Content uploaded by Peter König
Author content
All content in this area was uploaded by Peter König on Aug 05, 2020
Content may be subject to copyright.
Inter-individual Differences among Native Right-to-Left Readers and Native Left-to-
1
Right Readers during Free Viewing Task
2
3
Zaeinab Afsari1, Ashima Keshava1, José P. Ossandón1,2, and Peter König1,3
4
1 Institute of Cognitive Science, University Osnabrück Germany
5
2 Institute of Psychology, University of Hamburg, Germany
6
3 Institute of Neurophysiology & Pathophysiology, University Medical Centre Hamburg-
7
Eppendorf, Germany
8
Corresponding Author:
9
Zaeinab Afsari
10
Institute für Kognitionswissenschaft
11
Universität Osnabrück
12
Wachsbleiche 27, D-49090 Osnabrück
13
Germany
14
Phone: +49 (541) 9693380
15
zaeinab.afsari@gmail.com
16
17
18
Zaeinab Afsari’s work was supported by grant ERC-2010-AdG#269616-MULTISENSE.
19
Peter König’s and Ashima Keshava’s work was supported by grant H2020-FETPROACT-
20
2014, SEP-210141273, ID: 641321 (socSMCs).
21
José Ossandón’s work was supported by grant SFB 936, project B1
22
23
Abstract
24
Human visual exploration is not homogeneous but displays spatial biases. Specifically, early
25
after the onset of a visual stimulus, the majority of eye movements target the left visual space.
26
This horizontal asymmetry of image exploration is rather robust with respect to multiple
27
image manipulations, yet can be dynamically modulated by preceding text primes. This
28
characteristic points to an involvement of reading habits in the deployment of visual attention.
29
Here, we report data of native right-to-left (RTL) readers with a larger variation and stronger
30
modulation of horizontal spatial bias in comparison to native left-to-right (LTR) readers after
31
preceding text primes. To investigate the influences of biological and cultural factors, we
32
measure the correlation of the modulation of the horizontal spatial bias for native RTL readers
33
and native LTR readers with multiple factors: age, gender, second language proficiency, and
34
age at which the second language was acquired. The results demonstrate only weak or no
35
correlations between the magnitude of the horizontal bias and the previously mentioned
36
factors. We conclude that the spatial bias of viewing behaviour for native RTL readers is
37
more variable than for native LTR readers, and this variance could not be demonstrated to be
38
associated with interindividual differences. We speculate the role of strength of habit and/or
39
the interindividual differences in the structural and functional brain regions as a cause of the
40
RTL spatial bias among RTL native readers.
41
42
43
Introduction
44
Eye movement and attention are jointly studied in research on overt attention.
45
Specifically, the decision to shift the eyes towards specific locations and not at other locations
46
is considered a complex cognitive process. This process is influenced by several factors:
47
stimulus-dependent tasks, task-related factors, and spatial properties (Kollmorgen, Nortmann,
48
Schröder, & König, 2010).
49
Even though these factors can be activated simultaneously, each one has its own
50
specific characteristics. Image features such as colours, intensity, orientation, shape, and
51
motion are considered stimuli attracting attention and processed by a bottom-up directed
52
pathway (Corbetta & Shulman, 2002; Itti & Koch, 2000, 2001; Ossandón et al., 2012). This
53
pathway consists of ventral and dorsal streams of the visual cerebral cortex (Kastner &
54
Ungerleider, 2000), as well as subcortical structures such as the superior colliculus
55
(Ignashchenkova, Dicke, Haarmeier, & Thier, 2004). At the same time, memories, goals, the
56
individual’s emotional state, and expectations can redirect and control the attention by a top-
57
down signalling pathway (DeAngelus & Pelz, 2009; Kaspar & König, 2011). Initiated by
58
internal goals, the executive frontal cortex sends signals to the visual motor areas to perform
59
appropriate eye movements (Buschman & Miller, 2007; Corbetta & Shulman, 2002; Itti &
60
Koch, 2001). Additionally, spatial properties influence the selection of fixation locations. One
61
aspect of these properties is the tendency to fixate more often toward the central parts of a
62
scene rather than its periphery (Tatler, 2007). This bias has been observed not only in
63
association with static images, but also when freely exploring dynamic scenes (Hart et al.,
64
2009; Tseng, Carmi, Cameron, Munoz, & Itti, 2009). Furthermore, several studies have
65
reported a left/right horizontal asymmetry (Ossandón, Onat, & König, 2014) and a statistical
66
dependence on saccadic angles (Dorris, Taylor, Klein, & Munoz, 1999; Wilming, Harst,
67
Schmidt, & König, 2013). Jointly, these aspects explain a sizable fraction of the selection of
68
fixation points.
69
The spatial preference for the left visual field has been observed in many
70
behavioural studies. For example, in the chimeric faces task, participants have to choose
71
which half of mixed (happy & neutral) human faces pictures are happier. In general, the left
72
side of the pictures is preferred based on subject’s reports as well as the starting location and
73
number of fixation points (Butler & Harvey, 2005; Phillips & David, 1997). Another example
74
is the cancellation task, in which subjects cancel out targets that are distributed between
75
distractors as quickly and as accurately as possible. Subjects tend to have directional bias
76
toward the left visual hemispace (Rinaldi, Di Luca, Henik, & Girelli, 2014). In the SNARC
77
effect, people usually associate small numbers with the left hemispace and large numbers with
78
the right hemispace, which is associated with the habit of finger counting (Fischer, 2008;
79
Shaki & Fischer, 2008). The line bisection task, in which participants have to draw a vertical
80
line through the middle of a horizontal line, usually demonstrates a slight bias toward the left
81
(Bradshaw, Nettleton, Nathan, & Wilson, 1985; Rinaldi et al., 2014). Moreover, in the free
82
viewing task, in which subjects explore images freely, the subjects demonstrate an initial bias
83
toward the left side of the screen (Ossandón et al., 2014). What is interesting is that these
84
behavioural studies not only show a preference for the left hemispace, but they can also be
85
modulated to the opposite direction (from right to left) or reduced toward the centre when
86
performed by participants who are native RTL readers (Afsari, Ossandón, & König, 2016;
87
Dehaene, Bossini, & Giraux, 1993; Eviatar, 1997; Göbel, Shaki, & Fischer, 2011; Rashidi-
88
Ranjbar, Goudarzvand, Jahangiri, Brugger, & Loetscher, 2014; Rinaldi, Di Luca, Henik, &
89
Girelli, 2014).
