自閉症と感情コンピューティング

自閉症と感情コンピューティング大阪大学大学院竹村研究室 M2 藤原由来June 8, 2013

論文紹介R.W. Picard,“Future affective technology for autism andemotion communication,” Philosophical transactions of theRoyal Society of London. Series B, Biological sciences, vol. 364,no. 1535, pp. 3575–84, Dec. 2009.Future affective technology for autismand emotion communicationRosalind W. Picard*MIT Media Laboratory, Cambridge, MA 02139, USAPeople on the autism spectrum often experience states of emotional or cognitive overload that posechallenges to their interests in learning and communicating. Measurements taken from home andschool environments show that extreme overload experienced internally, measured as autonomicnervous system (ANS) activation, may not be visible externally: a person can have a resting heartrate twice the level of non-autistic peers, while outwardly appearing calm and relaxed. Thechasm between what is happening on the inside and what is seen on the outside, coupled with chal-lenges in speaking and being pushed to perform, is a recipe for a meltdown that may seem to come‘out of the blue’, but in fact may have been steadily building. Because ANS activation both inﬂu-ences and is inﬂuenced by efforts to process sensory information, interact socially, initiate motoractivity, produce meaningful speech and more, deciphering the dynamics of ANS states is importantfor understanding and helping people on the autism spectrum. This paper highlights advances intechnology that can comfortably sense and communicate ANS arousal in daily life, allowing newkinds of investigations to inform the science of autism while also providing personalized feedbackto help individuals who participate in the research.Keywords: affective computing; autism; autonomic nervous system; wearable sensors;Phil. Trans. R. Soc. B (2009) 364, 3575–3584doi:10.1098/rstb.2009.0143on March 16, 2013rstb.royalsocietypublishing.orgDownloaded from

自閉症スペクトラム障害と感情

自閉症スペクトラム障害AUTISM SPECTRUM DISORDER• 非定型なコミュニケーション• 社会性の障害• 制限や繰り返しのある行動、興味、活動

自閉症「スペクトラム」出典：自閉症スペクトラム - Wikipedia

自閉症スペクトラム障害者の見えない感情人の外部的な見た目と内部的な測定結果は異なりうる落ち着いており集中している外からの見た目内部の感情高い心拍数極度に興奮

VALENCE（情動価）と AROUSAL（覚醒度）Panagiotis C. Petrantonakis, Leontios J. Hadjileontiadis, "Emotion Recognition from Brain Signals Using Hybrid Adaptive Filtering andHigher Order Crossings Analysis," IEEETransactions on Affective Computing, vol. 1, no. 2, pp. 81-97, July-December, 2010↑ 情動価覚醒度 →「快・不快」「ポジティブ・ネガティブ」「活動的・非活動的」「興奮・冷静」

最も良い測定方法覚醒度： EDA(Electrodermal Activity;皮膚電気活動)情動価：顔• 眉間のしわ• ほお骨

波のような表情を解析するのは単純ではない

感情と自律神経系

自律神経系(Autonomic Nervous System;ANS)生理学的な覚醒を調整する交感神経副交感神経•外界の脅威に備えて、代謝を促す•‘ﬁght and ﬂight’ 戦いと躍動•心拍数・血圧が上がり、汗をかく•体のエネルギーの回復・保存•‘rest and digest’ 休息と消化

EDAで交感神経の覚醒を評価するmaking eye contact (he asked me, in a part that didn’tget aired, why I didn’t just look at him, and he was, Ioutcomes in general; in short, the ‘group analysis’can still be done. Thus, an individualized measure-ment approach, scaled to include anybody whowants to participate, can contribute to deepening theunderstanding of the science of autism.The science of individualized measurements—theidiographic approach—is not a new idea, andMolenaar and others have elsewhere argued for thebenefits of this methodology (Molenaar 2004). Theidea is to measure and characterize patterns overtime (not just minutes, but also days and weeks) foran individual, and then repeat this methodology formany individuals. The immense scientific value ofthis approach has long been known, yet it has beenimpractical. Today’s technological advances make theidiographic science newly practical. Most of all, withthe individualized data-intensive approach based onmeasurement in a person’s natural environment, it isnot just the science that benefits: each participantcan now benefit with information specific to his orher needs and situation. Participants need not wait ayear or more to see if a publication shows a finding,and wonder if their data were an outlier or a corepart of that finding. Instead, if you participate in astudy, you see your own data, and you can see ifyour patterns connect to those that the scientists findacross the group of individuals.5. DESIGN PRINCIPLES AND CHALLENGES:COMFORT, CLARITY AND CONTROLThe challenge is to design a device that can accuratelyand comfortably measure data outside the laboratoryas well as it does inside the laboratory. As shown in(a) (b)(c)Figure 1. Wearable EDA sensor. (a) Sensor is inside stretchybreathable wristband. (b) Disposable Ag/AgCl electrodesttached to the underside of the wristband. (c) The EDAensor can be worn comfortably for long periods of time.EDAセンサEDA（皮膚表面のコンダクタンス）＝交感神経の覚醒を示す便利な指標交感神経の覚醒度

