Visual cognitive function is important in the construction of executive function in daily life. Perception of visual number form (e.g. Arabic digits) and numerosity (numeric magnitude) is of interest to cognitive neuroscientists. Neural correlates and the functional measurement of number representations are complex events when their semantic categories are assimilated together with concepts of shape and color. Color perception can be processed further to modulate visual cognition. The Ishihara pseudoisochromatic plates are one of the best and most common screening tools for basic red-green color vision testing. However, there has been little study of visual cognitive function assessment using such pseudoisochromatic plates. 25 healthy normal trichromat volunteers were recruited and studied using a 128-sensor net to record event-related electroencephalogram. Subjects were asked to respond by pressing numbered buttons when they saw the number and non-number plates of the Ishihara color vision test. Amplitudes and latencies of N100 and P300 event related potential components were analyzed from 19 electrode sites in the international 10-20 system. A brain topographic map, cortical activation patterns, and Granger causation (effective connectivity) were analyzed from 128 electrode sites. No significant differences between N100 event related potential components for either stimulus indicates early selective attention processing was similar for number and non-number plate stimuli, but non-number plate stimuli evoked significantly higher amplitudes, longer latencies of the P300 event related potential component with a slower reaction time compared to number plate stimuli imply the allocation of attentional load was more in non-number plate processing. A different pattern of the asymmetric scalp voltage map was noticed for P300 components with a higher intensity in the left hemisphere for number plate tasks and higher intensity in the right hemisphere for non-number plate tasks. Asymmetric cortical activation and connectivity patterns revealed that number recognition occurred in the occipital and left frontal areas where as the consequence was limited to the occipital area during the non-number plate processing. Finally, results demonstrated that the visual recognition of numbers dissociates from the recognition of non-numbers at the level of defined neural networks. Number recognition was not only a process of visual perception and attention, but was also related to a higher level of cognitive function, that of language.
The brain and eye analyze the visual perception of brightness, movement, shape, and color of objects [1, 2] even the mental imagery of motion and static visual features [3, 4]. The three basic color receptors of blue, green, and red have their own photo pigments that react to light to evoke receptor potentials in the visual pathway. The Ishihara test is one type of test where a chart can be employed to detect weak or missing L-cones (red) and M-cones (green), but not S-cones (blue) [5, 6]. Different colored plates are used in this test, where primary and secondary dots form number or non-number shapes. Secondary dots play a role in the background of the plates [7]. Subjects are classified as having impaired color vision if two or more plates are recognized incorrectly [8]. Impaired color vision influences visual cognitive functions involved in object recognition, color memory, color consistency, and discounting of the illuminant [9, 10]. However, visual information processing can be divided into two general pathways: a neural pathway for vision from retina to cortex and a multi-synaptic corticocortical pathway for visual streams. The former consist of the retina, optic nerve, optic chiasm, optic tract, lateral geniculate body (LGN) of the thalamus, geniculocalcarine tract or optic radiations, and primary visual cortex (V1, also known as Brodmann's area 17). These anatomical visual pathways are of great interest to ophthalmologists and neurologists following injury or impairment of vision. The latter consist of the so-called `ventral stream' (also called the what pathway) and `dorsal stream' (also called the where pathway) [11, 12], which have attracted much attention from cognitive neuroscientists. The ventral stream originates from V1 of the occipital lobe and spreads along the ventral surface of the temporal lobe (occipitotemporal). It is involved in the processing of the recognition of objects, forms, face, color, and numerals. Conversely, the dorsal stream also originates from V1 and spreads along the dorsal surface of the parietal lobe (occipitoparietal). It is involved in the processing of the spatial location of objects. Likewise, numerical information from a visual scene is encoded and processed along the occipitoparietal (dorsal stream) pathway and finally projected to number-selective circuits of the bilateral intraparietal sulcus [13]. Such hemispheric activation or lateralization, identified by the triple code model (TCM), also originates not only from number processing but also from arithmetical skill processing which has been assessed recently in adults by functional transcranial doppler ultrasonography [14]. The TCM of number processing is a theoretical framework proposed to be organized through three circuits - a linguistically mediated verbal coding circuit associated with the left hemisphere, a visual number coding circuit for recognition of Arabic digits associated with the bilateral fusiform and lingual regions of the ventral stream of object recognition, and a bilateral number magnitude (cardinal value/numerosity) coding circuit [15].
Alternatively, cognitive function, can be simply described as the mental process that takes internal or external input and transforms, minimizes, elaborates, stores, retrieves, and finally utilizes it [16]. The term cognition covers a variety of functions such as attention, memory, learning, calculating, problem-solving, decision-making, reasoning, and planning [17,18,19]. Among all those sub-components of cognitive function, attention has a large role in regulating cognitive function [20]. The brain areas activated by visual stimuli have been investigated by different neuroimaging modalities such as MEG, fMRI, and PET, among others. A recent review article [21] of a functional neuroimaging meta-analysis highlighted that a number form area, proposed to be specialized for Arabic numerical processing, is in the right inferior temporal gyrus, bilateral parietal, and superior and inferior right frontal regions. Event-related potential (ERP) studies provide another method to measure visual cognitive function (attention) at the electrophysiological level. Due to higher temporal resolution, the electroencephalogram (EEG)/ERP has great importance in cognitive neuroscience research, as has fMRI, which has better spatial resolution. Moreover, using electrophysiological techniques, the connectivity of the brain has recently received increased attention, which has supported the development of brain cognition and action that can determine a temporal correlation [22].