90
Recently, we conducted series of eye-tracking experiments, which showed that
91
horizontal spatial bias is modulated not by the reading itself but by a habitual scanning
92
process. While exploring images freely, LTR readers did not show a change in the leftward
93
spatial bias after reading non-habitual texts (mirrored LTR texts that resembled the RTL texts
94
in the reading direction), nor after tracking RTL moving dots. On the other hand, LTR/RTL
95
bilinguals showed flexibility in changing the direction of the spatial bias according to the type
96
of text (LTR or RTL) they read prior to image exploration. This has been explained by the
97
process of developing two scanning direction habits. When starting to learn reading, a habit of
98
moving the eye toward the direction of the first word of the sentence is developed.
99
Sequentially, this long-term habit development affects the laterality of the visual attention
100
system to a certain extent (Afsari et al., 2016).
101
The dynamic nature of the spatial bias opens the door to investigate other biological
102
and cultural factors that might have an impact on the left/right visual spatial bias. One of the
103
factors that may contribute to the horizontal spatial bias is age. Several behavioural studies
104
have shown that older adults pay less attention to the local features of the images (Açık,
105
Sarwary, Schultze-Kraft, Onat, & König, 2010) and perform slower in visuospatial tasks than
106
younger ones (Robinson & Kertzman, 1990). They also preferred to focus on smaller regions
107
of space, contrary to young adults (Kosslyn, Brown, & Dror, 1999). Furthermore, the
108
HAROLD (Hemispheric Asymmetry Reduction in Older Adults) model suggested that the
109
activity of the prefrontal cortical area during cognitive performances is less lateralized in
110
older adults (Cabeza, 2002). Hence, during aging, the cognitive skills decline but with cortical
111
compensations (Li, Lindenberger, & Sikström, 2001; Madden, 2007; Robinson & Kertzman,
112
1990; Verhaeghen & Cerella, 2002).
113
Another suggested factor that could contribute to the left/right spatial bias is gender.
114
Sex differences were reported in landmark learning for virtual navigation (Andersen,
115
Dahmani, Konishi, & Bohbot, 2012; Chamizo, Artigas, Sansa, & Banterla, 2011). In addition,
116
female performance in the spatial attention task was superior to that of the male participants
117
when performing trials with endogenous cueing (Bayliss, Pellegrino, & Tipper, 2005; Merritt
118
et al., 2007). Many authors have attempted to explain the difference in results based on
119
evolutionary reasoning and the role of estrogen hormonal levels (Ecuyer-Dab & Robert, 2004;
120
Frischen, Bayliss, & Tipper, 2007; Robinson & Kertzman, 1990). Hence, we speculate that
121
the gender of the participant could have an effect on the horizontal spatial bias.
122
Moreover, we are interested in examining the role of second language proficiency
123
and the age of second language acquisition in the left/right spatial bias. Early bilinguals have
124
been shown to have a bilateral hemispheric interference, while late bilinguals who are less
125
proficient have higher interference of left hemisphere when conducting a dichotic listening
126
test (Hull & Vaid, 2006, 2007). In a different study, late bilinguals were reported to be less
127
accurate in English sentence judgment than early bilinguals (Birdsong & Molis, 2001). In the
128
same sequence, Yang and Lust showed that becoming an early bilingual can improve multiple
129
cognitive skills (Yang et al., 2011). Sequentially, we will focus on finding a correlation
130
between these two factors and the horizontal spatial bias.
131
In this paper, the goal is to investigate the interindividual differences among
132
RTL/LTR bilinguals’ and native LTR participant’s horizontal spatial bias by considering
133
multiple biological and cultural factors that might have an impact on the spatial attention. We
134
examined the interindividual variations by looking at the relationships between age, gender,
135
second language proficiency, and the age at which the second language was acquired, and
136
their relationship with the magnitude of the horizontal spatial bias. To our knowledge, this
137
study is the first study to examine the interindividual differences for a horizontal asymmetry
138
test.
139
Methods
140
Participants
141
Two groups of subjects participated in this study. The first RTL/LTR group consisted of 56
142
native RTL readers who learned a LTR language as a second language. The LTR/LTR group
143
consisted of 23 native LTR readers who learned a second LTR language. The LTR/LTR
144
group is also considered a control group. Part of the data was already used in Experiment 1(a)
145
and 2 in a previous study (Afsari, Ossandón, & König, 2016), which was extended by the
146
addition of 17 new subjects recruited specifically for this study. All participants performed the
147
experiment for 5-15 € or for student credit points. They filled out consent forms and
148
handedness tests (Edinbrugh Test; Oldfield, 1971) and performed a visual acuity test using
149
Sneller chart and dominant eye test (Miles Test; Miles, 1929). We verified that all of them
150
were right-handed and had normal or corrected-to-normal vision.
151
Stimuli
152
Visual stimuli in the form of texts and images were presented on a 21” CRT monitor
153
(Samsung SyncMaster 1100 DF, Samsung Electronics, Suwon, South Korea) at a refresh rate
154
of 85 Hz and a resolution of 960 x 1280 pixels. The texts served as primes for the images.
