[Purpose] The purpose was to examine relationships between variables characterizing the 20-second stepping test movement pattern, assessed with an infrared depth sensor (KinectTM), and measurements of dynamic and static balance. [Participants and Methods] A total of 27 independent-living, older adults (7 males and 20 females) participated in this study. For each participant, the stepping test total movement distance, movement displacement, knee movement distance, and step number were calculated from Kinect closed joint-point coordinate data. Dynamic and static balance were assessed using a NeuroCom Balance Master Platform system. [Results] The stepping test total movement distance had a moderate negative correlation with endpoint excursion (r=-0.469) and a moderate positive correlation with total movement distance, corrected for knee movement distance (r=0.557). Step numbers had a moderate negative correlation with stepping test total movement distance (r=-0.667) and total movement distance, corrected for knee movement distance (r=-0.531). Division into high and low-balance sub-groups (based on endpoint excursion or sway velocity scores) revealed that stepping test total movement distances and movement displacements were greater, and step numbers were fewer, in low balance groups. [Conclusion] The stepping test, combined with a KinectTM-assessed movement pattern provides a simple, objective, reliable means for assessing balance in community-dwelling, independent-living older adults.

The purpose of this study was to determine the effects of a MotoTiles exercise on balance ability in communitydwelling older adults. The participants were 11 women who lived in T city and had performed a community-group exercise. The exercise was carried out twice a week for a period of 12 weeks using the local community center. For the balance index, a 20-sec stepping test was performed with the eyes open, and the total head movment distance (TMD), maximum movment distance (MMD), and the distance of the knee joint point (KMD) were used using KINECTTM infrared sensor. The pace of both legs (PACE left and right) and its coefficient of variation (CV) were used using Optogait infrared sensor system. Functional reach (FR), up-and-go (TUG), and 30sec-chair stand (30-CS) were also examined. As a result of 12 weeks of exercise, there was significant improved on TMD, MMD, PACE left CV, FR, TUG, and 30-CS (effect size were moderate to large). No significant changes were observed in other infrared sensor-related indexes, but overall, the head and knee sway during the stepping test and the variation when raising the legs tended to be smaller. The MotoTiles Exercise appears to be useful for assessing one of balance ability in older adults.

Voice recognition technologies are widely applied to the general-purpose devices, such as car navigation systems or mobile phones. However, cognitive-loads of voice operations have not been discussed sufficiently. Since it is necessary to translate a command to a verbal phrase, cognitive-loads of voice operations for simple tasks may become higher than those of manual operations. In this study, we analyzed microsaccade rates to compare the cognitive-loads between voice and manual operations objectively. Subjects performed a parafoveal transition detection task of a geometry pattern in which they responded by their voice or mouse clicks. We also evaluated the task performances of simultaneously imposed an RSVP (Rapid Serial Visual Presentation) task to estimate effects on cognitive distractions. The results show that microsaccade rates increase synchronously with the timing of voice responses, and task performances of the RSVP task decline immediately after voice responses compared with the results of manual responses. These findings suggest that voice operations have an effect to disengage subjects' attention from visual targets, and this attentional disengagement might cause the cognitive distraction of gazing objects.

