Bioaffect:Current events

IAT881: BioAffective Computing & Interactive Media

Spring 2011

Dr. Diane Gromala

Structure of class meetings

In each class meeting, unless otherwise noted, we will have:

topics for discussion & readings, assignments, reports, examples (artifacts) and discussion of methods.

Topics

Topics for discussion will be informed by the assigned readings.

I will assume that everyone has a thorough understanding of the readings (or quizzes may result).

We will discuss each topic by traditional means: discussion and textual responses.

For each reading, one of you will be assigned to address the topic by non-traditional, non-textual means.

This may be a video, performance, prank, interactive artwork, computer application, game, sound,

or any other legal means you come up with.

Assignments

Some assignments will simply be turned in, but most will be addressed in class.

Reports

Reports are presentations of issues assigned to you. Examples are a sensory mode,

a technology, an example (artifact, such as an application or interactive artwork)

and ideas for your project. Most should be no longer than 5 to 7 minutes, with 10-15 minutes of discussion or demos.

Basic suggestions for these reports will be provided, but you are encouraged to be creative and experimental.

For instance, perhaps the person who is assigned to report on the sense of hearing may bring in

things to listen to, things that might modify the way we hear, or design a sonic demo.

Similarly, whomever is assigned, say, the biofeedback technology of GSR might set

up a scenario using the technology, and show videoclips from the Millgram experiment.

---

Week One January 12: Introductions to affective computing, course requirements and each other.

Readings for next week:

The Emotional Self, Deborah Lupton: Introduction, Chapter 1 & Chapter 3 Media:LuptonReading1.pdf

Affective Computing, Rosalind Picard, Introduction & Chapter 1 Media:LuptonChapter1.pdf

Assignments:

1. Add information about yourself and your interests + ideas you’d like to explore to this wiki, in the Community Portal section.

2. Reports regarding “the senses” assigned.

Reporting requirements: Each of you has chosen a sensory mode. For our next class meeting,

bring a 7 page PowerPoint or KeyNote slide set. Be concise -- the words we see should only

be touchstones that help you recall your recently acquired knowledge.

Slide 1: (large) the name of the sensory mode (small) your name, IAT881: Bioaffective Computing, Spring 2011.

Slide 2: An articulation of how the sensory mode works in physiological terms.

Think about how the sensory mode you are covering might relate to other sensory modes. For instance,

touch requires proximity and presence, but vision does not.

Go beyond general assumptions.

For example, many may assume that vision just requires a set of eyes, but the brain is also crucial.

Viscera is very complex, so whomever has this one should probably focus on the enteric system, or refer to Wilson's reading, below.

We will know if whomever covers taste has done their homework if they tell us that we have taste buds

that recognize 4 (not 5) basic kinds of taste. Images, sound, animations, diagrams and videoclips are

okay, but be sure you include proper citations/attributions in small text.

Slide 3: Historical considerations. For vision, for example, some ancient Greeks believed that our eyes

emitted rays. (Note: this slide may likely have some overlap with the subsequent slide.)

Slide 4: Cultural considerations. How was or is your sensory mode considered in diverse cultures

(evil eye, eyes are the windows to our souls)? What role does your sensory mode have in one or

more cultures? (It is argued that some centuries in Western cultures were dominated by

visual or so-called scopic regimes -- what does that mean?)

Slide 5: Find an exemplary artifact regarding your sensory mode. This might be a painting,

videogame, machine, interactive artwork, musical composition, foley effect, CGI,

piece of grafitti, or so on. Choose your artifact to make a point that interests you.

Slide 6: What existing technologies relate to, simulate or address your sensory mode? Tell us

about them generally and specifically. Are there any sci-fi technologies that are interesting? If so, why?

Slide 7: Brainstorm about one project you'd love to create that directly relates to your sensory mode.

Think outside of the box. Don't be shy.