155
English or German texts were the LTR stimuli. The English texts were quoted from
156
Wikipedia and the British Broad Casting Corporation (BBC), while the German texts were
157
quoted from German newspapers. Arabic, Urdu or Persian texts were the RTL stimuli
158
obtained from Wikipedia. All the texts in the experiment were centred and designed to cover
159
the whole screen. The image stimuli were: 60 urban scenes, 60 natural scenes, and 60
160
artificial fractal images. The urban scenes were taken in Zürich (Onat, Açık, Schumann, &
161
König, 2014). The natural scenes were from a calibrated colour image database (Olmos &
162
Kingdom, 2004). The artificial fractals were self-similar computer-generated shapes from
163
different Web databases: Chaotic N-Space Network
164
(http://www.cnspace.net/html/fractals.html), Elena’s Fractal Gallery (http://www.elena-
165
fractals.it/, in http://web.archive.org), and Maria’s Fractal Explorer Gallery
166
(http://www.mariagrist.net/fegal). All the images were presented in either original condition
167
or mirrored condition in order to cancel the effect of the bias towards the image contents
168
(Ossandón et al., 2014).
169
Experimental paradigm
170
Text stimuli appeared for 12 seconds followed by 9 test images, each shown for 6 seconds.
171
One text stimulus followed by 9 test images formed an experimental block. The whole
172
experiment consisted of a total of 20 blocks: 20 texts as primes and 180 images from the three
173
different categories. The experimental paradigm was sorted as the following: five blocks
174
contained texts from the first language, and five blocks contained texts from the second
175
language. After a 5-minute optional break, the experiment continued with five blocks of
176
second language texts followed by five blocks of first language texts (Figure 1). Prior to each
177
image or text presentation, a fixation point appeared in the middle of the grey background to
178
restore the participant’s gaze toward the center of the screen and to avoid the influence of the
179
spatial location of the written texts on the viewing behaviour. Calibration of the eye-tracker
180
took place prior to the very first trial and after the break. The images were presented to one
181
participant in the original condition and for the following participant in the mirrored condition
182
to balance the images’ content spatial biases.
183
(Insert Figure 1 about here)
184
Figure 1: The experimental paradigm: A) 20 blocks, where each block consists of a 1st
185
language or 2nd language text followed by 9 images, form the experimental paradigm. B)
186
Examples of RTL text (upper panel) and LTR text (lower panel) used as primes prior to image
187
presentation.
188
Procedure
189
The participants attended the eye-tracking lab in the Institute of Cognitive Science at the
190
University of Osnabrück. First, they signed an informed consent. Then, they filled out the
191
questionnaires and sat 80 cm away from the monitor. They had been instructed to read the
192
texts silently at their normal reading speed and to explore the images freely without moving
193
their head. A head-mounted video-based eye-tracker system of binocular pupil tracking at 500
194
Hz (Eyelink II, SR Research Ltd, Mississauga, Canada) was used to record the eye
195
movements. The Osnabrück University Internal Review Board approved the experiment.
196
Data analysis
197
Custom-made Matlab and Python scripts were used to analyse the data. We used
198
SPSS for statistical evaluation. For the purpose of this paper, we consider the subjects the unit
199
of observation. For each subject, we pooled the data across the images and calculated the
200
difference between left and right horizontal coordinates during the first second of trial
201
duration. First, we extracted the fixation points and their horizontal positions for each subject.
202
Second, we classified the fixation points into two categories: fixation points after reading
203
texts in native languages and fixation points after reading texts in second languages. Third, we
204
separated the fixation points for the images from the fixation points for reading text primes.
205
Fourth, to calculate the percentage of horizontal bias from the centre of the screen, the total
206
number of fixation points on the left side of the images was subtracted from the total amount
207
of fixation points on the right side of the images. Then, the result was divided over the
208
summation of the right and left fixation points. We multiplied the results by 100 to get the
209
fraction amount of the horizontal bias. We ended with two measurements for each individual:
210
the fraction of bias after reading native language primes (RTL spatial bias) and the fraction of
211
bias after reading second language primes (LTR spatial bias). Ultimately, the data in the
212
following sections represents the fraction of horizontal bias on the images during the first
213
second of trial duration after reading primes for each individual subject.
214
Results
215
To assess the impact of RTL language primes on the leftward spatial bias, we
216
analysed the data of 56 native RTL/LTR readers and 23 native LTR/LTR readers after they
217
read texts in their native and second language, followed by a free-viewing task. Starting with
218
the RTL/LTR group, reading RTL texts as primes shifted the mean score of the horizontal
219
spatial bias to the right side of the screen (1.19 ± 24.42, mean ± standard deviation), while
220
reading LTR texts as primes shifted the mean score of the horizontal bias to the left side of the
221
screen (-10.63 ± 22.78). On the other hand, the LTR/LTR group demonstrated a strong
222
leftward shift for the horizontal spatial bias after reading native LTR and second LTR text
223
primes; (-34.09 ± 19.23) and (-35.81 ± 17.65) respectively. Additionally, a one sample t-test
224
was conducted to determine whether a statistically significant difference existed between each
225
group and the no bias state (zero score). RTL/LTR group showed no significant bias toward
226
the right side of the monitor after primed with texts from their native language compared to
227
the centre of the monitor (t(55)= 0.365, p = 0.717). However, the LTR/LTR group showed a
228
significant leftward bias after primed with texts from their native language compared to the
229
center of the monitor (t(22) = -8.51, p≤0.001). Furthermore, an independent t-test indicated
230
that the mean of the RTL/LTR group was significantly different than the mean of LTR/LTR
231
group after reading texts in their native language (t (77) = 5.781, p < 0.01). Figure 2 shows
232
the distribution of spatial bias for the two groups after being exposed to native language
233
primes. The distribution of the data for the spatial bias of the RTL/LTR group is broad and
234
shifted toward the center of the images, compared to the distribution of the data for LTR/LTR
235
group, which is narrower and shifted to the left side of the screen.
236
(Insert Figure 2 about here)
237
Figure 2: Distribution of horizontal spatial bias for the RTL/LTR group and LTR/LTR group
238
after reading texts in their native language. The positive values on the abscissa represent the
239
fraction of bias toward the right, and the negative values represent the fraction of bias toward
240
the left from the viewer’s perspective.