In this study, a stimulus is being employed that has the capacity to simultaneously activate the processing of colored numbers and non-numbers in the brain to enable determination of how the brain processes activation, connectivity, integration, or segregation of different visual features in the visual system. Such a stimulus could provide an important test tool for understanding certain brain disorders, such as attention deficit disorder, and connectivity disorders in different sensory areas of the brain. Several investigators consider that connectivity can be reformed for various reasons in mental tasks [23], by sleep [24], by learning [25], and by consciousness [26]. Lower connectivity has been identified in the fronto-parietal network among subjects with reading difficulty [27, 28] and this network is vital for visual attention during reading [29]. There is also evidence that enhanced connectivity within visual processing components is further increased between visual processing and temporal-occipital components [30]. Moreover, left temporal-parietal connectivity is stronger during the processing of arithmetic principles and language than during computation, whereas parietal-occipital connectivities are stronger during computation than during the processing of arithmetic principles and language [31]. One example of connectivity disorders in different sensory areas of the brain is synesthesia, a neurological phenomenon in which an individual perceives a specific color when seeing a specific number [32]. Synesthesia can also occur when a letter is interpreted as a color and a taste for a shape. Grapheme synesthesia occurs when a color is experienced upon seeing letters and numbers. It is a phenomenon arising from cross connectivity between a color area and a number area in the fusiform gyrus. Research has verified that Arabic numerals but not Roman numerals induce color. This is interpreted as demonstrating that the visual image, or "grapheme" of a number that is predominant, rather than the numerical concept [33, 34]. Ramachandran and Hubbard [33] have also shown that synesthetic subjects are significantly better at identifying an embedded shape than Non-synesthetes, which is comparable to the way control subjects see colored patterns in the Ishihara color-vision test. Moreover, attention plays a vital role in modulating synesthesia, as during attention-demanding tasks colour often weakens synesthetic experience [35].
Visual attention is a one of the primary steps in the construction of executive function. Reaction or response time (RT) can be measured to examine visual attention [36, 37]. RT quantifies the temporal gap between the start of a stimulus and a subject's response. RT can be used as a measure of awareness in the activities of daily life. Gender, age, level of fatigue, and health influence RT [38]. There are three types of RT: simple, recognition, and choice, with one response, no response, and multiple responses, respectively [39, 40]. Choice RT is more important than the other RTs in everyday life. Visual or auditory inputs are important in the measurement of visual or auditory choice RTs. Due to different sensitivities to the wavelength of red, green, and blue light, RTs are faster or slower [41]. Visual choice RT is faster in males than females and in hand dominant subjects [42]. Faster RT emphasizes a faster processing time by the nervous system and faster muscular movement [43], which is defined as a higher cognitive function [44]. Interestingly, a recent study investigated the `mental number line' which is oriented from left to right on the SNARC (spatial-numerical association of response codes) effect (faster response to the small number on the left side and to the large numbers on the right side) using kinematics of fingers movement besides response time and found that numerical processing affects action execution [45]. Therefore, RT was used as a marker of cognitive processing speed in this study, together with ERPs. There are several components of ERP waveforms depending on the particular stimulus. ERP waveforms can be evoked during visual stimuli. The N100 or N1 ERP component is the negative deflection evoked approximately 100 ms after a stimulus [46, 47]. N100 is the sensory component that indicates selective attention and voluntary discrimination processes [48, 49] a priming stimulus is matched [50]. Alternatively, P300 or P3 is the positive ERP component evoked approximately 300 ms after a stimulus [51, 52]. P300 is a cognitive component, where a higher amplitude indicates higher cognitive involvement, which in turn is taken to mean greater attention [53].
Color and texture are processed in medial temporal-occipital areas and geometric features are processed in lateral temporal-occipital areas [54]. Though some studies have employed simple color stimuli [54, 55], there is a near complete absence of visual cognitive function assessment using pseudoisochromatic plates. For this reason, visual numerical cognitive function was investigated here using the Ishihara test and the ERP indexing procedure, and RT and EC effective connectivity analysis. Additionally, brain topography was defined to understand hemispheric lateralization associated with the two stimuli.
Ethical permission was obtained from the human ethical committee of the Universiti Sains Malaysia (USM) [USM/KK/PPP/JePeM 232.3(8)]. Subjects were recruited via the internet, email notification, and personal communication. All participants gave written informed consent before starting the experiment. A total of 25 (13 male and 12 female) healthy subjects were recruited who had normal or corrected-to-normal vision and no neurological or other major diseases. Sample size was calculated using power and sample size (PS) software (PS version 3.1.2) supported by data from a related study [56]. Inputs for sample size calculation were $ \alpha = $ 0.05, power = 0.9, $ \delta $ = 3.9 and $ \sigma $ = 5.8. The study was performed in the MEG/ERP laboratory of the Hospital Universiti Sains Malaysia (HUSM).
Each subject sat in a sound-proofed and dimly lit room with a 128-channel geodesic sensor net (GSN) from Electrical Geodesic Inc. (EGI) (Eugene, OR) on their head. Before application, the net head circumference of a subject was measured for calculation of the proper net size, and the net was soaked with an electrolyte solution. After each use, the net was rinsed and disinfected for the next subject. A simple visual oddball paradigm was employed by using 18 Ishihara color vision tests plates, including nine number (transformation and vanishing type) and nine non-number plates. The 18 plates were used three times, and a total of 54 stimuli (27 numbers and 27 non-numbers) were used in one session. E-prime software (v 2.0) was used for stimulus presentation.
