Frontal activation patterns during Tetris game play and differences between high and low performers: a preliminary functional near-infrared spectroscopy study

Background: Tetris has recently expanded its place of activity not only to the original entertainment but also to clinical applications such as prevention of trauma flashback. However, to our knowledge, no studies focused on the cortical activation patterns themselves when playing Tetris in a natural form. This study aimed to investigate the activation patterns in the frontal cortex during naturally-performed Tetris for 90 seconds in 24 healthy subjects using functional near-infrared spectroscopy robust to artifacts by motion and electric devices. We also calculated the correlations of behavioral data with cortical activations, and compared the differences in activations between the high and low performers of Tetris. Results: The results demonstrated that significant activations in the frontal cortex during Tetris play had two factors, each showing a similar activation pattern. One of the factors was distributed over the lateral prefrontal cortex bilaterally, and the other was localized to the right prefrontal cortex. Moreover, in the high performers, the activations of the areas centered on the right dorsolateral prefrontal cortex (DLPFC) were estimated to increase and correlations of the activations between those areas and the other areas decrease compared with the low performers. Conclusions: It is suggested that high Tetris performers might reduce functional connectivity between activations of the areas centered on the right DLPFC and the other areas, and increase the local activations compared with low performers. It would be necessary to consider whether its visuospatial cognitive loads stimulate the appropriate areas of the subject’s brain to effectively utilize Tetris play for clinical interventions.


Background
Tetris is a worldwide falling block puzzle computer game that has attracted many people over three decades. Studies using Tetris have been conducted across various disciplines mainly in the field of behavioral and cognitive sciences (Haier et al. 1992a(Haier et al. , b, 2009 There are few studies investigating brain activity related to Tetris play (Haier et al. 1992a fNIRS is a non-invasive neuroimaging technique, which requires little restriction compared with other neuroimaging methods such as positron emission tomography (PET), functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG). It shows low sensitivity to artifact by motion and electric devices. In addition, considering its high portability and low cost, the use of fNIRS is expected to spread worldwide (for a review see Pinti et al. 2018). In fNIRS, near-infrared light penetrates into tissues and is differentially absorbed by hemoglobin depending upon the oxygenation state and its optical path length in the tissues (modified Beer-Lambert Law). This relationship enables fNIRS to detect relative changes in concentration of oxygenated hemoglobin ([oxy-Hb]) and deoxygenated hemoglobin ([deoxy-Hb]) by emitting near-infrared light at several different wavelengths into the cortex and detecting its remnants (Jöbsis 1977;Hoshi 2003;Ferrari et al. 2004).
Areas with high neural activity show increased oxygen consumption followed by supply oxygenated hemoglobin (neurovascular coupling) (Fox and Raichle 1986;Hoshi et al. 2001). This means that neural activity is measured indirectly by using relative changes in regional cerebral blood volumes (rCBV). Using functional neuroimaging techniques, except for fNIRS, Tetris was first studied by Haier et al. (1992a), using PET in eight male healthy high performers trained intensively in Tetris. These subjects showed decreased metabolism over all brain areas induced by Tetris training. This led authors to conclude that activity of brain cortical areas was reduced by learning. They also reported that cortical metabolism after Tetris training could increase in areas needed for high Tetris performance, including the right precuneus and left cingulate (Haier et al. 1992b Only an electroencephalography (EEG) study investigated neural activation and network activity during naturally-performed Tetris. In the study, the activations were measured on Fz, F3, F4, C3, C4, T3, T4, P3, P4, O1, and O2 of the international 10/20 system for EEG, and the network activities were investigated between Fz and the other measurement points. Its findings showed increasing activations across the cortical areas and elevated network activities between the motor planning area in the frontal cortex (Fz) and the other cortical areas (the sensory and executive brain regions) as difficulty of Tetris increases (Rietschel et al. 2012).
Based on the findings of previous Tetris studies, we hypothesized that significant activation would be found in the lateral prefrontal cortex. Furthermore, as an exploratory investigation we attempted to identify frontal areas required to play Tetris successfully taking into account the relation between activation and performance.