241
242
Because subjects originated from 14 different countries, their heterogeneous
243
multilingual and cultural backgrounds could influence the direction and the magnitude of the
244
horizontal spatial bias. Therefore, the next goal is to assess the effect of biological and
245
cultural factors on the individual bias scores for the two groups presented in Figure 2
246
specifically.
247
Effect of age
248
To begin with, we studied the correlation between the age of the participants and the
249
horizontal bias after reading native language primes. For the RTL/LTR group, the age of the
250
participants ranged between 21 and 60 years. The data analysis showed no significant
251
correlation between RTL spatial bias and age (r (56) = 0.110, p = 0.418). As for the LTR/LTR
252
control group, the age of the participants ranged between 18 and 27 years. Again, no
253
significant correlation was detected between the age factor and the magnitude of bias after
254
reading texts in their native language (r (23) = 0.172, p = 0.432) (Figure 3). Therefore, given
255
our current sample size, we no indication that age did not contribute to the interindividual
256
difference in the RTL/LTR group.
257
(Insert Figure 3 about here)
258
Figure 3: The scatter plot represents the correlation between the magnitude of the horizontal
259
spatial bias after reading native language texts and the age of the participants.
260
Effect of gender
261
Likewise, we were interested in the relationship between the gender of the participants and the
262
interindividual differences in the horizontal spatial bias. For the RTL/LTR group, out of 56
263
participants who performed the task, 10 were female (Table 1). After testing the normality,
264
the mean score of the male group (2.41 ± 24.72) was not significantly different than the mean
265
of the female group (-4.40 ± 23.37) (t (54) = 0.796, p= 0.737). For the LTR/LTR control
266
group, 10 out of 23 participants were female, and the results showed no significant difference
267
in the mean score of the bias between males (-35.79 ± 20.79) and females (-31.87 ± 17.82)
268
after reading texts in their native language (t (21) = -0.48, p=0.639). Thus, given our current
269
sample size, we did not observe evidence of the effect of gender on the manipulation of the
270
horizontal spatial bias.
271
Effect of second language proficiency
272
The following step was to investigate if there is an influence of the proficiency of second
273
language on the horizontal spatial bias. In the questionnaire, subjects evaluated their second
274
language proficiency by choosing the best choice from four options: Excellent, Very Good,
275
Good, and Poor. One-way ANOVA showed that there were no statistically significant
276
differences between the means of different levels of second language proficiency (F (3, 52) =
277
0.263, p = 0.852). For the LTR/LTR group, all participants evaluated themselves as either
278
excellent or very good in their second language proficiency; hence, the mean score for the
279
horizontal spatial bias of the excellent group (-29.03 ± 16.71) was not significantly different
280
from the mean score of the very good group (-41.95 ± 21.17) (t(21) =1.631, p = 0.118).
281
For additional measurement of the second language proficiency, the median heights
282
of reading the second language text levels were calculated. We assume that the amount of
283
reading represents the level of proficiency. That means, the more proficient the reader is, the
284
more lines of the text are read, and the greater the value of the median height. For each
285
subject, the median score was calculated for the vertical fixation points extracted from the
286
second language text primes. Statistically, there was no correlation between the median height
287
for reading LTR texts and the RTL horizontal spatial bias (r (56) = -0.198, p = 0.143).
288
Consequently, for this particular sample, we do not have evidence that the large variance in
289
the horizontal spatial bias between RTL/LTR subjects was modulated by the second language
290
proficiency.
291
Effect of age of second language acquisition
292
We also evaluated the correlation between the age of the participants when they acquired their
293
second language and the magnitude of the horizontal spatial bias. For the RTL/LTR group,
294
out of 56 participants, 45 participants answered this question. There was no correlation
295
between the age at which the subjects learned to read/write their second language and the
296
RTL horizontal spatial bias (r (46) = 0.043, p = 0.774) (Figure 4). Hence, given the current
297
sample size, the age at which participants required their second language did not contribute to
298
the interindividual variations of the rightward spatial bias for the RTL/LTR group.
299
(Insert Figure 4 about here)
300
Figure 4: The scatter graph shows the correlation between the age at which native RTL
301
readers acquired a LTR language and the RTL horizontal spatial bias.
302
303
The calculated statistical power
304
The data presented earlier did not show to have a significant impact on the
305
horizontal spatial bias. A further step is taken in this research by measuring the effect size of
306
the sample above and then estimate the sample size required to reach a higher statistical
307
power and reduce type II error. For this purpose, we used G*power software with the results
308
that are obtained earlier to estimate the number of subjects necessary in order to detect a
309
possible effect.
310
For the correlative analysis as applied e.g. with the age at test or the age at second
311
language acquisition, the statistical power for a sample of 56 subjects to observe a significant
312
correlation of size 0.11 is 0.47. To achieve a statistical power of 0.80, 120 subjects are
313
needed.
314
For the binary divisions like the gender and second language proficiency factor, the
315
statistical power to observe a difference with an effect size of 0.28 is 0.20. To achieve a
316
statistical power of 0.80, an estimated sample size of 524 subjects are required.
317
Discussion
318
In this work, we analysed eye-tracking data of 56 native RTL/LTR readers and 23
319
LTR/LTR readers who performed free image exploration after being primed with texts. The
320
LTR/LTR group showed a strong leftward bias in the first second of image exploration. The
321
result of the LTR/LTR group is consistent with a previous eye-tracking study that showed an
322
initial leftward bias in a free viewing task without reading primes (Ossandón, Onat, & König,
323
2014). Hence, we claim that when native LTR readers read LTR primes, it strengthens the
324
natural leftward bias reported above.