[Abstract] It is believed that median filtering outperforms traditional linear filtering in preserving signal edges while reducing noise. However, this does not always mean that the median filter is superior to the linear filter as a smoother. By comparing the bias-variance characteristics of median filtering and linear filtering, we demonstrate that a linear filter (moving average filter) is a better smoother than the median filter for inputs that have Gaussian or uniform distributions and relatively small signal-to-noise ratios, whereas the median filter outperforms the linear filter for inputs with Laplace distributions. [要旨]メディアンフィルタは, それが提案されて以来, 決定論的信号に対するエッジ保存能力が注目され, エッジ保存が重視される状況下での平滑化フィルタとして, 線形フィルタに勝るものとして頻繁に用いられるようになってきている. しかし, ガウス雑音の仮定のもとでは, メディアンフィルタのエッジ保存能力が線形フィルタを凌駕するのは高SN比入力に対してのみであり, SN比が低下するに従いほとんど両者のエッジ保存能に差はなくなるとの報告がある. また, 同じくガウス雑音に対して, 高SN比でない限り, 出力の平均二乗誤差で見る性能評価では, むしろ線形フィルタの方が優れているとの報告もある. これらは両フィルタを比較するための重要な情報ではあるが, データ処理の現場で実際にどちらを使うべきかの判断材料とするには十分でない. そこで本報告では, 各種雑音に埋もれたエッジ信号入力に対する, メディアンフィルタと線形フィルタ (移動平均フィルタ) の出力のバイアスー分散特性を比較し, 長い尾を持たない雑音 (ガウス分布雑音と一様分布雑音) でかつ高SN比でない場合はむしろ線形フィルタ性能の方が良く, 長い尾を持つラプラス分布雑音の場合はSN比に関わらずメディアンフィルタが良いことを示す. これは一般に流布しているメディアンフィルタの線形フィルタに対する優位性「神話」への警告であり, また, 実際にどちらのフィルタを使うべきかの判断材料の提供でもある.

type:Departmental Bulletin Paper [要旨] 本研究では, 周辺視野の特性を考慮した網膜数理モデルを対象に, 計算資源の制限の中で計算処理を最大限に効率化することを目的として, 種々の並列計算アルゴリズムを適用し, 富士通製並列計算機PRIMEHPC FX10と汎用のパーソナルコンピュータとの間で計算時間の比較を行った. その結果, 各種の神経細胞層を個別のモジュールとして扱うことで，メッセージ通信によるサイクリック分割とパイプライン通信が実装可能であり，微分方程式の回帰参照の問題が回避できることが示された. また, メッセージ通信が困難な部分にマルチスレッド並列を適用することにより, 効率的な並列処理が実現できた. これらのことから, FX10の性能を最大限に引き出すためには, MPI（Message Passing Interface）を用いたメッセージ通信と, マルチスレッド処理とのハイブリッド並列化が有効であり, これにより高効率で高速な並列計算が可能となることが示された. [Abstract] The purpose of this study is to develop effective methods to compute a mathematical model of the retina which considers peripheral properties of retinal cells in the restriction of computation resources. We compared the machine time with regard to several parallel processing algorithms between the Fujitsu PRIMEHPC FX10 supercomputer system and general personal computers. The results indicate that the implementation of cyclically divided blocks and a pipeline processing with a message passing interface, by dealing with each cell layer as a module, is able to avoid problems of the regression reference of calculation of differential equations. For program codes which are difficult to implement message passing interfaces, we obtained a practical solution by applying multithread processing instead. Furthermore, these results suggest that the performance of the FX10 system is maximized by using a hybrid parallel processing method with message passing interfaces and multithread processing.