Subjects
Twenty-four right-handed healthy Japanese subjects participated in this study (13 men, 11 women; mean age ± standard deviation (SD) 27.3 ± 6.8 years). None of them had history of psychiatric or neurological disorders.
Written informed consent was obtained from all the subjects before the experiments. The procedures and methods in this study were consistent with the policies described in the Declaration of Helsinki. This research was approved by the 8 Ethics Committee of Osaka University Graduate School of Medicine.

fNIRS Measurement
Relative changes in [oxy-Hb] and [deoxy-Hb] were measured by using fNIRS (ETG-4000; Hitachi Medical Corporation, Tokyo, Japan) during playing Tetris. In this study we used the ETG-4000's probe mounted with fifty-two measurement points (channels). Seventeen laser diodes (emitters) and sixteen photodiodes (detectors) were arranged reciprocally at 3 cm intervals on a piece of thermoplastic shell (3 × 11) covering most of the frontal and part of the temporal surface areas (Fig. 1). Detection depth at the channels was 2-3 cm under the scalp. The lowest center photodiode was located on Fpz using the international 10/10 system for EEG. The

Tasks
In the present study Tetris was employed as the activation task. Tetris was performed using Nintendo Game boy pocket (Nintendo, Kyoto, Japan) which is a handheld game console. In this Tetris, on the monochrome liquid crystal display the 9 player moves and rotates blocks that fall one by one. There are seven kinds of blocks consisting of four square blocks called Tetrimino. Tetrimino is moved by the cruciform button under the left thumb and rotated by the two round buttons under the right thumb aiming to arrange Tetrimino without gaps at the bottom. The lines vanish at the moment when they are made, and numbers of the vanished lines are recorded as LINE, which represents the behavioral data collected in this study. In addition, the upcoming Tetrimino is shown in the Preview box of the right side on the display.
In the baseline task, consisting of a resting state before and after Tetris execution, subjects were instructed to gaze at the display and tap their right and left thumbs reciprocally on the buttons of the console at a uniform pace.

Experimental Procedure
Before the experiment, the subjects played Tetris for several minutes to confirm they knew how to play the game. Tetris used in this study can produce the game version from Level 0 to Level 9 in order from the one with the lowest falling speed of Tetrimino (blocks), and at first, we intended to have the subjects play Level 0 Tetris. However, if the subjects answered "Yes" to the question of whether you are good at Tetris and the fact was confirmed during the above pre-experimental play, we applied Level 1 Tetris to the subjects because we predicted that Level 0 would be too easy to induce the cortical activations characteristic of Tetris for them (Haier et al. 1992a; de Sampaio Barros et al. 2018). Accordingly, the subjects were given tasks closer to the individual optimal levels of difficulty, which had been shown to induce more right prefrontal activation than the easy or hard level during Tetris software of ETG-4000 was set at the "integral mode". In that configuration, mean changes in [oxy-Hb] for the baseline states 10 sec before the start of the activation task and after 50 sec from the finish of the activation task, were corrected to 0 mM•mm by using linear fitting. The activation task period was set at 90 sec, and the recovery period from the finish of the activation task period up to baseline stabilization was set at 50 sec. The moving average method for 5 sec was applied to smooth out short-term motion artifacts.

Detection of Activation Patterns
For statistical analyses, firstly, the authors determined frontal activation patterns during Tetris play. Mean changes in [oxy-Hb] during Tetris play were calculated for each subject in each channel, followed by a two-tailed single-sample t-test, which is equal to a paired t-test against zero (mean changes in [oxy-Hb] during baseline periods) to detect significant activations. The two-tailed single-sample t-test was conducted on all fifty-two channels; the significant α levels of 0.05 were corrected by the false discovery rate (FDR) (Benjamini and Hochberg 1995) to control multiple comparisons, which could be applied to fNIRS (Singh and Dan 2006). Next, we expected that components of the Tetris-induced activations could be classified by the functions, and this might give suggestion of the cortical areas to successfully perform Tetris, so that the following analysis was carried out. In order to extract channel groups showing similar patterns of changes in [oxy-Hb] from the significantly activated channels by the above single-sample t-test, factor analysis using major factor method with direct oblimin rotation was performed for mean changes in [oxy-Hb] during Tetris play of each subject in those channels. In each channel, the extracted factor, which indicated the absolute value of factor loading over 0.4 and higher than the other factors, was selected as the represented factor of the channel. Also, mean changes in [oxy-Hb] of each subject during Tetris play were averaged among the channels belonging to each extracted factor, and compared using paired Student's t-test.