325
On the other hand, the RTL/LTR group showed not only a rightward shifted
326
horizontal spatial bias after reading RTL texts, but also larger variance in comparison to the
327
LTR/LTR group. This raises the question of whether it would be possible to identify a factor
328
within the RTL group explaining the wide dispersion of the data. With this goal in mind, we
329
studied the relationship between the RTL spatial bias after reading RTL texts for the
330
RTL/LTR group and several parameters reported by the participants: age, gender, second
331
language proficiency, and the age at which the second language was acquired. However, we
332
found no significant correlation between these parameters and the spatial bias. Thus, the
333
higher variability might be explained by the fact that native RTL readers have two reading
334
habits that are conflicting in terms of reading direction, whereas the control group does not
335
have such a conflict.
336
The first factor we investigated in this paper was the age. We noticed no influence
337
of the “young and middle age” spectrum on the RTL spatial bias. One point to mention is that
338
we were able to recruit only one participant older than 50 years because of the limited
339
geographical area where it is difficult to find native RTL readers in general. In the study by
340
Açik, children, young adults, and older adults were requested to view natural and complex
341
images freely then perform a patch recognition task. Older adults (> 72 years old) were less
342
dependent on the low features of the images compared to children and young adults, and
343
relied more on a top-down mechanism, which suggests different strategies in exploring the
344
scenes (Açık et al., 2010). Thus, for this specific sample of the RTL/LTR group, the age
345
factor did not influence the horizontal spatial bias.
346
Gender was the second factor we analysed. We found no correlation with the RTL
347
spatial bias. The data of 10 females was compared to the data of 46 males and no significant
348
difference was noticed indicating no influence of gender factor on the general RTL bias. This
349
is in line with our previous report of there being no difference in viewing biases between
350
genders (Ossandón, Onat, & König, 2014).
351
Second language proficiency and age of second language acquisition are usually
352
investigated together, and their effect on several cognitive skills has been reported (Birdsong
353
& Molis, 2001; Hull & Vaid, 2006; Yang et al., 2011). However, in among this sample size,
354
we did not find an effect of these two factors on the RTL horizontal spatial bias for native
355
RTL/LTR readers. This might be explained by the strength of habit of reading direction,
356
regardless of the linguistic component. In contrast, Birdsong and Molis studied the
357
relationship between judging English sentences produced by early Spanish bilinguals (≤ 16
358
years old) and late Spanish bilinguals (> 17 years old) and found a negative correlation
359
between the late bilingual group and accurate English sentence judgment (Birdsong & Molis,
360
2001). Similarly, Hull and Vaid (2006 and 2007) conducted meta-analysis studies to compare
361
monolinguals with bilinguals and early bilinguals with later bilinguals in different
362
hemispheric laterality tasks and concluded that early bilinguals (< 6 years old) have bilateral
363
hemispheric interference. In addition, late bilinguals who are also less proficient have higher
364
interference of the left hemisphere when conducting a dichotic listening test (Hull & Vaid,
365
2006, 2007). In the same way, Yang and his team (2001) showed that becoming a bilingual
366
child at 4 years old can improve multiple cognitive skills (Yang et al., 2011). Although these
367
studies show an influence of the age of the participants and their second language proficiency
368
on some cognitive skills, these two factors did not demonstrate an impact on the horizontal
369
spatial bias.
370
The post-hoc statistical power analysis for the four factors showed to have a low
371
statistical power, which indicates that there could be an undetected effect on the horizontal
372
spatial bias. In order to increase the statistical power, larger sample size is required (from 120
373
to 524 subjects, depending on the relevant effect size) which is much more that we could
374
recruit. In a country where RTL language is uncommon, we are not able get these numbers.
375
Therefore, we made our data open to the science platform by uploading the data that we
376
obtained to the Open Science Framework (OSF) to open the opportunity for scientific
377
collaboration and replication for the study to obtain low type II error for these factors; please
378
visit the link: (https://osf.io/tnxme/).
379
Because none of the factors evaluated explained the larger bias variance of the RTL
380
subject, we discuss next the role of other factors that have not yet been evaluated.
381
Habit strength factor
382
Habit is an automatic response that requires no involvement of consciousness (Lally,
383
van Jaarsveld, Potts, & Wardle, 2010). It is activated immediately at the moment the cue that
384
is linked to the specific habit is displayed. Counting fingers and the SNARC effect are two
385
behaviours that have been linked directly to habit formation. For instance in a cross-cultural
386
study for the finger counting test, while LTR readers preferred to start counting with the left
387
thumb, the majority of RTL readers started counting with the right little finger (Lindemann,
388
Alipour, & Fischer, 2011). In the counting coins test, four identical coins are arranged in a
389
linear array. The task is to count the coins loudly while pointing at them. The main influence
390
on the counting direction was the habitual reading direction. Interestingly, the illiterate group
391
and the mixed language group (where letters are written RTL but numbers are written LTR)
392
did not show a counting direction preference (Shaki, Fischer, & Göbel, 2012). Additionally,
393
finger counting habit is associated with the SNARC effect. When the subjects performed the
394
finger counting test and the SNARC test (parity judgment test), the SNARC effect was
395
stronger among left counters, and a reversed SNARC effect was reported among right
396
counters (Fischer, 2008). Thus, culture can reshape behaviours through learning and
397
practicing.
398
The strength of a habit has a corresponding impact on the ability to change a
399
behaviour. For instance, when studying smokers with different levels of habit strength who
400
were trained to break their smoking habit, the effectiveness of minimizing the smoking habit
401
was dependent on the strength of the habit (Webb, Sheeran, & Luszczynska, 2009). Hence,
402
habit strength affects the process of changing the habit.
403
Thus, the reversed bias and larger variability of native RTL participants might be
404
explained by the result of averaging two strong habits with spatial biases effects that
405
antagonize each other. Assuming the existence of two habits in this task, LTR reading habit
406
and RTL reading habit, it could be that either these two habits are antagonizing each other,
407
ending with a net result of a certain percentage of RTL spatial bias, or it could be the strength
408
of RTL reading habit by itself that controls the RTL spatial bias. To further investigate this
409
point, a specific reading habit strength measurement would need to be developed to evaluate
410
the dependence between the magnitude of horizontal spatial bias with both RTL and LTR
411
reading habits.