[要旨] 従事すべき作業とは関係の無い考え事に注意が奪われている状態では, 覚醒水準とは無関係に判断や応答といった認知パフォーマンスが低下するため, 判断ミスなどのヒューマンエラーによる事故をもたらす危険性が高くなるが, 作業者がそうした状態に陥っていることを客観的に評価することは極めて困難である. 本研究では, 課題とは無関係な考え事をしている状態を課題非依存性思考状態と定義し, このような課題非依存性思考状態が, 認知パフォーマンスに与える影響の客観的な評価を目的として, あらかじめ提示された8桁の数字の記憶を維持しつつ, 同時に課せられた視覚探索課題を遂行する際の, 目標探索時間や前頭前野外側部の近赤外線スペクトロスコピー(Near-Infrared Spectroscopy: NIRS)信号を解析した. 実験の結果, 複数の視覚属性の組み合わせにより定義された視対象の中から唯一の特徴を持った目標を探索する場合においてのみ, 課題非依存性思考の有無により目標探索時間が影響されることが示された.また, この条件においてはNIRS信号から推定された前頭前野外側部の賦活度が高まることも示唆された.これらのことから, 課題非依存性思考によって前頭前野の情報処理リソースが消費されることが示唆され, こうした状況下では, 認知タスクのパフォーマンスが影響を受ける可能性が示された. [Abstract] Working with stimulus-independent thoughts(SITs) often causes the decline of cognitive performances, such as judgments of signals or response times to certain tasks. It might lead to fatal accidents called "human errors" independently of arousal level. The detection and prediction of these errors is extremely difficult, and the measurement technology of inattentiveness has not yet been established. In this study, we investigate the effect of SITs on the performance of visual search tasks. SITs are the streams of thoughts that are unrelated to immediate situations. The subjects memorized the numeral of eight digits as an SIT while carrying out visual search tasks. The reaction time and the near-infrared spectroscopy signals from the lateral prefrontal cortex(PFC) were analyzed to evaluate the effect of SITs objectively. As a result, the average reaction time for searching targets from distracters only in terms of their feature conjunctions increased when having SITs during the tasks. Also, the SITs in these tasks enhance the activity of the lateral PFC. These results suggest that the SITs occupy the resource of attention and it might affect the performances of certain cognitive tasks under these conditions.【注記】本研究は, 近畿大学生物理工学部戦略的研究NO.11-Ⅰ-3 2012, および, JSPS科研費24560534の助成を受けた.

type:Departmental Bulletin Paper [要旨]手振れのような振動や回転を含んだ映像は, 吐気や目眩などの乗り物酔いと同様の症状を引き起こす. この症状を映像酔いと呼ぶ. 本研究では, 映像酔を誘発する映像要素を特定することを目的として, 特定の運動成分からなる映像を観察中の主観的な酔いの程度と固視微動および心拍変動の解析を行った. 映像が水平方向に移動するパン, 回転方向に変移するロール, 拡大と縮小により生じるズームの3種の映像要素により構成される運動するランダムドットパターン(ダイナミックランダムドット)を観察中に計測された固視微動データに対して周波数解析を行った結果, 実験の進行に伴って徐々に視線の揺らぎが増し, 注視の維持が困難になることが示された. また, パンまたはロール成分のみからなる動画を観察する際の心拍変動を解析した結果, パン成分のみの映像の方が, 心拍変動のHRV(HF/LF)が低い値となることが明らかとなり, パン成分が交感神経の亢進をもたらすことが示唆された. これらの結果から, 水平方向の運動成分を持った映像を長時間視聴することによって生じる視覚系と前庭系および体性感覚系との矛盾の蓄積が, 映像酔いをもたらしているものと推察され, 固視微動や心拍変動から酔いの進行の定量化が可能であることが示された. [Abstract] Videos with camera shake containing strong vibrational or rotational motion may cause some symptoms similar to motion sickness such as nausea, dizziness and headache. It is called cybersickness, or visually induced motion sickness. In this paper, we have evaluated cybersickness by analyzing fixation eye movements and heart rate variability of subjects viewing videos of which content was restricted to certain visual factors in order to identify the effective motion component for cybersickness. Firstly, fixation eye movements while subjects were watching random dots which consist of each of three visual factors, such as Pan, Roll and Zoom were measured and evaluated by spectral analysis. The results show that variability of the eye movements is increased as the experiment session progresses and it suggests that it is difficult to maintain the attention of fixation as cybersickness progresses. Secondly, we have evaluated the heart rate variability while subjects were watching the motion pictures which are not random dots but well-controlled motion pictures of an indoor scene and an out outdoor scene. Spectral analysis of the heart rate variability demonstrates that HF/LF measure in the Pan condition has lower values than in the Roll condition and it suggests that Pan motion in the video increases sympathetic nerve activity. These results indicate that cybersickness is caused by cumulative inconsistencies among visual, vestibular, and somatosensory systems for a long time while watching videos. Also, we have shown that progression of sickness can be quantified by analyzing fixation eye movements and heart rate variability. 本研究は近畿大学生物理工学部戦略的研究No.07-I-02, 2007の助成を受けた.