Detection of Activations for High Performance
Pearson's correlation coefficients between LINE and the mean values of [oxy-Hb] changes in all subjects were calculated for each channel. Next, the twenty-four subjects were trisected into high, middle and low performer groups of 8 members based on their LINE values. One-way ANOVA using Tukey correction for multiple comparisons was performed with mean changes in [oxy-Hb] during Tetris play for each subject in each channel. In addition, to investigate differences in LINE in detail, independent Student's t-test was carried out to compare mean changes in [oxy-Hb] during Tetris play between the high and the low performer groups (8 members respectively) in each channel. Moreover, based on the findings of Rietschel et al. (2012), that the cortical network activity, i.e. functional connectivity measured by EEG increases as the difficulty of Tetris increases, we predicted that functional connectivity would be reduced in the high performer group compared with the low performer group, particularly in cortical areas required for playing Tetris successfully. Thus, the following analyses were carried out. After mean changes in [oxy-Hb] during Tetris play were determined for each subject, Pearson's correlation coefficients were calculated between each channel and the other 51 channels, and paired Student's t-test was performed on the 52 mean values of the Pearson's correlation coefficients over 51 channels, between both the low and high performer groups. Furthermore, for each channel, the 51 Pearson's correlation coefficients were compared using paired Student's t-test between the two groups.
In the above analyses, when controlling multiple comparisons, the significant α levels of 0.05 were corrected by FDR.

Estimation of Activations for High Performance
If the correction for multiple comparison using FDR in the previous section is too strict to detect significant channels, in order to estimate the frontal areas for successfully performing Tetris, we subsequently decided outlier channels on the basis of the statistic values larger than the mean + 2 SD.

Activation Patterns 13
The grand average waveforms of changes in [oxy-Hb] across all the subjects were obtained for each channel (Fig. 1). Fourteen channels (26.9%) demonstrated significant increases in [oxy-Hb] on a single-sample t-test with FDR correction (p = 0.0004 to 0.0120), and the largest t value was shown in channel 24 (t = 4.137) ( Fig. 1, Appendix Table 1). Factor analysis using major factor method with direct oblimin rotation was performed on mean changes in [oxy-Hb] during Tetris play of these significantly activated 14 channels, and two factors were extracted. While the channels with large factor loading of factor 1 were distributed on both sides of the prefrontal cortex, the channels with large factor loading of factor 2 were confined to the right side of the prefrontal cortex (Fig. 1). Additionally, there were no significant differences (t = 0.003, p = 0.998) among averages of mean changes in [oxy-Hb] during Tetris play in those channels belonging to each of the factor 1 and factor 2 (the average ± SD: 0.181 ± 0.229, 0.181 ± 0.247, respectively).

Correlation between Activations and LINE Pearson's correlation analyses between LINE and mean changes in [oxy-Hb] during
Tetris play for each channel were carried out, and no significant channels were found after FDR correction.

Difference in LINE
Based on LINE values (the mean ± SD: 8 ± 4), 24 subjects were divided into three groups of 8 subjects each, named high (13 ± 1), middle (7 ± 2) and low (3 ± 2) performer. One-way ANOVA for each channel found no significant channels after FDR correction. Independent Student's t-test of mean changes in [oxy-Hb] during Tetris 14 play for each channel between the high and low performer groups also showed no significant channels after FDR correction.

Comparison between High and Low Performers on Correlation Coefficients of Activations among Channels
The averages of the mean values of Pearson's correlation coefficients between each channel and the other 51 channels were 0.385 ± 0.161 for the high performer group and 0.554 ± 0.108 for the low performer group. Comparing with a paired Student's ttest, it was found that the correlation between channels was significantly larger in the low performer group relative to in the high performer group (t = 5.722, p < 0.001). For each channel, Paired Student's t-test for these Pearson's correlation coefficients between the two groups showed significant differences in twenty-nine channels after FDR correction (p = 1.981E-18 to 0.021) (Fig. 2).

Estimation of Differences in Activations between High and Low
Performers On the similar items described in Sect. 3.2., the frontal areas for successfully performing Tetris were estimated using the channels indicating the statistical values larger than the mean + 2 SD as outlier channels. Pearson's correlation analyses between LINE and Activation during Tetris play showed no outlier channels.
For reference, the channel with the largest correlation coefficient was channel 24 (r = 0.294, p = 0.163), and the channels belonging to factor 2 in the above factor analysis showed r > 0.20 while those belonging to factor 1 showed | r | < 0.135 (Fig. 3, Appendix Table 2). One-way ANOVA among the three groups found channel 24 as the only outlier channel (F > the mean + 2 SD = 2.221) (Fig. 3, Appendix Table 2). Multiple comparisons performed on channel 24 by Tukey's method 15 revealed that the activation of the high performer group was significantly larger than that of the low performer group (p = 0.0499). Independent Student's t-test between the high and low performer groups detected channels 12, 23, 24, 33 as outlier channels (t > 2.313) (Fig. 3, Appendix Table 2). In comparison between the high and low performer groups on the correlation coefficients of activations among channels, Paired Student's t-test for each channel showed channel 12 and 24 as outlier channels (t > 12.023) (Fig. 3, Appendix Table 2).