Suggestions:

a) Conduct Ypres-like research about that sensory mode. For exteroceptive senses, a good place to start is to

refer to Diane Ackerman's book A Natural History of the Senses. For interoceptive senses, a good place to

start is to refer to Drew Leder's The Absent Body. You will be confronted with a seemingly insurmountable

task -- to find and plow through an immense amount of information, and to make sense of it in a concise

but not dumbed-down manner. My advice is to maintain a hard-core, laser-like focus. Find the information

you need in the burning building, grab it, and get the heck out. One aspect of this assignment is for you to

figure out ways to quickly locate the best, most up-to-date knowledge from diverse domains, to determine

who the major players (theorists, scientists) are, what the main issues are, where exemplary works of

art, design, media or computing are, how to digest it all, and how to brainstorm ideas about that sense modality.

These are incredibly important skills.

b) Assume that you will have 30-45 seconds per slide, on average. Your report (slide presentation) will be

followed by a discussion, so use your reporting time in a concise manner. The rest of us will likely

expand upon what you have presented.

c) You can't possibly cover everything, so don't try. You DO want to make an incredibly well-informed

presentation, and you DO want to communicate the most important information. But at many points in

this process, you will have to make important decisions about WHAT is important and WHY it is important.

That's part of the assignment.

d) Often, it is easier to gain an understanding of how something works by looking at examples

of dysfunction, or so-called outliers. An example of this is the work of someone like Oliver Sacks.

e) You are strongly encouraged to enhance our understanding by helping us get our hands dirty as it were.

That means you are encouraged to bring stuff in, have us do things, or otherwise find ways to make

your information come out of the textual box. Remember that you don't have a lot of time, however.

One idea is to augment just ONE slide with something in the material world. For touch, for example,

you might consider bring in things to touch, or a technology that affects the way we experience touch.

f) Have fun. This stuff is really interesting. Seriously.

g) Remember, strategic risk-taking will be rewarded, in heaven, on earth and in class.

---

Week Two January 19: The Substrate of Sensory Experience: Intero- and Exteroception

Topics for discussion:

• Affective Computing, according to Picard

• Exteroception (Seeing, Tasting, Touching, Hearing, Smelling) and

Interoception (Proprioception, Temperature, Pain, Orientation, Kinesthetic Sense, Visceral Sense)

• Technologies related to affective computing: biofeedback, (and possibly: GPS, accelerometers)

Readings for next week:

1. (re-read Lupton)

2. CHI Paper: Humanities: Feminist HCI: Taking Stock and Outlining an Agenda for Design; doi: 10.1145/1753326.1753521

Media:Feminist_HCI.pdf‎

3. CHI Paper: Scientific: Predicting tie strength with social media; doi: 10.1145/1518701.1518736 (sent via email)

4. Barnes: "The Simulation of Smiles (SIMS) model: Embodied simulation and the meaning of facial expression" (sent via email) http://journals.cambridge.org/action/displayIssue?iid=7947586

Assignment:

Conduct an “autopsy” of the "Feminist HCI" and the "Predicting tie strength" papers.

For this first assignment, we will examine 2 papers that received best paper awards.

(For the next assignment, papers will come from more obviously different knowledge bases and fields.)

Bring the results of your autopsy in a form that can be easily shared with the rest of your colleagues in class.

The term autopsy: "to see for oneself" (Ancient Greek)

Goal: to uncover reasons why each paper was deemed to be a good paper, and to formulate reasons

for why each paper was recognized as a "best paper" by CHI decision-makers. It is assumed that few

of you are privvy to the internal politics or extenuating circumstances for best paper decision making.

This is loosely based on criteria used for judging academic papers generally, and much of it pertains to non-academic papers as well, so you will doubtless find pragmatic use for this exercise.

Directions for conducting a forensic autopsy (assume you are a SIAT-CSI):

1. Assess overall context: you know that each paper received the best paper distinction from the CHI conference.

Each of you have varying degrees of knowledge about that conference.