初期視覚神経系の情報処理の特徴を定式化した顕著性マップモデルは，顕著さの勾配に従って走査することで視線の移動を再現することができるが，特定の候補点の中から視線の移動先が選出されるために，ある程度長時間のシミュレーションでは循環が生じてしまう．本研究では，同じ視対象を繰り返し注視することを防ぐ復帰抑制と，周辺視野において生じる目標対象と同じ属性への応答を強化させる復帰増強と呼ばれる現象に着目し，従来の顕著性マップモデルに対して，これらの緒特性についての拡張を行った．さらに，網膜上の錐体細胞の密度分布に起因する第1次視覚野における皮質拡大を考慮し，視線移動のたびに生じる大脳皮質における網膜像の表象の変化を実装した．提案モデルによる視覚探索のシミュレーションを行った結果，従来モデルよりもターゲットの発見率が向上され，また，視線移動の軌跡が一定の戦略を持った探索時の視線移動に似たパターンとなること

Departmental Bulletin Paper

本研究では, 視覚系の情報獲得のプロセスにおける眼球運動の機能を解明することを目的として, ランダムに配置された複数の片仮名を同時に短時間呈示した際の, 被験者の注視点の位置と正しく認識された文字の空間的な分布との関係を解析する実験を行った. その結果, saccadeによって注視点の移動が生じる前後の時間帯において, 正しく認識された文字は, (1)視線が静止しておりsaccadeが計画されていない状態では, その時点の注視点を中心に, (2)視線は静止しているが次の注視点に向かってのsaccadeが計画されると, 視線の移動が生じる前の段階で新たに予定された注視点を中心に, (3)実際にsaccadeが生じた後は, 新たな注視点を中心に分布することが示された.

This study proposes a method for detecting microsaccades based on a discrete ulse transform analysis (DPT). The general methods for microsaccade detection need to decide the threshold for a differentiation signal of eye movements after enoising by the median filter. The median filter well maintain sharp edges configured by microsaccades, however, theoretical prediction of the behavior is difficult.Because of this unpredictability, the trial and error process is necessitated to decide the threshold value or the order of the filters. The proposed method is to detect microsaccades based on the features of the pulse series produced by DPT. The eye movement signal is divided into smoothed and detail components by multiple orders of LULU filters, then the pulse series which represents the dynamics of movements is calculated. The results show the detected microsaccades is more precise than general method. Furthermore,the start- and end-point of each microsaccades is clearly discriminated.

動画像中の激しい振動や急速な回転は，吐気やめまいなどの乗り物酔いと同様の病的な症状を引き起こす．これを映像酔いと呼ぶ．家庭用テレビの大型化や高精細化と相俟って，日常生活における映像酔いのリスクが高まっているものの，映像酔いが生体に及ぼす影響については，未だ全容の解明には至っていない．映像酔いは，視覚情報と前庭系の不調和に依存して発生するものと考えられ，眼球運動制御系への強い関与が予想されるものの，固視微動への影響に関する詳細な解析例はほどんど存在しない．そこで本研究では，映像酔いを生じさせる要素を含んだ映像を一定時間だけ観察する試行を繰り返し実施し，インターバル中の眼球運動を計測して固視微動の解析を行った．固視微動のうち，微小な視線のジャンプとして観測されるマイクロサッカード（MS）は，視覚的注意の集中度合いによって，その発生頻度が変化することが知られている．映像酔いは集中力の低下をもたらす