Subjects sat 80 cm from a 22" LCD computer monitor where all stimuli were presented at the centre of the screen for the subject's response. Subjects were asked to press either button '1’ or button '2' as quickly and correctly as possible if the stimulus was either a number plate or a non-number plate, respectively. Stimuli were presented for one second with a 0.5 sec fixation and 1.5 sec inter-stimulus interval (Fig.1).
Graphical representation of experimental procedure with number and non-number plates of the Ishihara color vision test.
RT recorded for the button press event, and a t-test was employed to analyze correct and incorrect responses.
EEG data was recorded with a Net Amps 300 amplifier and Net station software. Sampling rate was 250 Hz, and electrode impedances were below 50 K$\Omega$. Data were filtered with 30 Hz low-pass and 0.3 Hz high-pass filters, segmented 100 ms before and 800 ms after stimulus presentation, and baseline corrected before 100 ms of stimuli. Eye movement, eye blink, and movement artifacts were removed with an artifact removal tool. The amplitude and latency of N100 and P300 ERP components were then computed from 19 electrodes selected from the 128 electrode employed for the 10-20 international electrode placement system. All procedures were conducted with Net station software. After ERP data extraction, data were pre-processed for outliers, with no correction for N100 and P300 latency outliers. One subject exhibited outliers for N100 and P300 amplitudes of number and non-number stimuli and one subject showed outliers in N100 and P300 amplitude for number plate stimuli. Outliers were identified and replaced by an estimation method for series mean using SPSS version 22 statistical software. To determine the significance of amplitude and latency values of N100 and P300 ERP components for the two categories of number and non-number from the 19 EEG electrodes, both distribution-free nonparametric Wilcoxon signed rank tests and parametric paired T-tests were used.
For brain topography and cortical source localization from all 128 EEG electrodes, standardized low-resolution brain electromagnetic tomography (sLORETA) was performed with Brainstorm [57], which is documented and freely available for download online under the GNU general public license (http://neuroimage.usc.edu/brainstorm).
Analysis and visualization of effective connectivity (Granger causality) over all 128 EEG electrodes employed the MATLAB (MathWorks, Inc)-based connectivity analysis software HERMES [58]
(http://hermes.ctb.upm.es).
RT was significantly longer for non-number plate stimuli (570.33 $ \pm $ 158.43 ms) than number plate stimuli (557.03 $ \pm $ 151.91 ms) (p = 0.045) (Fig. 2).
Mean RT during presentation of number (white dotted column with standard error bars) and non-number (black column with standard error bars) plates of an Ishihara color vision test. RT was significantly increased for non-number plates indicating subjects responded to number plates more rapidly than non-number plates.
The N100 ERP component was not significantly increased in response to number plate stimuli at 12 electrode sites (Fp1, Fp2, F4, F8, C4, P3, P4, O1, O2, Fz, Cz, and Pz) compared with the non-number stimuli. A significantly higher amplitude was found at the Fp2 electrode position, p = 0.028 (t = 2.33, df = 24) with a paired t-test but there was no significance with the Wilcoxon test (Table 1).
Number plate | Non-number plate | Wilcoxon | Paired T-test | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Sites | Mean | ±SD | Variance | Mean | ±SD | Variance | Z-value | p-value | t | df | p-value | |||
Fp1 | 1.91 | 2.22 | 4.91 | 1.43 | 1.92 | 3.7 | -1.197b | 0.231 | 1.36 | 24 | 0.186 | |||
F3 | 1.47 | 2.43 | 5.92 | 1.66 | 1.66 | 2.76 | -0.632c | 0.527 | -0.54 | 24 | 0.598 | |||
F7 | 1.18 | 1.49 | 2.23 | 1.32 | 1.31 | 1.72 | -1.117c | 0.264 | -0.45 | 24 | 0.657 | |||
Fp2 | 2.22 | 2.3 | 5.28 | 1.37 | 2.05 | 4.19 | -1.897b | 0.058 | 2.33 | 24 | 0.028* | |||
F4 | 1.94 | 1.81 | 3.29 | 1.51 | 1.72 | 2.97 | -1.009b | 0.313 | 1.3 | 24 | 0.206 | |||
F8 | 1.64 | 1.14 | 1.29 | 1.14 | 1.25 | 1.57 | -1.628b | 0.104 | 1.82 | 24 | 0.082 | |||
C3 | 1.58 | 1.41 | 1.98 | 1.73 | 1.61 | 2.6 | -0.256c | 0.798 | -0.61 | 24 | 0.55 | |||
C4 | 1.79 | 1.26 | 1.58 | 1.6 | 1.23 | 1.53 | -0.444b | 0.657 | 0.75 | 24 | 0.464 | |||
T3 | 0.99 | 1.11 | 1.23 | 1.57 | 1.16 | 1.34 | -1.870c | 0.061 | -1.95 | 24 | 0.063 | |||
T4 | 1.72 | 1.21 | 1.47 | 2.02 | 1.82 | 3.33 | -0.309c | 0.757 | -0.81 | 24 | 0.429 | |||
P3 | 1.77 | 1.3 | 1.69 | 1.6 | 1.16 | 1.34 | -0.659b | 0.51 | 0.73 | 24 | 0.473 | |||
T5 | 2.08 | 2.19 | 4.81 | 2.15 | 1.47 | 2.18 | -0.498c | 0.619 | -0.19 | 24 | 0.853 | |||
P4 | 2.61 | 1.07 | 1.14 | 2.3 | 1.36 | 1.84 | -1.251b | 0.211 | 1.05 | 24 | 0.306 | |||
T6 | 3.09 | 1.65 | 2.71 | 3.17 | 2.45 | 5.99 | -0.309b | 0.757 | -0.18 | 24 | 0.855 | |||
O1 | 2.42 | 2.77 | 7.66 | 2.18 | 2.08 | 4.34 | -1.170b | 0.242 | 0.57 | 24 | 0.577 | |||
O2 | 2.88 | 2.68 | 7.18 | 2.62 | 2.34 | 5.5 | -0.605b | 0.545 | 0.72 | 24 | 0.479 | |||
Fz | 1.68 | 1.86 | 3.46 | 1.63 | 1.84 | 3.39 | -0.040c | 0.968 | 0.16 | 24 | 0.871 | |||
Cz | 1.9 | 1.89 | 3.58 | 1.89 | 1.59 | 2.52 | -0.256c | 0.798 | 0.04 | 24 | 0.971 | |||
Pz | 2.02 | 1.48 | 2.2 | 1.82 | 1.48 | 2.19 | -0.955b | 0.339 | 0.55 | 24 | 0.586 |
b, c based on positive & negative ranks respectively, * significant if $ p < $ 0.050.
