Individual differences in mental imagery tasks : a study of visual thinkers and verbal thinkers

RIEB, Kobe University, Kobe, 657-8501, Japan; Santa Fe Institute, Santa Fe, New Mexico, 87501, USA Institute of Economic Research, Kyoto University, Kyoto, 606-8501, Japan Medical Welfare Center, St. Joseph Hospital, Kyoto, 603-8323, Japan Elegaphy, Inc., Otsu, 520-0225, Japan Human Technology Research Institute, National Institute of Advanced Industrial Science and Technology, Tsukuba, 305-8566, Japan


Introduction
In the field of economics, neuroeconomics has received a great deal of interest.This field examines economic behavior and its association with accompanying brain activity by integrating research in experimental economics and behavioral economics, fusing them with that of neuroscience.For example, one study of neuroeconomics found that the prefrontal cortex is active when choices are made about the distant future, while the limbic system of the cerebrum is active when choices are made about the near future [2] .In other words, different regions of the brain are used when making choices pertaining to the near versus distant future.These studies seem to take the approach of obtaining the overall perspective concerning the average human being.
However, conventional economics places a strong emphasis on conducting studies that address resource distribution in a society comprising heterogeneous agents [3] .Individuals can be divided into two completely different groups of risk lovers and risk averters based on attitudes toward danger [4] .In game theory, some have studied

RESEARCH HIGHLIGHT
interdependent relationships such as conflicts of interest or collaboration between different individuals, as well as decisions on strategy [5] .
In general, comprehension of another individual's thought process is difficult.The psychological state that would allow one to stand in another's shoes and understand him or her is called the "theory of mind" in psychology [6] .The theory of mind has been studied extensively in young children in their formative stages, and in individuals with developmental disorders such as autism spectrum disorder [7] .However, given the diversity of individual personalities and demeanors, even healthy individuals in the general population require high level communication skills to grasp another's personality or sensitivities and interpret their emotional dynamics.For example, we must collect information related to another's personality and actions, or predict long-term their personality and sensitivities as the relationship develops with time.
While the importance of distinguishing between individuals based on differences in thinking patterns (regardless of whether they are congenital or acquired) is widely acknowledged, only a few studies have addressed economic behaviors among different individuals.In particular, studies that examine how differences in individual thinking characteristics affect decision-making are incredibly scarce.Against this backdrop, our research has focused on understanding why different individuals end up making different choices and how brain function differs in these situations.We anticipate that research on thinking characteristics and decision-making will form the foundation for understanding judgments and decisions made by consumers and investors regarding economic challenges.
Thinking comprises verbal thinking (using words to think) and visual thinking (using images, rather than words, to think), and some have studied differences between the two using brain measurements [8] .In the same way that some people can remember the faces of those they have just met, while others cannot, individuals also vary in their capacity to evoke images of certain things.It is known that individuals who can remember a large amount of information in a short period do so using visual images.When playing chess, a professional chess player can evoke images of the patterns of chess pieces and move their chess pieces accordingly without verbalizing these thoughts.In general, these individuals use more dominantly the right hemisphere of their brain over the left hemisphere, as well as the visual area over the verbal area.In addition to professional chess players, many who work in design and imaging occupations share this tendency toward visual dominance.thinking tend to comprehend things in order.They have difficulty starting a novel halfway through, or listening to a mathematical lecture halfway through.These individuals must read from the beginning, or listen from the start, in order to understand the content.When they think, they think while talking to themselves.This differs from visually dominant thinkers, who think while moving images around.

Individuals whose thinking is dominated by verbal
In a recent study, we divided subjects into two groups based on whether they had strong or weak visualization ability to form mental imagery.We then proceeded to examine how this ability at the time of thinking affected global brain activation patterns, as well as differences in local activation patterns that emerged in a region of interest (ROI) [1] .With the prediction that the strength of visualization ability is associated with a tendency toward visual or verbal thinking, we hypothesized that "when undergoing a visual or verbal task, those with strong imaging capacities will have more activation in the visual area, while those with weak imaging capacities will show more activation in the frontal language area, as assessed with high-β and low-γ bands."To test this hypothesis, we administered block design tasks to our subjects and conducted magnetoencephalography (MEG) using SQUID.This experimental task has been used in previous studies [9][10][11] .Subjects were divided into groups according to whether they had strong or weak visualization abilities using a survey developed in a previous study that uses visual and verbal factors to classify individuals in this manner [8] .The survey we used was created independently, but was based on essentially the same principle [9][10][11] .Comparison of brain activity in the two groups revealed that those with strong visualization ability tended to be visual thinkers, while those with weak visualization ability tended to be verbal thinkers.Below, we will discuss the research in light of results from our recent paper, together with some complementary findings [1]   .