1.1 Articulate what you know and what you have discovered about that conference. (Given time constraints, it may be useful to ask

your colleagues and faculty members, to visit CHI's site, and to visit the URLs that lists best paper awards

for the past few years. http://jeffhuang.com/best_paper_awards.html,

http://www.chi2006.org/bestofchi.php, www.chi2007.org/pressroom/0425-best-papers.pdf

1.2 Make an initial assessment about factors that may have played roles in the decision to recognize each paper as "best paper."

2. Identification.

2.1 Who are the authors?

2.2 What factors may be associated with assessing their credibility?

2.3 What is the subject of the paper?

3. The Y-incision:

3.1 Why may this subject be relevant to the CHI community?

3.2 Is the subject innovative, original or one which may have only been rarely addressed before? Does it provide new insights? If yes, articulate your reasons; if no, articulate your reasons.

4. Structural integrity:

4.1 Is each paper technically sound? While you may not be familiar with this field, look for clues anyway.

Is the technology dated (look at the citations)?

Do you see the terminology elsewhere on the CHI site or in a CHI search? If the technology is experimental, does it seem realistic and doable?

(If the authors describe, say, a computer based on DNA soup or on jet-pack technology, or if such technologies are not visible in the citations, the answer is probably no.)

4.2 Is the article structurally sound in terms of readability, structure of argument, grammar and use of images & graphics? Taken together, these should promote comprehension, even if you are not an expert.

5. Assessing major organs:

Does each component make sense or cohere?

Are the claims backed up by appropriate citations? Every claim needs to be supported unless it is "common knowledge."

Is there a component that leaps out at you because it is exceptionally confusing, not backed up, not articulated well, too broad or too specific? Or does it just not seem to fit? Often, even good papers have problems (or "issues") -- usually one glaring issue that needs to be addressed by rewriting and resubmitting. Is there an identifiable weakness? If so, what is it?

If possible, it may be helpful to structure your answers in parallel columns, or on two parallel pages, one per article.

You may use words, phrases or bulleted lists for some of your answers, but use sentences when you need to really explain something in detail.

Remember, you are to read (and re-read) and assess the "Feminism HCI" and the "Predicting Social Tie Strength" articles, NOT the "Simulation of Smiles" article. THe more effort you put into this, the better your results will be now and later. Like a new CSI (crime scene investigator), you are expected to use specialized tools and methods -- in this case, to search for information elsewhere. Over time, this becomes a habit, but you will almost never judge a paper simply by reading it, especially when you are in interdisciplinary realms. Knowledge is constantly updated, so even the most prominent specialists habitually make sure that they keep current with knowledge in their field (in some fields like medicine, one needs to regularly accrue continuing education credits and certification), scours journals and databases and generally seeks knowledge for pain and pleasure.

GRADING: I will take your background and current knowledge into account, but I expect everyone to demonstrate that they have made serious efforts. Thus, answers such as "this is crap," and "I disagree with the basic premise," and "I have no idea how to even begin to figure this out" will doom you to the depths of very bad grades.

Like the grading schema you witnessed during the slide presentations, grades in the realm of an A mean that you not only do the assignment, but that you do it exceptionally well and go beyond the instructions. Some of that is sheer insight and being able to make profound connections, while some of it means you've spent hours slogging through information, and/or you demonstrate superior investigative techniques -- like finding experts for advice or sending the authors or editors a friendly email. (Note: busy people will generally not reply to emails unless you provide them with a reason to do so, and make it easy for them to do so. So asking them open-ended questions usually won't elicit a response.) Grades in the B range mean that you have done a good, solid job.

Note: As witnessed in the slide presentations, you may choose a relatively "objective" stance, or you may choose to integrate knowledge you are expert in, or you may choose a position (point of view) that may not be obvious or usual. That is encouraged -- it just has to be relevant and really good. Contrary to popular belief, you CAN follow the assignment specifications AND be creative. And stay within presentation time limits.