412
This result supports the notion that learning to read at an early age (around 6 or 7
413
years) will form a habit that is difficult to deteriorate. Cunningham ran a 10-year follow-up
414
study for the 1st-grade students to measure their language skills and noticed that learning to
415
read quickly can affect the lifetime reading habit (Cunningham & Stanovich, 1997). Because
416
of this, we believe that another country’s cultural effect will not overcome the effect of the
417
native country’s influence.
418
Variability of brain structure and function among healthy individuals
419
The hypothesized effect of reading habit might be related to patterns of lateralization
420
and connectivity between left visual hemifields and the right hemisphere for attentional
421
mechanisms (Kastner & Ungerleider, 2000). A leftward bias induced by a LTR habit
422
coincides with and reinforces a leftward (in the visual field) attentional bias which is probably
423
secondary to the right lateralization of attentional networks. However, in RTL readers, these
424
two mechanisms have opposite consequences and therefore lead to a weaker and more
425
variable spatial bias. Thus, the low variability among LTR readers would be explained by the
426
congruence of lateralization of cortical regions and reading habits. The high variability of
427
spatial bias among RTL readers would be explained by the lack of congruence of the effect of
428
module lateralization on viewing bias and the effect of reading habits on attentional bias.
429
Conclusion
430
To summarize, our hypothesis suggests that the power of reading direction habit is
431
strong enough to manipulate horizontal spatial bias. We assume that developing a ”habit” to
432
scan a text with the eyes from one direction to the other, beginning in 1st grade, is an
433
important factor to implant a dynamic horizontal bias. Although this study is considered a
434
small multicultural experiment due to the diversity of the subjects’ cultural background, we
435
did not find any indication of the effect of the exogenous factors on the interindividual
436
variance on the rightward horizontal spatial bias such as age, gender, second language
437
proficiency, and the age at which the second language was acquired. As suggested by
438
previous works, we assume that the hemispheric lateralization of the spatial attention system
439
might be the leading role for the leftward preference of the naturally existing horizontal bias
440
(Afsari et al., 2016; Ossandón et al., 2014). In addition, we suggest the role of the scanning
441
habit on modulating the leftward preference of the horizontal bias. Lastly, we suggest two
442
other factors that may influence the wide variability of the RTL spatial bias: the strength of
443
the habit and interindividual differences at the cortical level for language/attention.
444
For future work, we suggest replicating similar studies by recruiting more subjects
445
who are RTL/LTR bilinguals. In addition, recruiting illiterates can add a lot to understand the
446
nature of the spatial bias without the effect of reading habit. Reporting the natural viewing
447
bias among illiterates will generalize the notion of the spatial attention with a minimal effect
448
of habitual scanning influence. RTL monolinguals could also contribute to this type of study
449
by revealing the effect of different scanning habits on the visual spatial bias without
450
conflicting with contrary scanning habit.
451
Acknowledgement
452
We thank Matti Krüger and Mattias Hampel Holzer for their technical assistance in the early
453
phase of the project.
454
References
455
Açık, A., Sarwary, A., Schultze-Kraft, R., Onat, S., & König, P. (2010). Developmental
456
Changes in Natural Viewing Behavior: Bottom-Up and Top-Down Differences
457
between Children, Young Adults and Older Adults. Frontiers in Psychology, 1.
458
https://doi.org/10.3389/fpsyg.2010.00207
459
Afsari, Z., Ossandón, J. P., & König, P. (2016). The dynamic effect of reading direction habit
460
on spatial asymmetry of image perception. Journal of Vision, 16(11), 8–8.
461
https://doi.org/10.1167/16.11.8
462
Andersen, N. E., Dahmani, L., Konishi, K., & Bohbot, V. D. (2012). Eye tracking, strategies,
463
and sex differences in virtual navigation. Neurobiology of Learning and Memory,
464
97(1), 81–89. https://doi.org/10.1016/j.nlm.2011.09.007
465
Bayliss, A. P., Pellegrino, G. di, & Tipper, S. P. (2005). Sex differences in eye gaze and
466
symbolic cueing of attention. The Quarterly Journal of Experimental Psychology
467
Section A, 58(4), 631–650. https://doi.org/10.1080/02724980443000124
468
Birdsong, D., & Molis, M. (2001). On the Evidence for Maturational Constraints in Second-
469
Language Acquisition. Journal of Memory and Language, 44(2), 235–249.
470
https://doi.org/10.1006/jmla.2000.2750
471
Bradshaw, J. L., Nettleton, N. C., Nathan, G., & Wilson, L. (1985). Bisecting rods and lines:
472
Effects of horizontal and vertical posture on left-side underestimation by normal
473
subjects. Neuropsychologia, 23(3), 421–425. https://doi.org/10.1016/0028-
474
3932(85)90029-6
475
Buschman, T. J., & Miller, E. K. (2007). Top-Down Versus Bottom-Up Control of Attention
476
in the Prefrontal and Posterior Parietal Cortices. Science, 315(5820), 1860–1862.
477
https://doi.org/10.1126/science.1138071
478
Butler, S. H., & Harvey, M. (2005). Does inversion abolish the left chimeric face processing
479
advantage? Neuroreport, 16(18), 1991–1993.
480
Cabeza, R. (2002). Hemispheric asymmetry reduction in older adults: The HAROLD model.
481
Psychology and Aging, 17(1), 85–100. https://doi.org/10.1037/0882-7974.17.1.85
482
Chamizo, V. D., Artigas, A. A., Sansa, J., & Banterla, F. (2011). Gender differences in
483
landmark learning for virtual navigation: The role of distance to a goal. Behavioural
484
Processes, 88(1), 20–26. https://doi.org/10.1016/j.beproc.2011.06.007
485
Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention
486
in the brain. Nature Reviews Neuroscience, 3(3), 201–215.