交感神経感情認知注意関連づけ

EDA感情認知注意感情に関係無い要素精神的負荷による認知負荷周囲の熱や湿気

コンピュータは生理学的信号から感情を認識できるのか？

FRIDLUND & IZARD (1983)• 線形判別による生理学的特徴量からの感情の自動分類を行ってきた• 被験者に依存した、与えられた4つの顔の筋電位信号による、4つの表情（happy, sad, anger, fear）の分類• 38-51%の精度 (via cross-validation)

PICARD ET AL. (2001)• パターン認識技術と4つのmodalityを用いた• 皮膚コンダクタンス• 心拍数• 呼吸• 顔の筋電位• 被験者1人の6週間分の測定により、nertralを含めた8つの感情を81%の精度で自動分類した

LIU ET AL. (2008)• 12種類の生理学的指標に対してパターン分析• liking, anxiety, engagementの分類を、ロボットと交流した自閉症児に対して行った

学習とコミュニケーションのためのパーソナライズド感情計測

自律神経系のデータは日々明らかに変化している必要なのは、絶え間なく変わるベースラインを認識することM. Poh, N. C. Swenson, and R.W. Picard,“A wearable sensor forunobtrusive, long-term assessment of electrodermal activity.,” IEEEtransactions on bio-medical engineering, vol. 57, no. 5, pp. 1243–52,May 2010.

問題：ベースラインはいつ測定すべきか？通常の実験：実験室にて短い時間で被験者の自己申告により「リラックスしている時間」で取るしかし、この状況は作為的であり人物を特徴付けるには短すぎる

24時間・毎日測定することが重要睡眠時間などにlow stable な時間帯が自然に現れるM. Poh, N. C. Swenson, and R.W. Picard,“A wearable sensor forunobtrusive, long-term assessment of electrodermal activity.,” IEEEtransactions on bio-medical engineering, vol. 57, no. 5, pp. 1243–52,May 2010.

もう一つの大きな問題：慣れない環境での測定• 非日常的な環境• 慣れないクリニックや研究室で• 初対面の人の前でテストを受け• 不快な電極、ファスナー、ワイヤーなどを付けたりさせられる

正確な測定へのニーズは、被測定者の邪魔をせず、数週間から数ヶ月測定できるデバイスへの興味を沸かせている

ウェアラブルEDAセンサmaking eye contact (he asked me, in a part that didn’tget aired, why I didn’t just look at him, and he was, Ian individual, and themany individuals. Ththis approach has longimpractical. Today’s teidiographic science nethe individualized datmeasurement in a persnot just the sciencecan now beneﬁt withher needs and situatioyear or more to see ifand wonder if their dpart of that ﬁnding. Istudy, you see your oyour patterns connectacross the group of ind5. DESIGN PRINCIPLCOMFORT, CLARITYThe challenge is to desand comfortably measc)gure 1. Wearable EDA sensor. (a) Sensor is inside stretchyeathable wristband. (b) Disposable Ag/AgCl electrodesached to the underside of the wristband. (c) The EDAnsor can be worn comfortably for long periods of time.(a) (b)(c)3580 R. W. Picard Future affective technology for autisrstb.royalsociDownloaded from

実用化：AFFECTIVA Q SENSORhttp://www.qsensortech.com/features/

EDA測定値のストリーミング機能Bluetooth（無線）によりデータを受信可能

自閉症の成人女性であるBAGGSによる感想

人が多い部屋覚醒度 UP

座って体を揺り動かし、人々を見ないようにする覚醒度 DOWN

アイコンタクトをとられる覚醒度 UP

自分の感覚でストレスの変化に気づく前に、グラフで気づくことができた

ウェアラブルEDAセンサの利点自閉症者の社会的インタラクションが自身の生理学的状態に影響を与えることを学ぶ機会を与える自閉症者にとって助けになるように、ケアする側の対応を改善するための情報が得られる