---

Week Three January 26: Biopsychology of Emotion

Guest lecturer: Dr. Steven Barnes, neuroscientist & artist, UBC/SFU (sjb@nervouscreation.com)

Discussions:

1. Lupton: Discussion leaders: Veronica & Jay

2. Biopsychological Theories of Emotion: Dr. Steven Barnes Reading: Simulation of Smiles: http://journals.cambridge.org/action/displayIssue?iid=7947586

Please send your autopsy assignments to Dr. Gromala via email. Due: Wed. Jan 26th, 5:30pm. Don't forget to title the email, beginning with BIOAFFECT.

(NOTE: slide presentations from Daniel & Andrea will be forthcoming.)

---

Week Four: February 2 biofeedback basics & review

Discussions: Review: Lupton & Picard: What counts as "emotion"? Mood? Affect? Temperment? + Barnes' reading

Biofeedback demos: GSR

Readings for this week:

Picard, Affective Computing, Chapter 2

NOTE: We will be reading most of Picard's book, so it would be a good idea to plan to read 2 chapters per week (say, during the reading break).

---

Week Five: February 9

Class Trip: Vancouver Aquarium, SEE THE 4D Show: www.visitvanaqua.org/news/4D

Note: the show runs every 20 minutes or so in the afternoon. www.visitvanaqua.org/visitors/show-descriptions

THE DOORS CLOSE AT 4:45pm, & the LAST SHOW TIME IS at 6:00pm -- be sure to bring your student ID.

Assignment 1: Write a one page paper about the so-called "4D" aspect of the show, roughly in 500 words or so.

Your paper should include a concise description (1 paragraph) of the show, and should be an analytical. Bonus points: add a paragraph of what you WOULD do to create a more affective experience (define), with any technology, real or imagined.

Assignment 2 (self-monitoring prep):

a) Find 3 examples of art or art+science projects that demonstrate an area of self-monitoring (see: quantified self.com, Dunne & Raby or Scofidio & Diller for starters). Write a list and provide URLs or book, article or movie titles where possible. For each, provide a one sentence description of what was monitored, how/what it was mapped to, and what it resulted in. Add a sentence about what you thought was interesting about each one.

b) Explore (20) self-monitoring technologies. At least half MUST be outside of the box of "norm" (i.e., commonly considered self-monitoring technologies like commonly available pedometers, biofeedback). Make a list of at least 20 (you'll hit a wall at 10, but it will open up when you hcange how you think about them). Describe, in a phrase, what the technology measures so that any of your colleagues would understand it if they read it with your verbal explanation. To jog your habits of thinking, refer to MIT's Media Lab, any scientific lab that deals with animals (your list is about self-monitoring technologies, but many analyses of animals apply to humans; don't forget that we are a host and have close links to other species), and Theo Gray's book Mad Science (available at Dr. Shaw's desk).

c) Think about (10) connections between self & world, and how when you self-monitor, you MAY be consciously or not consciously affected by moment-to-moment changes in your not-self context (world). Make a list of 10 aspects of world, from phenomena to places, states, effects or zones. What aspects of your "external" world are you interested in? Things you can't see, like radio, gamma or EMF waves? Bacteria that live on your eyelashes (is that itch from something that is on your skin?). Or, why are some places so seemingly "creepy" that your fine body hairs stand up, while others are not -- is it all in your head? Electromagnetic vibes from earthquake faults? Concentrated pressure systems? Ectoplasm residue from ghosts or jellyfish? Static electricity? Subtle oddities in light and atmosphere? Petro deposits? Insects? Subsonic noise? This part of the assignment is more speculative -- a technology to measure phenomena may not exist. It might be useful to think about specific places and ask why you may respond to them, and think about technologies that may measure phenomena that are not (easily) perceptible.

BOTH ASSIGNMENTS ARE DUE IN EMAIL FORM: February 16 5:00 p.m. (via email), include BIOAFFECT FEB.16 in title of email.