487
https://doi.org/10.1038/nrn755
488
Cunningham, A. E., & Stanovich, K. E. (1997). Early reading acquisition and its relation to
489
reading experience and ability 10 years later. Developmental Psychology, 33(6), 934–
490
945. https://doi.org/10.1037/0012-1649.33.6.934
491
DeAngelus, M., & Pelz, J. B. (2009). Top-down control of eye movements: Yarbus revisited.
492
Visual Cognition, 17(6–7), 790–811. https://doi.org/10.1080/13506280902793843
493
Dehaene, S., Bossini, S., & Giraux, P. (1993). The mental representation of parity and number
494
magnitude. Journal of Experimental Psychology: General, 122(3), 371–396.
495
https://doi.org/10.1037/0096-3445.122.3.371
496
Dorris, M. C., Taylor, T. L., Klein, R. M., & Munoz, D. P. (1999). Influence of Previous
497
Visual Stimulus or Saccade on Saccadic Reaction Times in Monkey. Journal of
498
Neurophysiology, 81(5), 2429–2436.
499
Ecuyer-Dab, I., & Robert, M. (2004). Have sex differences in spatial ability evolved from
500
male competition for mating and female concern for survival? Cognition, 91(3), 221–
501
257. https://doi.org/10.1016/j.cognition.2003.09.007
502
Eviatar, Z. (1997). Language experience and right hemisphere tasks: the effects of scanning
503
habits and multilingualism. Brain and Language, 58(1), 157–173.
504
https://doi.org/10.1006/brln.1997.1863
505
Fischer, M. H. (2008). Finger counting habits modulate spatial-numerical associations.
506
Cortex, 44(4), 386–392. https://doi.org/10.1016/j.cortex.2007.08.004
507
Frischen, A., Bayliss, A. P., & Tipper, S. P. (2007). Gaze cueing of attention: Visual attention,
508
social cognition, and individual differences. Psychological Bulletin, 133(4), 694.
509
https://doi.org/10.1037/0033-2909.133.4.694
510
Göbel, S. M., Shaki, S., & Fischer, M. H. (2011). The Cultural Number Line: A Review of
511
Cultural and Linguistic Influences on the Development of Number Processing.
512
Journal of Cross-Cultural Psychology, 42(4), 543–565.
513
https://doi.org/10.1177/0022022111406251
514
Hart, B. M. ’t, Vockeroth, J., Schumann, F., Bartl, K., Schneider, E., König, P., & Einhäuser,
515
W. (2009). Gaze allocation in natural stimuli: Comparing free exploration to head-
516
fixed viewing conditions. Visual Cognition, 17(6–7), 1132–1158.
517
https://doi.org/10.1080/13506280902812304
518
Hull, R., & Vaid, J. (2006). Laterality and language experience. Laterality: Asymmetries of
519
Body, Brain and Cognition, 11(5), 436–464.
520
https://doi.org/10.1080/13576500600691162
521
Hull, R., & Vaid, J. (2007). Bilingual language lateralization: A meta-analytic tale of two
522
hemispheres. Neuropsychologia, 45(9), 1987–2008.
523
https://doi.org/10.1016/j.neuropsychologia.2007.03.002
524
Ignashchenkova, A., Dicke, P. W., Haarmeier, T., & Thier, P. (2004). Neuron-specific
525
contribution of the superior colliculus to overt and covert shifts of attention. Nature
526
Neuroscience, 7(1), 56–64. https://doi.org/10.1038/nn1169
527
Itti, L., & Koch, C. (2000). A saliency-based search mechanism for overt and covert shifts of
528
visual attention. Vision Research, 40(10–12), 1489–1506.
529
https://doi.org/10.1016/S0042-6989(99)00163-7
530
Itti, L., & Koch, C. (2001). Computational modelling of visual attention. Nature Reviews.
531
Neuroscience, 2(3), 194–203. https://doi.org/10.1038/35058500
532
Kaspar, K., & König, P. (2011). Overt attention and context factors: the impact of repeated
533
presentations, image type, and individual motivation. PloS One, 6(7), e21719.
534
https://doi.org/10.1371/journal.pone.0021719
535
Kastner, S., & Ungerleider, L. G. (2000). Mechanisms of Visual Attention in the Human
536
Cortex. Annual Review of Neuroscience, 23(1), 315–341.
537
https://doi.org/10.1146/annurev.neuro.23.1.315
538
Kollmorgen, S., Nortmann, N., Schröder, S., & König, P. (2010). Influence of low-level
539
stimulus features, task dependent factors, and spatial biases on overt visual attention.
540
PLoS Computational Biology, 6(5), e1000791.
541
https://doi.org/10.1371/journal.pcbi.1000791
542
Kosslyn, S. M., Brown, H. D., & Dror, I. E. (1999). Aging and the Scope of Visual Attention.
543
Gerontology, 45(2), 102–109. https://doi.org/10.1159/000022071
544
Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. (2010). How are habits
545
formed: Modelling habit formation in the real world. European Journal of Social
546
Psychology, 40(6), 998–1009. https://doi.org/10.1002/ejsp.674
547
Li, S.-C., Lindenberger, U., & Sikström, S. (2001). Aging cognition: from neuromodulation to
548
representation. Trends in Cognitive Sciences, 5(11), 479–486.
549
https://doi.org/10.1016/S1364-6613(00)01769-1
550
Lindemann, O., Alipour, A., & Fischer, M. H. (2011). Finger Counting Habits in Middle
551
Eastern and Western Individuals: An Online Survey. Journal of Cross-Cultural
552
Psychology, 42(4), 566–578. https://doi.org/10.1177/0022022111406254
553
Madden, D. J. (2007). Aging and Visual Attention. Current Directions in Psychological
554
Science, 16(2), 70–74. https://doi.org/10.1111/j.1467-8721.2007.00478.x
555
Merritt, P., Hirshman, E., Wharton, W., Stangl, B., Devlin, J., & Lenz, A. (2007). Evidence
556
for gender differences in visual selective attention. Personality and Individual
557
Differences, 43(3), 597–609. https://doi.org/10.1016/j.paid.2007.01.016
558
Miles, W. (1929). Ocular dominance demonstrated by unconscious sighting. Journal of
559
Experimental Psychology, 12(2), 113–126. https://doi.org/10.1037/h0075694
560
Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory.