研究は「グループ分析」から個人アプローチへ

従来の研究複数のサンプル（自閉症者）をひとまとめにし、グループとして解析していた

新しい研究方法：個人アプローチ各個人のパターンに注目することができ、あるテクニックがどのように自閉症者へ影響を与えるかを示す

EDAセンサ：デザインの原則

making eye contact (he asked me, in a part that didn’tget aired, why I didn’t just look at him, and he was, Ithink in an attempt at friendliness, leaning into me thewhole time in a way that was making me very stressycan still be done. Thus, an individualized measure-ment approach, scaled to include anybody whowants to participate, can contribute to deepening theunderstanding of the science of autism.The science of individualized measurements—theidiographic approach—is not a new idea, andMolenaar and others have elsewhere argued for thebenefits of this methodology (Molenaar 2004). Theidea is to measure and characterize patterns overtime (not just minutes, but also days and weeks) foran individual, and then repeat this methodology formany individuals. The immense scientific value ofthis approach has long been known, yet it has beenimpractical. Today’s technological advances make theidiographic science newly practical. Most of all, withthe individualized data-intensive approach based onmeasurement in a person’s natural environment, it isnot just the science that benefits: each participantcan now benefit with information specific to his orher needs and situation. Participants need not wait ayear or more to see if a publication shows a finding,and wonder if their data were an outlier or a corepart of that finding. Instead, if you participate in astudy, you see your own data, and you can see ifyour patterns connect to those that the scientists findacross the group of individuals.5. DESIGN PRINCIPLES AND CHALLENGES:COMFORT, CLARITY AND CONTROLThe challenge is to design a device that can accuratelyand comfortably measure data outside the laboratoryas well as it does inside the laboratory. As shown infigure 1, researchers at MIT have developed an EDA(c)Figure 1. Wearable EDA sensor. (a) Sensor is inside stretchybreathable wristband. (b) Disposable Ag/AgCl electrodesattached to the underside of the wristband. (c) The EDAsensor can be worn comfortably for long periods of time.EDAは情動価(VALENCE)に関与しないEDA好き嫌い本当嘘楽しい怒り快不快覚醒度情動価

ある人の覚醒度が上がったとき、どうやって快/不快を知ることができるだろうか？ある自閉症児のケースでは、指を叩くパターンのわずかな違いで判別できる一般には、人によって判別方法は違うため、それぞれのケースで見いだす必要がある

一般に、覚醒度は他の情報と組み合わされるべきである例えば、（正しく正確な）顔の表情

EDA計測はコインの表裏科学者は個人の感情を正確に測定したい多くの人々にとって、感情を測定されることは不安感情の状態を細かく知られることを望まない

プライバシー・コントロール感情を調整するツールが開発されたとする。もし壊れたり、ユーザが「自分の感情がコントロールされている」と感じたら？ツールの有効性が信用できなくなる

ウェアラブルEDAセンサの‘OPT-IN’ポリシーユーザが許可しない限り、EDAの情報は送信されないoutcomes in general; in short, the ‘group analysis’can still be done. Thus, an individualized measure-ment approach, scaled to include anybody whowants to participate, can contribute to deepening theunderstanding of the science of autism.The science of individualized measurements—theidiographic approach—is not a new idea, andMolenaar and others have elsewhere argued for thebenefits of this methodology (Molenaar 2004). Theidea is to measure and characterize patterns overtime (not just minutes, but also days and weeks) foran individual, and then repeat this methodology formany individuals. The immense scientific value ofthis approach has long been known, yet it has beenimpractical. Today’s technological advances make theidiographic science newly practical. Most of all, withthe individualized data-intensive approach based onmeasurement in a person’s natural environment, it isnot just the science that benefits: each participantcan now benefit with information specific to his orher needs and situation. Participants need not wait ayear or more to see if a publication shows a finding,and wonder if their data were an outlier or a corepart of that finding. Instead, if you participate in astudy, you see your own data, and you can see ifyour patterns connect to those that the scientists findacross the group of individuals.(a) (b)(c)Figure 1. Wearable EDA sensor. (a) Sensor is inside stretchy3580 R. W. Picard Future affective technology for autismon March 16, 2013rstb.royalsocietypublishing.orgDownloaded fromEDA ユーザ：不許可研究者データ解析者

結論

日常生活での感情計測は実生活の経験を映し出す

実環境における感情の確実な科学的分析は超高密度で大規模な個人データの塊と動的なパターン認識ツールがもたらす

新しいコミュニケーション技術は特に自閉症者にとってパーソナライズされた感情学習をもたらし、自閉症の理解を深める

まとめ• 自閉症スペクトラム障害と感情• 感情と自律神経系• 学習とコミュニケーションのためのパーソナライズド感情計測• EDAセンサ：デザインの原則• 結論

自閉症と感情コンピューティング

by Yuki Fujiwara, Graduate Student at Osaka Univ

on Jun 16, 2013

Statistics

Views

Actions

0 Embeds 0

Accessibility

Categories

Upload Details

Usage Rights

Report content

自閉症と感情コンピューティング Presentation Transcript