In the case of N100 latencies, a trend for shorter latencies wasobserved at 10 electrode sites (Fp1, F7, Fp2, F4, F8, T3, T4, T6, O2, and Fz) during number plate stimuli when compared to non-number plate stimuli for which only the Fp2 site showed significance for the Wilcoxon test (p = 0.032) (Table 2).
Number plate | Non-number plate | Wilcoxon | Paired T-test | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Sites | Mean | ± SD | Variance | Mean | ± SD | Variance | Z-value | p-value | t | df | p-value | |||
Fp1 | 109.44 | 19.93 | 397.17 | 111.84 | 17.79 | 316.64 | -0.419b | 0.675 | -0.52 | 24 | 0.610 | |||
F3 | 105.44 | 17.88 | 319.84 | 104.8 | 19.83 | 393.33 | -0.284c | 0.776 | 0.13 | 24 | 0.900 | |||
F7 | 112.48 | 25.75 | 663.09 | 120.48 | 25.07 | 628.43 | -1.22b | 0.222 | -1.38 | 24 | 0.182 | |||
Fp2 | 107.36 | 22.59 | 510.24 | 117.28 | 22.62 | 511.63 | -2.14b | 0.032* | -1.8 | 24 | 0.085 | |||
F4 | 102.88 | 16.15 | 260.69 | 105.92 | 20.3 | 412.16 | -0.263b | 0.792 | -0.81 | 24 | 0.423 | |||
F8 | 107.84 | 20.86 | 435.31 | 116.8 | 23.41 | 548 | -1.88b | 0.06 | -1.58 | 24 | 0.128 | |||
C3 | 102.88 | 17.3 | 299.36 | 98.4 | 17.93 | 321.33 | -0.681c | 0.496 | 1.05 | 24 | 0.306 | |||
C4 | 103.04 | 19.71 | 388.37 | 99.68 | 20.33 | 413.23 | -0.163c | 0.871 | 0.68 | 24 | 0.506 | |||
T3 | 109.12 | 25.76 | 663.36 | 120 | 27.54 | 758.67 | -1.77b | 0.076 | -1.85 | 24 | 0.076 | |||
T4 | 110.08 | 31.18 | 972.16 | 115.84 | 28.55 | 815.31 | -0.987b | 0.324 | -0.82 | 24 | 0.422 | |||
P3 | 103.68 | 26.86 | 721.23 | 100.32 | 25.84 | 667.89 | -0.618c | 0.537 | 0.47 | 24 | 0.641 | |||
T5 | 116.64 | 30.65 | 939.57 | 116.48 | 32.28 | 1041.76 | -0.443b | 0.658 | 0.03 | 24 | 0.980 | |||
P4 | 112.32 | 30.29 | 917.23 | 106.56 | 28.07 | 787.84 | -1.10c | 0.27 | 1.01 | 24 | 0.321 | |||
T6 | 110.4 | 31.03 | 962.67 | 118.72 | 32.84 | 1078.29 | -1.08b | 0.28 | -1.14 | 24 | 0.267 | |||
O1 | 121.44 | 27.73 | 769.17 | 120.8 | 28.59 | 817.33 | -0.130b | 0.897 | 0.14 | 24 | 0.891 | |||
O2 | 122.08 | 27.43 | 752.16 | 125.76 | 26.96 | 726.77 | -0.919b | 0.358 | -0.73 | 24 | 0.473 | |||
Fz | 105.76 | 18.48 | 341.44 | 112.16 | 19.3 | 372.64 | -1.76b | 0.078 | -1.41 | 24 | 0.172 | |||
Cz | 105.28 | 19.76 | 390.29 | 103.52 | 17.75 | 315.09 | -0.192c | 0.848 | 0.66 | 24 | 0.515 | |||
Pz | 115.36 | 25.89 | 670.24 | 116.32 | 27.3 | 745.23 | -0.114b | 0.909 | -0.17 | 24 | 0.870 |
b, c based on positive & negative ranks respectively, * significant if $ p < $ 0.050.
