Future Affective Technology for Autism and Emotion Communication

Author: Rosalind W. Picard
Year: 2009
Theme of the paper: The importance of improvements of affective communication for autistic population to help them communicate their emotions.
Important ultimate goal for autism: 
development of tools that help autistic people learn how to better understand and regulate their own ANS activation without having to share their internal changes with others, allowing them more control and autonomy, which is also very important in  autism.
 
Picard’s view on use and design of a wearable: 
  • design a wearable that is comfortable to wear over an extensive period of time to capture physiological data in order to create better emotion classifier or emotion recognition AI.
  • design a tool/wearable l to help someone regulate negative arousal but also be curious about that they should not have trouble controlling the tool, or feel controlled by the tool or by others when using the tool. Because this is compromise the effectiveness of the tool.
    • it should not collect physiological data without person of the user.
    • it should be taken on/off easily.
    • it should be easily turned off.
 
 
Significant feature of an autistic person: an autistic person can have a resting heart-rate (120 beats per minute or more) twice the level of non-autistic peer (60-80 b.p.m), while outwardly appearing calm and relaxed with no visible signs of seating, heavy breading, or outward stress. Furthermore, an autistic person’s electrodermal activity (EDA), a measure of sympathetic arousal, can swing very high. Given that an autistic person have a hard time communicating through speech or facial expressions, (s)he experiences increased internal frustration from failed attempt to be understood, stress, and anger to the point where (s)he  may erupt self-injury.
  • difficulty producing speech and getting it to  be mean what they want it to mean
  • they may find that their typed speech comes out more accurately reflecting their thoughts than their spoke speech
  • at the moment of mounting pain and overload, it may also be the case that a person cannot physically move in a way to operate their communication devices, much less navigate to correctly worded choices.
  • discomfort making eye concats and and looking at faces (which is  associate with increased ANS activation and hpyer-arousal of associated brain regions.
Therefore, deciphering the dynamics of autonomic nervous system (ANS) state is important for understanding and helping people on the autism spectrum.
note: self-reported rating studies are are anticipated to be inaccurate for many autistic individuals.
Arousal and valance damnations do not fully capture the space of emotion, they can be considered a second-order approximation to the emotion space.
What are the best physical measure for capturing the two dominos of arousal and valence? 
over the course of many years of research, researchers found that face is good indicator valence and EDA is a good indicator of arousal.
for the face, with the corrugated, and zygomatic activity, separation between pleasure/displeasure, liking/disliking, joy,sadness is feasible. With the EDA, as it goes up the sympathetic nervous system activation goes up as well — causing “flight or flight” response.
Can we rely on these two matrices to measure the experienced emotion? 
No, meaning emotion is not as simple as measuring a few bodily parameters. So emotions are complex; for example exhuerant which involves combination of smiles, shrieks, bodily bouncing, or arm gestures, and even treated of joy. No algorithm exist yet desvirinbg how to precisely combine many contributions channels into a full space of emotions. never the less, context plays an important role. For example a person trying to act exuberant because of their drama teacher.
 
Context is even more important than facial expression when it comes to emotion recognition: 
Ekman-Friesen’s basic faces of emotion paired with context showed that context plays a significant role in detecting the emotion.
If a disgust face is paired with holding a gun vs underwear, it is highly likely to be interpreted as anger instead of disgust.
This simple study is an indicator of the complexity of emotion recognition based on facial expressions.
Note: EDA is a measure of arousal dimension of emotion, although it can change when mental talks increases cognitive load, or when ambient heat and humidity are suddenly increased.
 
Is change of physiological measure sufficient for an emotion to come on surface? 
No, Emotion is not just physiologically changes in the body but it requires cognitive reprisal for the current situation ( Schachter and Singer 23)
 
Could emotion be recognized by computer with which you choose to share your physiological signals? 
Yes and it seems like this direction is quite promising.
Initially  the accuracy rate of a classifier built was (38% to 51%) for four different classes of  happy sad anger and fear given skin conductance, heart-rate, and respiration, with facial electromyogram. Then using patter analysis, the accuracy was jumped up to 81%. While all these studies restricted measurements to laboratories and small sets of emotions, they demonstrate that here is a significant emotion-related information that can be recognized through physiological activities.
Challenges:
  • setting a baseline is not easy as the ANS data related to emotion varies significantly from day to day in the same person.
  • experiments done in labs are not that accurate: the whole process of asking the participant to relax for 15 minutes to set the base line is not sufficient.  it was found that when measuring a person 24/7 the low stable peroiso occur naturally deurtion during portion of sleep and other times of the day, and an hour in as strange lab may be entry non-represnative of this.
Posted in Emotion Regulation.

Leave a Reply

Your email address will not be published. Required fields are marked *