Summary of Results
In this study, we used fNIRS to determine frontal activation patterns in healthy subjects while playing Tetris. We found bilateral significant activations in 14 (26.9%) of the 52 channels investigated in both hemispheres (Fig. 1, Appendix Furthermore, factor analysis to extract channel groups showing similar activation patterns allowed us to extract two factors from these significantly activated 14 channels. Those channels with a large factor loading for the factor 2 were confined to the right-sided lateral prefrontal cortex (Fig. 1). Subsequently, exploratory analyses were carried out assuming the frontal cortical areas necessary for successfully performing Tetris, and channels with such a possibility appeared to overlap with those showing a large factor loading of factor 2 derived from the aforementioned factor analysis. 16 Among the frontal cortical areas necessary for successfully performing Tetris assumed in this study, the channel 24 was the only channel coincidently estimated by all the exploratory analyses, and the other estimated channels were radially distributed rearward around channel 24 (Fig. 3, Appendix Table 2 Tetris play, activation of the right DLPFC and the right inferior parietal lobe were high when the measurement focused on the two areas involved in the frontoparietal attentional network. This supports our finding that the subjects in the high performer group, many of who are thought to achieve a large flow-state during Tetris play, exhibited an increase in activation of the right DLPFC while playing the game. Based on these arguments, we consider that at least the right DLPFC (BA 46) might be a crucial frontal cortex area for successfully performing Tetris.

Function of the Right DLPFC during Tetris play
In previous studies, attention, mental rotation, working memory, planning and decision making during speeded manipulations for visuospatial tasks were mentioned as the main cognitive functions required for smooth performance of

Functional Connectivity
It is noteworthy that in this study by using a method to infer the frontal cortex areas necessary for successfully performing Tetris based on the EEG evidence that cortical network activity increases as the difficulty level of Tetris increases (Rietschel et al. 2012), we quantitatively compared functional connectivity between the high performer group and the low performer group. An interesting finding obtained was that functional connectivity declines in the high performer group, mainly in the channels involved in successful performance of Tetris. This phenomenon is likely related to the neural efficiency hypothesis, indicating that high performers exclude inefficient cortical functional connectivity, using the brain more efficiently than low performers (Haier et al. 1992a In the studies to compare high performers with low performers of cognitive tasks as this study, differences in activations across the brain lobe level may be found due to differences in strategies. Therefore, it would be worth investigating across wide brain areas. Fourth, recent evidence indicates that signals

Conclusion
The activations of the frontal cortex during naturally-performed Tetris were distributed over the lateral prefrontal cortex in both hemispheres. The activations included two factors, which were the activations estimated to associate with the Tetris performance only in the right hemisphere and performance-independent activations in both hemispheres. Furthermore, our findings suggest that high Tetris performers while producing increased activations of the areas centered on the right DLPFC (BA 46) may disengage those from unnecessary functional connections with other cortical areas to use the brain more efficiently, compared with low performers. Thus, since the cortical activation patterns during Tetris play seem to be considerably different depending on the subject's performance, in order to effectively utilize Tetris play for clinical interventions, it would be necessary to 20 consider whether its visuospatial cognitive loads stimulate the appropriate areas of the subject's brain (e.g. competing with visuospatial trauma memory). This might help explore better interventions and avoid wasting time applying ineffective interventions. Future studies may confirm and extend the findings of this study by using larger samples or by exploring brain activity during Tetris play in wider brain areas.

Declarations
Footnotes 1 The web site where Tetris is certified as a "most ported videogame" by Guinness World Records.

Ethics approval and consent to participate
The study was approved by the Ethics Committee of Osaka University Graduate School of Medicine. Written informed consent was obtained from all the subjects.
The procedures and methods in this study were consistent with the policies described in the Declaration of Helsinki.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. The high (thick line) and low (thin line) performer groups' grand average waveforms of chang

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.