The majority of electrode positions (12 of 19 sites: F4, C3, C4, P3, P4, T3, T4, T5, T6, O1, O2, and Cz) showed higher amplitudes of the P300 ERP component during non-number plate stimuli, for which T4 and T6 were highly significant for both non-parametric and parametric tests, when compared to the number plate stimuli, a tendency for higher amplitude was observed in the left frontal hemisphere (Fp1, Fp2, F3 and F7) for number stimuli and for which Fp1 was significant (t = 2.35, df = 24) for only the parametric test (Table 3). At about the P300 ERP latency, non-number plates evoked longer latencies at 15 sites (Fp1, F3, F7, F8, C3, T3, P3, P4, T5, T6, O1, O2, Fz, CZ, and Pz). A significantly longer latency was found at the Pz location for both Wilcoxon (p = 0.036) and paired t-test (t = $-$2.07, df = 24), while the remaining electrode locations showed a tendency towards longer latencies (Table 4).
Number plate | Non-number plate | Wilcoxon | Paired T-test | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Sites | Mean | ± SD | Variance | Mean | ± SD | Variance | Z-value | p-value | t | df | p-value | |||
Fp1 | 4.28 | 3.5 | 12.27 | 2.84 | 3.62 | 13.1 | -0.256b | 0.798 | 2.35 | 24 | 0.027* | |||
F3 | 3.08 | 3.18 | 10.09 | 2.77 | 2.35 | 5.51 | -1.709b | 0.088 | 0.49 | 24 | 0.632 | |||
F7 | 1.48 | 2.73 | 7.47 | 1.45 | 2.43 | 5.92 | -0.525c | 0.6 | 0.04 | 24 | 0.966 | |||
Fp2 | 4.91 | 3.8 | 14.48 | 4.21 | 4.01 | 16.09 | -1.144b | 0.253 | 1.38 | 24 | 0.179 | |||
F4 | 3.19 | 1.79 | 3.19 | 3.23 | 1.65 | 2.73 | -0.013b | 0.989 | -0.11 | 24 | 0.913 | |||
F8 | 1.86 | 2.16 | 4.68 | 1.81 | 1.48 | 2.18 | -0.094c | 0.925 | 0.12 | 24 | 0.907 | |||
C3 | 2.97 | 2.2 | 4.84 | 3.28 | 2.55 | 6.51 | -0.309c | 0.757 | -0.62 | 24 | 0.542 | |||
C4 | 3.85 | 1.89 | 3.59 | 4.2 | 1.95 | 3.78 | -0.982c | 0.326 | -0.99 | 24 | 0.331 | |||
T3 | 1.99 | 2.33 | 5.43 | 2.43 | 2.59 | 6.69 | -1.520c | 0.128 | -0.79 | 24 | 0.44 | |||
T4 | 2.25 | 3.02 | 9.15 | 3.76 | 3.94 | 15.55 | -2.919c | 0.004* | -3.26 | 24 | 0.003* | |||
P3 | 2.82 | 2.4 | 5.77 | 2.88 | 1.81 | 3.28 | -0.605c | 0.545 | -0.15 | 24 | 0.879 | |||
T5 | 2.12 | 1.96 | 3.84 | 2.79 | 2.24 | 5.04 | -1.359c | 0.174 | -1.38 | 24 | 0.179 | |||
P4 | 2.83 | 1.88 | 3.55 | 3.31 | 2.08 | 4.34 | -1.574c | 0.115 | -1.68 | 24 | 0.105 | |||
T6 | 1.62 | 2.65 | 7.03 | 2.98 | 3.02 | 9.11 | -2.839c | 0.005* | -2.66 | 24 | 0.014* | |||
O1 | 1.7 | 2.76 | 7.64 | 1.99 | 2.65 | 7.05 | -1.063c | 0.288 | -0.63 | 24 | 0.533 | |||
O2 | 1.24 | 2.5 | 6.24 | 1.61 | 3.01 | 9.04 | -0.632c | 0.527 | -0.85 | 24 | 0.404 | |||
Fz | 3.35 | 2.51 | 6.3 | 3.06 | 2.32 | 5.38 | -0.767b | 0.443 | 0.84 | 24 | 0.409 | |||
Cz | 3.99 | 2.93 | 8.6 | 4.41 | 2.29 | 5.23 | -0.740c | 0.459 | -0.71 | 24 | 0.485 | |||
Pz | 3.3 | 2.92 | 8.51 | 2.89 | 2.68 | 7.17 | -0.955b | 0.339 | 0.63 | 24 | 0.535 |
b, c based on positive & negative ranks respectively, * significant if $ p < $ 0.050, * bold & italic significant at both Wilcoxon & paired t-test.