Readings for next week:

Picard's Affective Computing, Part I

---

Week Six: February 16 READING BREAK

Self-monitoring project: investigate a monitoring technology and design a self-monitoring project.

Readings for next week:

Finish reading Picard's Affective Computing book.

Assignment for next week:

Develop and roughly flesh out ideas for your class project. Each of you will be required to discuss them in class next week.

The more developed the idea, the more useful feedback will be for you. (You can work with a colleague or two. Group size limit is 3 people, unlimited other organisms.)

---

---

Week Seven: Feb. 22: Catch-up

Initial project ideas articulated.

---

Week Eight: March 1:Catch-up continued

First half of class: Assignments discussed; examples of bioaffective art, self-monitoring projects & technologies.

Picard reading part 2

Second half of class: testing technologies & discussing projects as they evolve.

Readings for next week:

Farquhar's, Beyond the Body Proper, Introduction (available online at Google Books - with some pages missing: [1]) Also, PDF: Media:Lock_Farquhar_Intro.pdf: Gromala

Springgay: Fantastical BodyMedia:Fatastical_body_small.pdf Lonnie

Noe: Out of our HeadsMedia:Out_of_our_heads2.pdf D'Arcy

The Body: Orlan: D'Arcy

The Inner Touch: Of Flying Animals (Ch 21) D'Arcy and Jay

The Inner Touch: The Anaesthetic Animal (CH 24) Tyler

(note: the last 4 essays are very short but meaty)

---

Week Nine: March 8: Beyond the Normative:Disabilities, Synesthesia & (co-)adaption

First half of class: Discussions: readings, bioaffective art & technology examples continued

The Inner Touch: Of Flying Animals (Ch 21) D'Arcy and Jay

Discussions: Humans and their technologies.

Second half of class: Terry will bring in different film/video clips to discuss the rhetoric of his project

Readings for next week:

Psychosomatic: Stress & Neurobiological Kindling: Terry

---

---

Week Ten: March 16

Discussions: Psychosomatic: Stress & Neurobiological Kindling: Terry

The Body: Bodies, Disabilities: Terry

Heller-Roazen: The Inner Touch: The Anaesthetic Animal: Tyler

Picard: Part 1: Mark

Picard Part 2: (Ch.5 Affective Signals & Systems) Jay, Veronica, Lonnie, Daniel

Readings for next week:

Picard Part 2: (Ch. 6 Rec. & Expressing Affect) Jay, Veronica, Lonnie, Daniel

---

Week Eleven: March 23: Body Image/Body Schema; Habituation, Healing: Shamanism, CAM

Discussions: Picard Part 2: (Ch. 6 Rec. & Expressing Affect) Jay, Veronica, Lonnie, Daniel

Readings for next week:

Picard Part 2: (Ch. 7) Jay, Veronica, Lonnie, Daniel (Ch.8 Wearables+) Suk Kyoung

Haraway, Cyborg Manifesto: Terry

The Taboo of Subjectivity Scientific Paradigms: Intro & Chapter 1: Suk Kyoung & Lonnie

(note on this reading: identify the 4 basic approaches)

---

Week Twelve: Medical Technologies & Affect

Discussions: Haraway, Cyborg Manifesto: Terry

The Taboo of Subjectivity Scientific Paradigms: Intro & Chapter 1: Suk Kyoung & Lonnie

1. Medical Frontiers: Research re: brain-computer interfaces, drug delivery technologies, genetics, nanotechnology.

2. CAM (Complementary & Alternative Medicine)

4 CLASS PRESENTATIONS

---

Week Thirteen: April 6 Review & Archiving, Wet & Dry

Course Review

5 CLASS PRESENTATIONS

FINAL PROJECT ARCHIVAL INFORMATION TURNED IN, IN CLASS.

Be sure everything is labelled and dated. No label, no grade.