561
Neuropsychologia, 9(1), 97–113. https://doi.org/10.1016/0028-3932(71)90067-4
562
Olmos, A., & Kingdom, F. A. A. (2004). A biologically inspired algorithm for the recovery of
563
shading and reflectance images. Perception, 33(12), 1463–1473.
564
Onat, S., Açık, A., Schumann, F., & König, P. (2014). The contributions of image content and
565
behavioral relevancy to overt attention. PloS One, 9(4), e93254.
566
https://doi.org/10.1371/journal.pone.0093254
567
Ossandón, J. P., Onat, S., Cazzoli, D., Nyffeler, T., Müri, R., & König, P. (2012). Unmasking
568
the contribution of low-level features to the guidance of attention. Neuropsychologia,
569
50(14), 3478–3487. https://doi.org/10.1016/j.neuropsychologia.2012.09.043
570
Ossandón, J. P., Onat, S., & König, P. (2014). Spatial biases in viewing behavior. Journal of
571
Vision, 14(2), 1-26. https://doi.org/10.1167/14.2.20
572
Phillips, M. L., & David, A. S. (1997). Viewing Strategies for Simple and Chimeric Faces:
573
An Investigation of Perceptual Bias in Normals and Schizophrenic Patients Using
574
Visual Scan Paths. Brain and Cognition, 35(2), 225–238.
575
https://doi.org/10.1006/brcg.1997.0939
576
Rashidi-Ranjbar, N., Goudarzvand, M., Jahangiri, S., Brugger, P., & Loetscher, T. (2014). No
577
horizontal numerical mapping in a culture with mixed-reading habits. Frontiers in
578
Human Neuroscience, 8. https://doi.org/10.3389/fnhum.2014.00072
579
Rinaldi, L., Di Luca, S., Henik, A., & Girelli, L. (2014). Reading direction shifts visuospatial
580
attention: An Interactive Account of attentional biases. Acta Psychologica, 151, 98–
581
105. https://doi.org/10.1016/j.actpsy.2014.05.018
582
Robinson, D. L., & Kertzman, C. (1990). Visuospatial attention: Effects of age, gender, and
583
spatial reference. Neuropsychologia, 28(3), 291–301. https://doi.org/10.1016/0028-
584
3932(90)90022-G
585
Shaki, S., & Fischer, M. H. (2008). Reading space into numbers – a cross-linguistic
586
comparison of the SNARC effect. Cognition, 108(2), 590–599.
587
https://doi.org/10.1016/j.cognition.2008.04.001
588
Shaki, S., Fischer, M. H., & Göbel, S. M. (2012). Direction counts: A comparative study of
589
spatially directional counting biases in cultures with different reading directions.
590
Journal of Experimental Child Psychology, 112(2), 275–281.
591
https://doi.org/10.1016/j.jecp.2011.12.005
592
Tseng, P.-H., Carmi, R., Cameron, I. G. M., Munoz, D. P., & Itti, L. (2009). Quantifying
593
center bias of observers in free viewing of dynamic natural scenes. Journal of Vision,
594
9(7), 4. https://doi.org/10.1167/9.7.4
595
Verhaeghen, P., & Cerella, J. (2002). Aging, executive control, and attention: a review of
596
meta-analyses. Neuroscience & Biobehavioral Reviews, 26(7), 849–857.
597
https://doi.org/10.1016/S0149-7634(02)00071-4
598
Webb, T. L., Sheeran, P., & Luszczynska, A. (2009). Planning to break unwanted habits:
599
Habit strength moderates implementation intention effects on behaviour change.
600
British Journal of Social Psychology, 48(3), 507–523.
601
https://doi.org/10.1348/014466608X370591
602
Wilming, N., Harst, S., Schmidt, N., & König, P. (2013). Saccadic Momentum and
603
Facilitation of Return Saccades Contribute to an Optimal Foraging Strategy. PLOS
604
Computational Biology, 9(1), e1002871. https://doi.org/10.1371/journal.pcbi.1002871
605
Yang, S., Yang, H., & Lust, B. (2011). Early childhood bilingualism leads to advances in
606
executive attention: Dissociating culture and language. Bilingualism: Language and
607
Cognition, 14(03), 412–422. https://doi.org/10.1017/S1366728910000611
608
Parameters
RTL/LTR group
LTR/LTR group
Gender
Female
10
10
Male
46
13
Second language proficiency
Excellent
23
14
Very good
15
9
Good
11
0
Poor
7
0
Table 1. The table represents the different sample sizes for the subgroups of RTL/LTR group and
LTR/LTR group tested for several parameters. (*) (The age when acquired the second language)
parameters was answered by 46 participants only.
5x 1st language
Blocks
5x 1st langauge
Blocks
5x 2nd language
Blocks
5x 2nd language
Blocks
5x 1st language
Blocks
Block = 1x Text + 9x Images
(A) (B)
Horizontal Spatial Bias (%)
-100 -80 -60 -40 -20 0 20 40 60 80
Frequency
0
2
4
6
8
10
12
14
16
18
20 RTL/LTR
gaussian fit
LTR/LTR
gaussian fit
Age when acquired the second language
0 5 10 15 20 25 30
Horizontal Spatial Bias (%)
-80
-60
-40
-20
0
20
40
60
Age
15 20 25 30 35 40 45 50 55 60 65
Horizontal Spatial Bias (%)
-80
-60
-40
-20
0
20
40
60 LTR/LTR
linear regression
RTL/LTR
linear regression
609