自動車運転等の視覚依存性の高い作業下における覚醒水準の減衰は，視覚機能や運動能力の低下をもたらし，ヒューマンエラーによる事故のリスクを増加させるため，その変動を客観的に捕らえるための判断基準の確立が望まれている．覚醒水準と眼球運動の動的特性との関係を調査した報告によると，高速で断続的な視線のジャンプであるサッカードの最高速度と継続時間の比 PV/D が，覚醒水準と強い相関を示すことが明らかにされているが，ヒューマンエラーに結び付くような危機的状況の直前の変動については検討されていない（植野ら，2000）．そこで本研究では，覚醒水準が低下して睡眠に至る直前に発生するとされる SEM（slow-rolling eye movement）を覚醒維持の限界点として定め，一定間隔 でステップ的に呈示位置を変える視標をサッカードにより追尾させる実験を実施し，SEM 発生に至るまでの PV/D と，サッカード間の固視微動の解析を行った．その結果，PV/D の値が減少す

2023年度の研究目的は，「ウェアラブル型センサーや画像処理による受講動作の検出」および「検出した受講動作に対する時系列モデリング」であった．本研究目的に対し，オンデマンド型講義におけるビデオ講義を対象として，学習者による講義ビデオの操作 (再生/一時停止/送り/戻り) およびノーティング動作を検出するための手法を構築した．また，検出した受講動作に対する特徴量を定義することにより，受講動作の時系列モデリングに寄与した．さらに，オンライン環境において，受講者が PC を用いて文書を作成するような学習環境を対象として，受講者の文章作成の過程を可視化することを目的に対して，(a) キーボードおよびマウスのイベント検出，(b) 画像処理によるアクティブなアプリケーションの検出，(c) 文章の編集操作の検出，(d) 文章の編集操作における特徴量の定義，(e) 定義した特徴量に基づいた受講者の編集スタイルの分類のための手法を構築した． 研究期間における研究目的は，(a) 講義ビデオに対する操作ログの抽出と時系列モデリング，(b) ウェアラブル型センサーや画像処理による受講動作の検出，および (c) 検出した受講動作に対する時系列モデリングであった．本研究目的に対して，(a) 画像処理によるアプリケーションのウィンドウ検出，(b) Word ウィンドウ内の変動 に基づいた文章の編集操作 (読む/スクロール/領域選択など) の分類を行うための特徴量の導入，(c) 導入した特徴量により表現される編集履歴に基づいた受講者の編集スタイルの分類，(d) ビデオ講義における講師の動作 (板書/説明動作) および受講者のノーティング動作の検出，(e) 両者の相互関係を表現するための時系列モデルの構築，(f) 時系列モデルにおける感度分析のための手法を構築し，それらの有用性を実験により評価した．

In this study, we analyzed fNIRS signals and fixation eye movements during attention control tasks to create an objective method for measuring the information processing involved in sustained attention. The results of the fNIRS signal analysis suggest that the activation of the brain regions involved in visual attention depends on the task. However, the analysis of fixation eye movements revealed no apparent differences in microsaccade frequency or direction. These findings indicate that fNIRS analysis can be used for attention-based cognitive rehabilitation.

We used the KINECT sensor to evaluate the movement patterns of the 20-second stepping (ST) and chair standing test (CS), and identified that these two variables have strong discriminatory performance for detecting “need for assistance” in older adults. We specifically focused on ST from multiple tests for nursing/nursing care groups, reevaluated its reliability, reconfirmed the conditions that the test should have, and compared and examined the differences in testing depending on the presence or absence of nursing care. We also compared the ST data with the Balance Master Evaluation Method to confirm the validity of ST as a balance test. Finally, using data from a large-scale group health checkup in Tokyo, we created a draft of a five-level evaluation method for ST and CS tests, and attempted to set standards. Screening using this evaluation method seems to be effective, and we would like to introduce it to many communities in Japan.