Number plate | Non-number plate | Wilcoxon | Paired T-test | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Sites | Mean | ± SD | Variance | Mean | ± SD | Variance | Z-value | p-value | t | df | p-value | |||
Fp1 | 508.8 | 160.43 | 25737.33 | 511.2 | 162.47 | 26397.33 | -0.568b | 0.57 | -0.08 | 24 | 0.938 | |||
F3 | 488.64 | 113.02 | 12772.91 | 522.24 | 112.24 | 12597.44 | -1.339b | 0.181 | -1.60 | 24 | 0.124 | |||
F7 | 416.16 | 131.12 | 17191.31 | 433.12 | 118.8 | 14114.03 | -0.639b | 0.523 | -0.58 | 24 | 0.569 | |||
Fp2 | 516.16 | 160.41 | 25729.97 | 488 | 155.05 | 24041.33 | -0.834c | 0.404 | 1.01 | 24 | 0.325 | |||
F4 | 518.08 | 121.31 | 14716.16 | 511.04 | 116.47 | 13564.37 | -0.557c | 0.577 | 0.23 | 24 | 0.820 | |||
F8 | 504.16 | 152.21 | 23168.64 | 512.64 | 146.27 | 21394.24 | -0.341b | 0.733 | -0.27 | 24 | 0.789 | |||
C3 | 501.12 | 132.44 | 17540.69 | 528.16 | 95.16 | 9055.307 | -0.672b | 0.502 | -0.87 | 24 | 0.391 | |||
C4 | 549.12 | 117.73 | 13859.36 | 544.16 | 93.18 | 8681.973 | -0.757c | 0.449 | 0.22 | 24 | 0.829 | |||
T3 | 475.52 | 132.17 | 17468.43 | 483.84 | 147.53 | 21764.64 | -0.390b | 0.697 | -0.40 | 24 | 0.691 | |||
T4 | 508.16 | 138.56 | 19197.97 | 497.6 | 125.14 | 15660 | -0.304c | 0.761 | 0.31 | 24 | 0.761 | |||
P3 | 468.64 | 127.63 | 16290.24 | 494.72 | 95.17 | 9056.96 | -0.972b | 0.331 | -0.93 | 24 | 0.363 | |||
T5 | 419.52 | 101.2 | 10241.76 | 438.4 | 119.78 | 14346.67 | -0.900b | 0.368 | -0.90 | 24 | 0.375 | |||
P4 | 456.64 | 116.85 | 13652.91 | 509.28 | 124.93 | 15606.29 | -1.272b | 0.203 | -2.00 | 24 | 0.057 | |||
T6 | 404.16 | 97.98 | 9599.307 | 459.36 | 125.62 | 15780.91 | -1.758b | 0.079 | -2.04 | 24 | 0.053 | |||
O1 | 401.76 | 119.86 | 14365.44 | 418.24 | 121.41 | 14740.11 | -1.188b | 0.235 | -0.74 | 24 | 0.467 | |||
O2 | 407.52 | 150.21 | 22564.43 | 423.04 | 134.2 | 18009.71 | -1.413b | 0.158 | -0.56 | 24 | 0.584 | |||
Fz | 499.84 | 136.75 | 18700.64 | 503.36 | 130.76 | 17098.24 | -0.061b | 0.951 | -0.1 | 24 | 0.921 | |||
Cz | 531.84 | 113.29 | 12835.31 | 548.64 | 89.61 | 8030.24 | -0.486b | 0.627 | -0.68 | 24 | 0.506 | |||
Pz | 461.12 | 144.99 | 21023.36 | 521.12 | 122.52 | 15011.36 | -2.100b | 0.036* | -2.07 | 24 | 0.049* |
$ b $, $ c $ based on positive & negative ranks respectively, * significant if $ p < $ 0.050, * bold & italic significant for both Wilcoxon & paired t-test.
Grand averaged waveforms from twenty five subjects are shown in Fig. 3 with their N100 and P300 ERP components from 19 electrode channels: Fp1, Fp2, F3, F4, F7, F8, C3, C4, T3, T4, T5, T6, P3, P4, O1, O2, Fz, Cz, and Pz.
A scalp topography map for the N100 ERP components showed similar patterns for both number and non-number plate conditions except for a more intense map at the occipital area for number plates, which was also reflected in the non-significant amplitude increase, as seen in Fig. 4 and Fig.5. However, interestingly, a scalp topography map for the P300 components showed different relationships, such as in the case of the number plates. That was more intense in the left hemisphere, which was opposite the non-number plates which were more intense in the right hemisphere.
Cortical activation patterns during number (Fig. 4) and non-number plate tasks (Fig. 5) in relation to N100 and P300 ERP components are shown in Table 5. Interestingly, in terms of Brodmann's cortical area, the activation pattern of P300 ERP components for the number plate was limited to two Brodmann's regions for each hemisphere, whereas for non-number plates, there were three for each hemisphere; moreover, inter- and intra-hemispheric differences were noticed in the responses to these two stimuli.