Materials and methods
Experiments were conducted on four dates from August 2 through September 13, 2011, at the National Institute of Advanced Industrial Science and Technology in Ikeda City (Osaka, Japan).Individuals subject to analysis totaled 13 (11 males, 2 females), and were divided ahead of time into two groups according to their dominant thinking pattern as determined by the survey.Actual question items used to divide subjects are shown in Table 1(a).For each "A" selected for any of the questions, the respondent received 1 point, and each "B" received 0 points.Those who chose more "A"s than "B"s (i.e., had a total score of 3 or higher) were classified as Group I (strong visualizers).Those with a total score of 2 or lower were classified as Group L (weak visualizers).
Each experimental session comprised 2 back-to-back repeated sequences in which a subject was asked to do the following: envision Kiyomizu-dera (a famous temple in Kyoto), envision the Japanese House of Parliament, recall the 12 signs of the Chinese zodiac, recall a conversation they had with someone that day, and cease thinking at rest (Table 1b).Ten seconds was allotted to each task, with no breaks in between.The reasoning behind each task was as follows: envisioning Kiyomizu-dera and the Japanese House of Parliament required them to imagine a book or photograph (1, 2).Recalling the 12 signs of the Chinese zodiac led them to chant nouns in their head (3).Recalling a personal conversation challenged them to remember an encounter with someone (4).Tasks 5 and 6 were done to have them rest.Our intent was to measure neurological activity while subjects were thinking as they performed tasks 1 and 2 for visual imagery (hereafter, visual conditions) and 3 and 4 for verbal recollection (hereafter, verbal conditions), and then compare these measurements with those taken while subjects were still and rest during tasks 5 and 6 (rest conditions) in order to b1, b2 and b3 monitored frontal language areas.More specifically, Nos.v1 and v2 designate primary visual and early visual areas, respectively.Nos.b1, b2 and b3 designate frontal language areas in the middle frontal gyri (b1) and in the left inferior frontal gyri (b2 and b3).No. b3 corresponds to so-called Broca's area.
identify any increases in measurements relative to those at baseline (tasks 5 and 6).
A whole-cortex-type 122-channel direct current SQUID system (Neuromag 122, Elekta-Neuromag, Helsinki, Finland) was used for MEG measurements.Several groups have published protocols, theorems, and spectrum analyses according to MEG [12][13][14][15][16][17][18][19] .For each channel, the MEG signal was measured, and a short-time Fourier transform was applied to each 1/5 second interval.We thereby derived an estimated spectral density for all 61 sensor locations in 5 Hz widths.For each condition (visual, verbal, and rest), means for the two groups were determined.In order to examine activation patterns under visual conditions and verbal conditions relative to baseline, we calculated the mean ratios of these measurements relative to the mean from the rest conditions for each sensor.