For face-to-face collaborative learning, we constructed methods for extracting the relationship between nonverbal behaviors of learners and cooperation. Specifically, we constructed (i) methods for detecting non-verbal motion using a glasses-type device and/or image processing, and (ii) a time-series model for expressing the interaction between behaviors of learners. Furthermore, we constructed a method to clarify the relationship between the parameters in the time series model and the cooperation between the learners, and confirmed their effectiveness by experiments.

The purpose of this study is to apply the gaze behaviors as an objective measure of neuroendovascular of cerebral aneurysms to therapy training. We constructed an experimental environment and analyzed gaze data obtained from three pairs of specialists and trainees neurosurgeons. First, trainees were less likely to switch their gaze in response to information, and second, specialists were more likely to disperse their gaze beyond the aneurysm being therapy. Future work is needed to determine if these results are not just an occurrence in a subset of the study population, but occur in many neurosurgeons, and to apply them to training for neuroendovascular therapy.

Network analysis of multidimensional biological signals is a very useful tool to understand the mechanism of signal generation. Although network analysis has been widely used in brain function analysis, most of them are based on linear correlation analysis using the Pearson product-moment correlation coefficients. In this study, we conducted a research on multidimensional network analysis methods using network analysis of intractable epilepsy as a touchstone. We introduced a partial correlation function to accurately capture the network structure in multivariate connectivity analysis, introduced a directional graph that takes into account time delays. We have also developed a macroscopic network analysis method called "average connectivity strength tracking.

Two objective indices of ADL named the chair-standing and the marching test were assessed by optical sensing technology in physically independent versus physically dependent older adults with the aim of examining the difference in movement performance by these two groups of people. Based on the results of ROC analysis, the Kinect-v2 system that used optical sensing technology was found effective in screening physically dependent older adults with a diagnostic accuracy of 70%. However, detection of functional independence as well as level of care needed in rapidly growing frail older adults in Japan by performance of only one chair stand movement by Kinect-v2 indicates that this technique could be a suitable and user friendly mode of testing for these people. In order to analyze a large amount of data at a time, it may be necessary to modify the program of analysis that could be addressed in our future studies.

The purpose of this study is to establish an objective evaluation of the functions in neural network involved in visual attention control. We conducted spatial cueing tasks in which the subjects had to concentrate or move their visual attention while maintaining their gaze. The concentration of attention was quantitatively evaluated based on an analysis of the fixation eye movements measured in the tasks. We also analyzed the event-related fNIRS signals to objectively assess the brain activity associated with attentional control. The results showed that the microsaccade-rate varied with passive attention shifts, indicating that the microsaccade-rate is a potential candidate for an objective criterion of attentional state. In addition, the results of the event-related fNIRS analysis suggest that the responses of the brain neural network associated with attention maintenance or interruption can be evaluated quantitatively.

In this study, we developed a method for removing or separation of additive and convolutive artifacts artifact from multi-dimensional biosignals with independent vector analysis. We applied the method to extract fetal electrocardiogram from maternal abdomen as a test of our proposed method. As a results, the proposed method showed better performance rather than the conventional independent component analysis method. In addition, we developed a method for estimating a distribution of fetal cardiac potential which was observed on the maternal abdomen. The distributions are of help in order to find electrodes’ coordination which has better performance on the maternal abdomen.

The purposes of this study were to develop an index of measuring independence objectively in ADL of elderly by using an optical sensor (OS) and to determine the effect of exercise using OS. During the course of the study, the OS named KINECT was upgraded to (KINECTv2) by the manufacturer which increased the device’s accuracy. To address the compatibility issue of KINECTv2, we recollected an ADL (chair stand (CS)) data and analyzed the coronal plane angle (CPA) data while performing CS in apparent healthy elderly (HE), and in frail elderly (FE). The CPA exists between a line that connects one’s shoulder-center and waist versus the person’s vertical axis (sagittal plane). The CPA was found to be significantly smaller in HE than in FE while performing CS indicating that OS is effective in assessing independence to complete the CS. However, no exercise intervention efficacy was noted on CS when assessed by the OS, thus further study is needed to address this issue well.