ERP components | MNI x, y & z coordinates (millimetres) | Brodmann's area (BA), L-left, R-right | Cortical regions | EEG electrode positions |
---|---|---|---|---|
Number | ||||
N100 | -5768 | LBA39 | Inferior Parietal Lobule - Intraparietal sulcus - Angular gyrus | P3, T5 |
-56 -28 14 | LBA41 | Temporal lobe - Auditory cortex | T3 | |
17 70 6 | RBA10 | Prefrontal Cortex | Fp2 | |
P300 | -44 -29 14 | LBA40 | Inferior Parietal Lobule - Supramarginal gyrus | C3, P3 |
-47 17 0 | LBA45 | Inferior Frontal gyrus - Pars Triangularis (Broca's Area) | F7, F3 | |
50 -49 24 | RBA20 | Inferior temporal, Fusiform and Parahippocampal gyri | T4, F8 | |
17 69 2 | RBA10 | Prefrontal cortex | Fp2 | |
Non-number | ||||
N100 | -28 -94 7 | LBA18 | Occipital lobe - Secondary visual cortex | O1 |
-19 -61 5 | LBA23 | Posterior Singulate Gyrus | Pz | |
P300 | -55 -56 14 | LBA39 | Inferior parietal lobule - intraparietal sulcus - Angular gyrus | P3, T5 |
-3070 | LBA22 | Superior Temporal Gyrus (Wernicke's area) | T3 | |
-7624 | LBA17 | Occipital lobe - Primary visual Cortex | O1 | |
46 -78 14 | RBA19 | Occipital Lobe (Secondary visual cortex) | O2, T6 | |
58 -49 14 | RBA39 | Inferior parietal lobule - intraparietal sulcus - Angular gyrus | P4, T6 | |
45 -25 14 | RBA40 | Inferior parietal lobule (Supramarginal gyrus) | C4, P4 |
Effective connectivity (Granger causality measure), showed a causal effect, interestingly, in the area of the left inferior frontal lobe (near F7) and temporal lobe (near T3), near the mid-parietal (Pz) and occipital lobe (O2) during number plate tasks (Fig. 6), but the causal effect was almost solely localized in the occipital area near O1 during non-number plate tasks (Fig. 7).
Arrangement of the grand average waveform of ERP components at 19 scalp sites of electrode channels during the presentation session of number (grey line) and non-number plates (black line), small vertical solid and dotted bars indicate segment time = 0 & stimulation presentation time, respectively. Central rectangle gives scale.
EEG waveform, scalp topography, and cortical activations during number plate task. Upper left panel of figure (A), EEG time series of all 128 channels, y-axis and x-axis give amplitude ($ \mu $V) & time (ms), respectively, white and red dotted lines indicate stimulation start time marked at 0 ms peak N100 ERP response (104 ms), respectively. Upper right panel gives scalp topographic map plotted from N100 peak. Lower panel of figure (A), cortical activations of N100 peak response are displayed as MRI 3D view with color bar (left side) and as axial MRI view with color bar (right side). Lower left panel of figure (B), EEG time series of 128 channels, y-axis and x-axis gives amplitude ($ \mu $V) & time (ms), respectively. White dotted line gives stimulation start time (0 ms) and red line denotes peak P300 ERP response (492 ms), upper right panel shows scalp topographic map plotted from P300 peak. Lower panel of figure (A), cortical activations of P300 peak response are displayed on MRI 3D view with color bar (left side) and with axial MRI viewer, color bar (right side).
EEG waveform, scalp topography, and cortical activations during non-number plate task. Upper left panel (A), EEG time series of all 128 channels, y-axis and x-axis give amplitude ($ \mu $V) & time (ms) respectively, white dotted line gives stimulation start time (0 ms) and red line gives peak N100 ERP response (100 ms), upper right panel shows scalp topographic map plot at N100 peak. Lower panel (A), cortical activations of N100 peak response are displayed as MRI 3D view with color bar (left side) and as axial MRI view with color bar (right side). Lower left panel (B), EEG time series of 128 channels, y-axis and x-axis give amplitude ($ \mu $V) & time (ms), respectively, white dotted line gives stimulation start time (0 ms) and red line indicates peak P300 ERP response (540 ms), upper right panel gives scalp topographic map plotted at P300 peak. Lower panel (A), cortical activations of P300 peak response are given as MRI 3D view with color bar (left side) and as an axial MRI view with color bar (right side).
Effective connectivity visualization from grand averaged wave form during presentation of number plates of an Ishihara color vision test from all recorded 128 channels of EGI system, circle indicates 19 electrode sites in 10-20 international system.
Effective connectivity visualization from grand averaged waveform during presentation of non-number plates of an Ishihara color vision test from all recorded 128 channels of EGI system, circle indicates 19 electrode sites in 10-20 international system.
Visual cognitive function has been investigated using Ishihara color vision test plates, where number and non-number plates were the stimuli. RT, amplitudes, and latencies of N100 and P300 ERP components, brain topography, and Granger causation (effective connectivity) were analyzed. There were no significant differences between the N100 ERP components obtained for either stimulus, but the non-number plate stimuli evoked significantly higher amplitudes and longer latencies for the P300 ERP component compared to number plate stimuli. Similarly, the reaction time was slower for the non-number plate task. Interestingly, a scalp topography map for P300 components showed different patterns of asymmetry, with a higher intensity in the left hemisphere for the number plate task and higher intensity in the right hemisphere for the non-number plate task. Similarly, analysis of effective connectivity revealed a clear and strong interaction between occipital and left frontal cortices during number plate tasks. During number recognition, the direction of coupling was from the visual occipital area towards the language area of semantic coding, but the effect was restricted to the occipital area during the non-number plate task. Non-number plates evoked longer RTs, higher amplitudes, and longer latencies for the P300 ERP component at most electrode sites, of which two were significant when compared with number plate stimuli. Moreover, left/right asymmetry in the areas of cortical activation, interaction, and direction of the effect of attention control during number and non-number tasks were clearly demonstrated in this study.