Results
In this section, we will report some of our findings, with new results on additional frequency bands and sensors [1] . Figure 1 shows the distribution of the 122 sensors made into 61 pairs.Figure 2 is a color-coded display of the low-γ band (27.5 Hz-32.5 Hz, central frequency of 30 Hz) spectral density for each sensor under visual conditions (upper panel) and verbal conditions (lower panel) in each group, presented as ratios to that for rest conditions.Group I is the first row, Group L is in the second row, and the ratio of Group I to Group L is in the third row.For each of these figures, smoothing by nonparametric regression was performed according to the sensor layout map shown in Figure 1.
In rows 1 and 2 of Figure 2, Group I and Group L both show activation from the parietal region to the frontal lobe under both visual and verbal conditions, and represent "thinking" characteristics.In addition, as shown in row 3 (far right), the left side near the visual area of Group I shows more activation (reddish tint) relative to that of Group L. This suggests that relative to Group L, those in Group I (visual thinkers as defined by the survey) had more neural activity near the visual area.Observations of the frontal language area, which is located within the middle frontal gyri and left inferior frontal gyri, revealed more activity in Group L (bluish tint).The high-β bands (22.5 Hz-27.5 Hz, central frequency of 25 Hz) showed a similar tendency to that of low-γ bands (central frequency 30 Hz).For more on this, refer to Figure 1 of our recent study [1] .
From among the sensors thought to be near the visual area and frontal language area, we selected some from each area (Figure 1): v1 and v2 from the visual area and b1, b2, and b3 from the frontal language area.The primary visual area and early visual area were indicated by v1 and v2, respectively.Frontal language areas in the middle frontal gyri were indicated by b1, while those in the left inferior frontal gyri were indicated by b2 and b3.In particular, b3 in the left inferior frontal gyri corresponded to Broca's area.
Figure 3 shows the Group I spectrogram (ratio against Group L) for sensors v1, b1, and b3 for visual and verbal conditions.The y-axis of the spectrogram shows the frequency (9 frequency bands, central frequency of 10-50 Hz), while the x-axis shows the measurement time (0-20/3 sec).The y-axis is divided into 5 Hz intervals, and the x-axis into 1/15 sec intervals, presenting a color-coded version of spectrum density (ratio).Information on sensors v2 and b2 can be found in Figure 3a of our recent study [1] .Sensors in the visual area show an overall reddish tint, and are consistent with the results of Figure 2, as they indicate more activity in Group I. Particularly for v1, β bands with central frequencies of 20 and 25 Hz (y axis) show a more highly activated band (reddish) in the direction of the x-axis (time).This corresponds to our previous observation of the presence of low-γ band activity (around 30 Hz) during image processing [20] .On the other hand, compared to sensors in the visual area, those in the frontal language areas show an overall bluish tint, with Group L showing markedly higher activation, particularly under verbal conditions.Sensors b1 and b3 in particular show band-like areas of higher activation (more blue) in the direction of the x-axis (time), specifically for high-β/low-γ bands (y-axis) with central frequencies of 25 and 30 Hz.
Statistical analysis revealed that for both the sensor-based and spatial filter-based approaches, consistent findings obtained are as follows.Group I showed more activation in the visual area compared to Group L, markedly so for the high-β band (25 Hz).Group L showed more activation in the frontal language area than Group I, markedly so for the low-γ band (central frequency, 30 Hz).

Group-dependent differences in global activation patterns
The present study aimed to demonstrate that individuals who use more imaging in their spontaneous thinking show more activity in the visual area when undergoing tasks, relative to those who use less imaging.Research using fMRI has found a significant correlation between an individual's subjective "vividness" of visual imagery and activity in the visual area [26] .In addition, visual imagery activities involve not only the visual area but associated areas from the frontal lobe to the parietal area, involving activation of these areas as well [27] .Our analysis revealed activation, not only in the low-γ bands (central frequency 30 Hz; Figure 2), but for α, β, and γ bands in the region spanning the frontal lobe to parietal area.This was observed in both groups, regardless of visual or verbal conditions (Figure 2, rows 1, 2).On the other hand, both groups showed reduced activation near the visual area (areas centered around sensor pairs v1 and v2) under both conditions [1] .For both visual and verbal conditions, the ratio of Group I to Group L revealed more activation in Group I (Figure 2, row 3).We therefore conclude that our hypothesis was supported, as Group I had higher activity than Group L in the occipital area including the visual area while performing the tasks.This result was consistent with that of our previous study [9] .