The purpose of this study is to elucidate the information processing mechanism in visual attention system of brain. We analyzed involuntary fixation eye movements to quantify the degree of focal attention, and near-infrared spectroscopy (NIRS) signal to establish an objective monitoring method of the attention control system. As results from the experiments conducted in this study, we obtained a signal processing technique for extracting microsaccades with high accuracy and a mathematical model necessary for drift eye movement analysis. Moreover, we constituted experiment conditions suitable for brain function measurement by NIRS and its signal processing methods to avoid various artifacts. These results suggest that measurement techniques to elucidate the neural mechanism of attention were confirmed.

In lectures by using blackboards, lecturers give lectures in combination of explanation by speech and writing on blackbirds. On the other hand, students take notes and/or listen to lectures corresponding to behaviors by lecturers. In this research, we have analyzed the relations between behaviors by students corresponding to them by lecturers. Concretely, we have constructed the image processing methods for the extraction of behaviors by lecturer and students and the modeling methods for the relations between behaviors by them. Moreover, we have clarified the relations between behaviors by lecturer and students and the relations between the understandings and behaviors by students.

(1) We introduced L and U filters having not been well-known as morphological filters and the digital pulse transform (DPT) constructed of L and U filters. (2) We proposed a new method for detecting randomly generated signals buried in noise, in particular, a new method to detect saccades in human eye movement data. We attained error-free detection of the saccades by this method which has not been achieved by traditional linear filters. This method can be widely applied to detection problems in biomedical environmental data processing. (3) Agaist the commomly accepted theory that median filtering outperforms traditional linear filtering in preserving signal edges while reducing noise, we theoretically proved that moving average filter is a better smoother preserving signal edges than the median filter under a practical wide range condition.

The main objective of the research is to develop a method for analyzing functional brain networks using electroencephalography (EEG) and functional Near-Infrared Spectroscopy (fNIRS). We introduced fNIRS signal because fNIRS signal have the advantage that there is no interference from electro-oculogram (EOG) signal. However, it has some drawbacks, i.e., it contains not only functional signals of the brain but also systemic physiological signals evoked by physiological changes and body motion. We evaluated a separation method based on differences in hemodynamic modalities and based on independent component analysis (ICA). We, then, propose a new method which combines the above two methods in order to compensate for the drawbacks of both the methods in a complementary style. The results showed that the proposed method could be removed various artifact which often contaminated fNIRS signals.

The purpose of this study is to evaluate the influence of ``undetectable motions'', the motion information under the perception threshold, on cognitive functions by statistical analyses of the property of fixation eye movements. Subjects performed rapid serial visual presentation (RSVP) tasks with undetectable or ambiguous motions in the background while maintaining their fixation. We evaluated the influence of these motions on top-down attention mechanism by analyzing fixation eye movements. The results suggest that microsaccade rates were correlated with subliminal motion information, and fluctuations of drift eye movements were modulated by low-contrast motions in the peripheral vision.

We have measured multi-channel bio-signals, i. e., electroencephalogram, heart rate variability, blood pressure, eye movement and pupil diameter variability in order to evaluate the influence of visual stimuli to the wakefulness maintenance state against sleepiness. We performed time-frequency analysis, conventional spectral analysis, and feedback modeling to those signals. All results showed that physiological activities in wakefulness maintenance state have different features from those in onset of natural sleep. In addition, we have introduced new artifact removal methods to the EEG analysis.

