate Analysis. Theory
Received: 8 August 2011 Accepted: 12 March 2012 and Practice Cambridge: MIT Press; 1975.
Published: 12 March 2012 28. Wilson LB, Tregellas JR, Slason E, Pasko BE, Rojas DC: Implicit phonological
priming during visual word recognition. Neuroimage 2011, 55:724-731.
References 29. Humphries C, Binder JR, Medler DA, Liebenthal E: Time course of semantic
1. Hegarty M, Kozhevnikov M: Types of visual-spatial representation and processes during sentence comprehension: an fMRI study. Neuroimage
mathematical problem solving. J Educ Psychol 1999, 91:684-689. 2007, 36:924-932.
2. Lee KM, Kang SY: Arithmetic operation and working memory: Differential 30. Stoeckel C, Gough PM, Watkins KE, Devlin JT: Supramarginal gyrus
suppression in dual tasks. Cognition 2002, 83:63-68. involvement in visual word recognition. Cortex 2009, 45:1091-1106.
3. LeFevre JA, Liu J: Numerical cognition: Single-digit multiplication skills of 31. Pei X, Leuthardt EC, Gaona CM, Brunner P, Wolpaw JR, Schalk G:
adults from China and Canada. Mem Cognit 1997, 3:31-62. Spatiotemporal dynamics of electrocorticographic high gamma activity
4. Riding RJ, Cheema I: Cognitive styles - an overview and integration. Educ during overt and covert word repetition. Neuroimage 2011, 54:2960-2972.
Psychol 1991, 11:193-215. 32. Loui P, Li HC, Schlaug G: White matter integrity in right hemisphere
5. Blazhenkova O, Kozhevnikov M: The new object-spatial-verbal cognitive predicts pitch-related grammar learning. Neuroimage 2011, 55:500-507.
style model: theory and measurement. Appl Cogn Psychol 2009, 33. Stice E, Spoor S, Bohon C, Veldhuizen MG, Small DM: Relation of reward
23:638-663. from food intake and anticipated food intake in obesity: a functional
6. Burbaud P, Camus O, Guehl D, Bioulac B, Caille J-M, Allard M: Influence of magnetic resonance imaging study. J Abnorm Psychol 2008, 117:924-935.
cognitive strategies on the pattern of cortical activation during mental 34. Stice E, Spoor S, Ng J, Zald DH: Relation of obesity to consummatory and
subtraction. A functional magnetic resonance study of mental anticipatory food reward. Physiol Behav 2009, 97:551-560.
subtraction in human participants. Neurosci Lett 1999, 273:195-199. 35. Brown S, Ingham RJ, Ingham JC, Laird AR, Fox PT: Struttered and fluent
7. Kraemer DJM, Rosenberg LM, Thompson-Schill SL: The neural correlates of speech production: An ALE meta-analysis of functional neuroimaging
visual and verbal cognitive styles. J Neurosci 2009, 29:3792-3798. studies. Hum Brain Mapp 2005, 25:105-117.
8. Kirby JR, Moore PJ, Schofield NJ: Verbal and visual learning styles. 36. Koelsch S, Fritz T, Cramon DYV, Muller K, Friederici AD: Investigating
Contemp Educ Psychol 1998, 13:169-184. emotion with music: an fMRI study. Hum Brain Mapp 2006, 27:239-250.
9. Feigenson L, Dehaene S, Spelke E: Core systems of number. Trends Cogn 37. Ischebeck A, Friederici AD, Alter K: Processing prosodic boundaries in
Sci 2004, 8/7:307-314. natural and hummed speech: an fMRI study. Cereb Cortex 2008,
10. Eger E, Sterzer P, Russ MO, Giraud A-L, Kleinschmidt A: A supramodal 18:541-552.
number representation in human intraparietal cortex. Neuron 2003, 38. Indefrey P, Brown CM, Hellwig F, Amunts K, Herzog H, Seitz RJ, Hagoort P:
37:719-725. A neural correlate of syntactic encoding during speech production. PNAS
11. Pinel P, Dehaene S, Riviere D, Le Bihan D: Modulation of parietal 2001, 98:5933-5936.
activation by semantic distance in a number comparison task. 39. Veroude K, Norris DG, Shumskaya E, Gullberg M, Indefrey P: Functional
Neuroimage 2001, 14:1013-1026. connectivity between brain regions involved in learning words of a new
12. Nieder A: Counting on neurons: The neurobiology of numerical language. Brain Lang 2010, 113:21-27.
competence. Nat Rev Neurosci 2005, 6:177-190. 40. Meyer M, Steinhauer K, Alter K, Friederici AD, von Cramon AD: Brain
13. Dehaene S: Varieties of numerical abilities. Cognition 1992, 44:1-42. activity varies with modulation of dynamic pitch variance in sentence
14. McCloskey M: Cognitive mechanisms in numerical processing: Evidence melody. Brain Lang 2004, 89:277-289.
from acquired dyscalculia. Cognition 1992, 44:107-157. 41. Meyer M, Alter K, Friederici AD, Lohmann G, von Cramon DY: Functional
15. Grefkes C, Fink GR: The functional organization of the intraparietal sulcus MRI reveals brain regions mediating slow prosodic manipulations of
in humans and monkeys. J Anat 2005, 207:3-17. spoken sentences. Hum Brain Mapp 2002, 17:73-88.
Zarnhofer et al. Behavioral and Brain Functions 2012, 8:13 Page 15 of 15
http://www.behavioralandbrainfunctions.com/content/8/1/13