Importance of β and γ bands for spontaneous imagery
One study examined α (8-14 Hz) and β (14-24 Hz) band brain activity during visual imagery [28] .During visual imagery resulting from external stimulation, increased activity is evident for θ waves and β waves in the frontal lobe (particularly the medial superior frontal gyrus), as well as α waves and β waves of the parietal area (particularly the superior parietal lobe) [27] .In addition, γ bands play an important role in higher brain function, especially in visual cognition [29,30] .Other research studies using EEG/MEG have demonstrated that high-β bands and low-γ bands are appropriate indicators that can be used to evaluate the degree of individual competency with regard to visual and verbal task processing [20,[31][32][33] .
The present study analyzed α waves (central frequency 10 Hz), β waves (central frequency 15, 20, 25 Hz), and γ waves (central frequency 30, 35, 40, 45, 50 Hz) (see spectrogram in Figure 3a and Supplementary Figures from our recent study [1]   ).This revealed that Group I had higher activation with regard to α waves, but particularly the high-β bands (central frequency, 25 Hz) and low-γ bands (central frequency, 30 Hz) in the occipital area including the visual area.

Visual condition for visual imagery and verbal condition for verbal recollection
While it is certainly necessary to distinguish between perception of external stimuli and spontaneous imagery, the two share similarities (in particular, activity near the visual area) as well as differ with regard to brain activity [28,[34][35][36][37][38][39][40] .
We focused on spontaneously occurring thinking, and divided this into visual thinking and verbal thinking.The visual thinking considered here is that which creates images, while verbal thinking is that comprising self-talk.Some studies have analyzed visual thinking and verbal thinking as they relate to the strength of an individual's ability for imagery [1,8,[41][42][43] .One such study examined the interaction between activity in the frontal cortical area and that in Wernicke's area during verbal thinking [41] .We found, especially under verbal condition, that Group I showed higher activation in the visual area associated with visual cognition, while Group L had higher activation in the frontal language area associated with verbal expression in the middle frontal gyri and left inferior frontal gyri.This indicates that strong visualizers are visual thinkers, while weak visualizers are verbal thinkers.In other words, this study presents evidence for an association between one's ability for mental imagery and brain activity.

Conclusions
Compared to measurement of neural activity evoked by external stimuli of the visual or auditory organs, measuring spontaneous neural activity has been considered difficult.However, the present study used a SQUID device that enables highly sensitive and non-invasive measurements of spontaneous brain activity, which allowed for successful examination of individual differences.Specifically, global activation patterns for spontaneous thinking activities revealed that strong visualizers showed marked differences in brain activity from weak visualizers.When the analysis was limited to a region of interest (ROI), the group with strong visualization showed activation near the visual area, while the group with weak visualization showed activation near the frontal language area.It is easy to imagine how an individual's thinking pattern (visual or verbal) might greatly affect that individual's thinking in everyday life.The present study demonstrates that these individual characteristics would significantly correlate with changes in global brain wave activity as well as local activation patterns.We anticipate that these findings will be applicable to fields related to economic activity and education.

Table 1 .
Questionnaire and Tasks: Five-item questionnaire (a) was used to assess subjects' ability to form mental images.Tasks (b) were used in the experiment

Figure 2 .
Figure 2. Plots of the sensor-specific spectral densities under the visual condition (upper row) and the verbal condition (lower row), for 27.5-32.5Hz (central frequency: 30 Hz).Fig.2plots the sensor-specific spectral densities (the ratios) under the visual condition (upper row) and the verbal condition (lower row).They are relative to the resting condition.All figures represent the results of nonparametric regression smoothing on the layout map as shown in Fig.1.In order to improve the MEG measurement sensitivity for detecting global changes in neural activity, the spectral densities are normalized by the mean of densities of all sensors measured under the corresponding conditions.In each figure, the left-most images present the data for Group I, the middle images present the data for Group L, and the right-most images present the ratios of the Group I data relative to Group L data.

Figure 3 .
Figure 3. Temporal variations of spectral densities in the ROI's (visual and frontal language areas).Spectrograms represent the spectral intensities (the ratios) of the major regions of interest (v1: primary visual area, b1: frontal language area in the middle frontal gyri, and b3: Broca's area) for Group I relative to Group L. The horizontal axis designates the measurement time (interval 1/15 second, total duration 20/3 seconds).The vertical axis designates the central frequency of the frequency bands (overall range of the central frequency 10-50 Hz, with steps of each 5 Hz wide).The left column presents the data under the visual condition, and the right column presents the data under the verbal condition.