本研究は,視覚的注意機構と眼球運動神経系とがほぼ共通した神経機序に基づくことを示した心理物理学・神経生理学的知見に基づき,視覚的注意機構の機能を考慮した眼球運動神経系の制御モデルの構築を目的とするものである.昨年度までの成果として,一点を固視している状態で発生するsquare-wave jerkに着目し,従来のサッカード眼球運動モデルを拡張して,大脳皮質からのサッカード計画に関する信号と,中脳上丘の注視制御機によってsquare-wave jerkの発生メカニズムを説明する数理モデルを提案した.Square-wave jerkとは,マイクロサッカードと呼ばれる微小な視線のジャンプが,ごく短期間に連続して生じることにより形成される矩形波状の眼球運動であり,マイクロサッカードにより生じた注視目標からの視線のズレ(ポジション・エラー)に対し,別のマイクロサッカードによって自動的に補正する機構により生成されるものと考えられる.本研究では,ポジション・エラーとして発生するマイクロサッカードが,上丘中間層における注視細胞とサッカード細胞との相互抑制的制御機構のノイズ的活動に起因するものと仮定し,注視細胞の活動の低下によって,近辺のサッカード細胞が十分に抑制しきれなくなったとき,マイクロサッカードが発生してポジション・エラーとなると考え,この動作のモデル化に取り組んだ.昨年度までのモデルには,サッカード期間中にのみ,サッカードの振幅や速度に関する信号を発生する上丘の応答特性に問題があり,常に一定の期間だけ発火するような実装となっていた.今年度の成果として,この上丘の応答に関するモデルの記述を見直し,サッカードの速度・振幅特性をコード化した信号を発生するよう修正を施し,その成果を「Asian Conference on Vision」,および,「日本神経回路学会」にて発表した.

近年増加の一途にある自動車運転中の電話利用時の交通事故は,運転者が会話に注意を奪われたことが原因とされる場合が多い.これまでにも,視覚課題への反応時間の測定などによって注意の客観的に測定する試みがなされているが,応用価値の高い生体信号解析に基づいた測定手法の確立には至っていない.このような背景のもとで,本研究では,ある視覚対象を凝視する際に生じる不随意の微小眼球運動(固視微動)を解析することにより,注意の集中度合いの定量的測定を試みた. 本研究では,被験者の注意を,1)注視位置に集中させた場合,2)注視位置の周辺に分散させた場合,3)周辺視野の特定の位置に注視を集中させた場合,の3つの状態にコントロールする実験課題を提案し,各実験課題のもとで得られた固視微動から,発生したすべてのマイクロサッカードを抽出し,その発生頻度,振幅,ピーク速度について実験課題間での比較を行なった.また,固視微動からマイクロサッカードを取り除いて連結したslow cumulative eye position curveを求め,FFTによる周波数解析を行うことにより,ドリフト成分の周波数解析を行った. 解析の結果,マイクロサッカードの発生頻度は,実験課題1)と実験課題3)ではほぼ等しく,実験課題2)ではこれらよりも有意に高くなることが示された.また,マイクロサッカードの振幅は,実験課題2)と実験課題3)でほぼ等しく,これらに比較して実験課題1)では有意に小さくなることが示された.すなわち,視野内の特定の位置に注意を集中することによりマイクロサッカードの発生が抑制され,このとき注視位置に注視を集中させることによりマイクロサッカードによる固視のブレも押さえられることが明らかとなった.さらに,一般に主系列と呼ばれているマイクロサッカードの速度・振幅特性を比較した結果,実験課題1)におけるマイクロサッカードのピーク速度は,他の2条件に比較して有意に大きいことが示された.一方,ドリフトと呼ばれる注視点が緩やかに揺れる眼球運動に対しては,FFTによる周波数解析では条件間の差を見いだすことができなかった.したがって,視覚的注意は,サッカードシステムにのみ選択的に影響を与えていることが示唆される. これらの結果は,注視システムがマイクロサッカードのトリガーをコントロールしていることや,マイクロサッカードの大きさや速度といったダイナミクスを修飾していることを示しており,また,ドリフト成分のように緩やかな固視の揺れには直接的に関与していないことを示唆している.これらの結果より,注視時に生じるマイクロサッカードを解析することにより,注意の状態をある程度予測することが可能であると思われる.