RTs quantify the time interval between stimulus occurrence and the voluntary response of a subject. Ghuntla et al. [59] proposed reaction time is a substantial physiological parameter that contributes information about how quickly a subject can respond to the perception of a stimulus. Faster RTs indicate shorter nervous system processing times associated with faster muscular movement [43] and are an indicator of higher cognitive function [44]. Correspondingly, with this study, reaction time was different during the visual presentation tasks of number and non-number plates. Reaction time was significantly slower for non-number plates than for the number plates. Due to the similarly colored dot shape design of number and non-number plates, numerical error [60] and confusion [61] are obvious with Ishihara testing, even for trichromats; such difficulties added to visual search and identification of a non-number plate, because after encoding a stimulus, subjects need to match the target with the memorized shape of the correct Arabic digit to achieve accurate responses for non-number pseudoisochromatic plates a decision is made. A significant increase in the mean RT indicated that subjects were taking more time to elicit a convenient intentional response for the relatively complex non-number plates compared with number plates. Luck [62] noted that reaction time is usually affected by the complexity of the stimulus.
The key ERP components for this study were the N100 and P300 ERP. Different colors, shapes or any visual stimuli evokes a N100 ERP component [63] when the subjects are previously experienced with this stimulus [50]. N100 is a marker for visual perception and P300 is an indicator of the higher cognitive functions of attention processing [64]. No significant differences in N100 between number and non-number plates demonstrated that selective attention and a sensory gating mechanism of attention are almost similar for these two stimuli. Regarding the neural sources of N100, based on the result of a scalp topographical map N100 was rather similarly bilaterally distributed for both stimuli, but with a higher intensity for the number plates at occipital (O1 and O2), parietal (P3, P4, and Pz), and frontal (Fp1, Fp2, F4, and F8) locations. Though the differences in this finding were not significant, they were consistent with previous studies [65,66,67].
The P300 ERP component can be evoked during visual stimulation [68, 69]. Significant differences in P300 between number and non-number plates suggested that during the non-number plate task, subjects allocated more attention and memory processing than for the number plate task [70]. Elevated P300 amplitudes were found for increased attention [71], and they indicate increased consciousness [72] with the quality of selection [73]. Reduced amplitudes and longer latencies and processing times were found during periods of lower attention [74, 75]. In this study, non-number plate stimuli evoked higher amplitudes and longer latencies of the P300 ERP component at most electrode sites This result suggests that subjects paid more voluntary attention to quality selection, which took more information processing time for non-number plate stimuli compared with number stimuli.
Concerning the neural sources of P300, based on the result of the scalp topographic map, interestingly, the distribution of magnitudes for the non-number plate was greater in the right hemisphere, whereas, conversely for the number plate, it was greater in the left hemisphere, indicating a left/right asymmetry of a neural response pattern for these two stimuli in the brain. Additionally, based on the results of cortical activation patterns, remarkably, some brain areas are activated both during number and non-number plate tasks, such as the bilateral inferior parietal lobule and temporal lobe. Strikingly, the frontal lobe activation especially in the left inferior frontal gyrus (Broca's area) was found only during the number plate task. Number recognition was not only a process of visual perception and attention, but was also related to the higher level of the cognitive function of language. This propensity was further supported by the effective connectivity analysis undertaken for this study, as it was in a previous study of blind people [76]. Neural correlates and functional measurement of number representations are complex occurrences when their semantic category is assimilated with shape and color categories. Number recognition, which can be associated with numerosity and the identification of the inherent magnitude of a number, can be a visual number shape (e.g. Arabic digit) that is culturally learned through education after birth. Activation of bilateral occipitotemporal cortex associated with the shape coding of non-numbers as opposed to numbers, was found in this study. Conversely, bilateral occipitotemporal activation based on the shape coding of Arabic digits and other numeric symbols was found in a study of macaque monkey brain that investigated neurophysiological evidence for a neural code for numbers [77]. However, in the current study, the right inferior temporal cortex activation during number plate tasks was consistent with findings reported by others [78, 79]. Additionally, during number plate tasks, the language area (LBA45) was activated, which is a similar result to that reported for a study of Arabic number reading [80]. Nonetheless, language area activation did not occur during non-number tasks in this study. Taken together, however, visual recognition of numbers dissociates from the recognition of non-numbers for both behavioral data and at the neural level, as similarly found in a study of neural double dissociation between letter and number recognition [81].
Granger causality analysis has received much attention in the study of language processing and the interactions of brain areas in large neural networks [82]. Throughout connectivity analysis, results reported here for effective connectivity revealed that there was clearly a strong interaction between occipital (O1 \& O2) and left lateral frontal cortices (near the F7 electrode) during number plate tasks. The direction of coupling was from a visual occipital area towards a language area associated with semantic coding during number recognition, thus showing a stimulus-driven bottom-up effect in the defined neural networks. However, this effect was restricted to the occipital area during the non-number task.
In summary, to assess attention during numerical information processing, the visual cognitive function of ERP waveforms was investigated by use of Ishihara pseudoisochromatic plates. Results suggest that the visual identification of the two different stimuli were almost identical. However, the number information was processed spontaneously and was faster than the processing of non-number information. Attention load was relatively higher and delayed in the decision-making process of the non-number task. More interestingly, distribution of the source of attention control and the effects of attention on information processing were different for number and non-number tasks. Finally, this study is the first to report the use of easily administrable and readily available pseudoisochromatic plates to evaluate visual numerical cognition. These plates could provide an important tool to test attention deficit and connectivity disorders in anatomically and physiologically distinct brain networks.
This work was supported by a short term grant of Universiti Sains Malaysia (USM) (304/PPSP/ 61311092) for the author T.B.
All authors declare no conflicts of interest.