42. Morosan P, Rademacher J, Schleicher A, Amunts K, Schormann T, Zilles K:

Human primary auditory cortex: cytoarchitectonic subdivisions and
mapping into a spatial reference system. Neuroimage 2001, 13:684-701.
43. Eriksen CW, Pollack MD, Montague WE: Implicit speech: Mechanism in
perceptual encoding? J Exp Psychol 1970, 84:502-507.
44. Klapp S: Syllable-dependent pronunciation latencies in number naming:
A replication. J Exp Psychol 1974, 102:1138-1140.
45. Knops A, Viarouge A, Dehaene S: Dynamic representations underlying
symbolic and non-symbolic calculation: Evidence from the operational
momentum effect. Atten Percept Psychophys 2009, 71:803-821.
46. Grabner RH, Ansara D, Koschutnig K, Reishofer G, Ebner F, Neuper C: To
retrieve or to calculate? Left angular gyrus mediates the retrieval of
arithmetic facts during problem solving. Neuropsychologia 2009,
47:604-608.
47. Chochon F, Cohen L, van de Moortele PF, Dehaene S: Differential
contributions of the left and right inferior parietal lobules to number
processing. J Cogn Neurosci 1999, 11:617-630.
48. Lee KM: Cortical areas differentially involved in multiplication and
subtraction: A functional magnetic resonance imaging study and
correlation with a case of selective acalculia. Ann Neurol 2000, 48:657-661.
49. Cohen Kadosh R, Walsh V: Numerical representation in the parietal lobes:
abstract or not abstract. Behav Brain Sci 2009, 32:313-328.
50. Szucs D, Csepe V: Access to numerical information is dependent on the
modality of stimulus presentation in mental addition: a combined ERP
and behavioral study. Cogn Brain Res 2004, 19:10-27.
51. Cohen Kadosh R, Muggleton N, Silvanto J, Walsh V: Double dissociation of
format-dependent and number-specific neurons in human parietal
cortex. Cereb Cortex 2010, 20:2166-2171.
52. Campbell JID: Architectures for numerical cognition. Cognition 1994,
53:1-44.
53. Noel M-P, Fias W, Brysbaert M: About the influence of the presentation
format on arithmetical-fact retrieval processes. Cognition 1997,
63:335-374.
54. Riding RJ: On the nature of cognitive style. Educ Psychol 1997, 17:29-49.
55. Peterson ER, Rayner SG, Armstrong SJ: Researching the psychology of
cognitive style and learning style: Is there really a future? Learn Individ
Diff 2009, 19:518-523.
56. Riding RJ, Pearson F: The relationship between cognitive style and
intelligence. Educ Psychol 1995, 14:413-425.
57. Peterson ER, Deary IJ, Austin EJ: Are intelligence and personality related
to verbal-imagery and wholistic-analytic cognitive styles? Person Individ
Diff 2005, 39:201-213.
58. Rayner S, Riding RJ: Towards a categorization of cognitive styles and
learning styles. Educ Psychol 1997, 17:5-29.
59. Riding RJ, Mathias D: Cognitive styles and preferred learning mode,
reading attainment and cognitive ability in 11-year-old children. Educ
Psychol 1991, 11:383-393.
60. Riding RJ, Read G: Cognitive style and pupil learning preferences. Educ
Psychol 1996, 16:81-106.

doi:10.1186/1744-9081-8-13
Cite this article as: Zarnhofer et al.: The Influence of verbalization on
the pattern of cortical activation during mental arithmetic. Behavioral
and Brain Functions 2012 8:13.

Submit your next manuscript to BioMed Central

and take full advantage of:

• Convenient online submission

• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution

Submit your manuscript at

www.biomedcentral.com/submit