---

---

Required Readings for the course

Picard Affective Computing, http://troy.lib.sfu.ca/record=b1876254 (Full text available online) The basis for all discussion, updated research and extensions of her model.

Lupton: The Emotional self : a sociocultural exploration, http://troy.lib.sfu.ca/record=b1904667

Farquhar: Beyond the Body Proper, Introduction (available online at Google Books - with some pages missing: [2]) Also, PDF: Media:Lock_Farquhar_Intro.pdf: Gromala

Noe: Out of our Heads Media:Out_of_our_heads2.pdf D'Arcy

Massumi: Parables for the Virtual: Movement, Affect, Sensation http://troy.lib.sfu.ca/record=b2483710 (Full text available online)

Niedenthal: "The Simulation of Smiles (SIMS) model: Embodied simulation and the meaning of facial expression" http://journals.cambridge.org/action/displayIssue?iid=7947586

Springgay: Fantastical Body Media:Fatastical_body_small.pdf Lonnie

Wilson: Psychosomatic: Feminism and the Neurological Body, Chapter 2: The Brain in the Gut http://troy.lib.sfu.ca/record=b3617961 (Full text available online)

Suggested Readings for the course: Intero- and Exteroception

The Absent Body, Drew Leder http://troy.lib.sfu.ca/record=b1465856

A Natural History of the Senses, Diane Ackerman http://troy.lib.sfu.ca/record=b2461966

Suggested Readings for the course: Beyond the Norm

The Mind's Eye, Oliver Sacks http://troy.lib.sfu.ca/record=b5664168

Suggested Readings for the course: technologies

Physical Computing: Sensing and Controlling . . . , Tom Igoe and Dan O'Sullivan http://troy.lib.sfu.ca/record=b3872653 (Full text available online)

The Affect Theory Reader, Melissa Gregg & Gregory J. Seigworth, eds. http://troy.lib.sfu.ca/record=b5532632

The Taboo of Subjectivity: Toward . . . B. ALan Wallace http://troy.lib.sfu.ca/record=b2137665 (Full text available online)

"'Thought Technologies Manual"'

http://www.thoughttechnology.com/pdf/manuals/SA7560%20Rev%203.pdf

Interesting Readings?

10 Most Bizarre Scientific Papers

http://www.oddee.com/item_90683.aspx

---

---

ASSIGNED READINGS: BRIEF ANNOTATIONS

---

Everyone should write a very concise paragraph covering the reading -- somewhere between a dictionary and encyclopedia entry.

If you share a reading with another student, you should collaborate on the entry.

Goal 1: to be a resource for all students in class

Goal 2: for you to inculcate the reading

Goal 3: for you to speak in every class "as" the author(s) assigned to you. I.e., "Author X would say . . . "

---

example: (Veronika & Jay)

Lupton, The Emotional Self: Jay and Veronica

Lupton asserts that emotions have been historically been viewed as embodied sensations, so they were (and still are) are considered to be the antithesis of reason and rationality (p.3).

Lupton's main aim is to discuss how emotions contribute to our sense of self. In doing so, she covers many divergent theories of emotions.

In the first chapter, she divides knowledge about emotions into 2 major areas:

1. emotions as inherent: humans have a set of basic emotions, even though the expression of them may vary according to culture & history

• emotions are generally viewed as:

genetically inscribed, inherited rather than learned (positivist, essentialist, organismic)

linked to physiological changes, the result of stimuli

instinctive reflexes ranging from uncontrollable to moderately controllable through willpower/self-control

• evolutionary psychologists: the function of emotions for survival & continuation of the human species

• inherent perspectives connect to this idea of emotions being "animalistic" tendencies, implying that emotions should be suppressed for a progressive society.

• Cognitive theorists: the interest is on the interrelationship among bodily response, context & individual recognition/awareness; emotions are viewed as coping mechanisms

• William James: physical responses precede the emotion, interpreted based on judgement of a situation

The citation of Mesquita and Frijda suggests theories which connect an emotion to "preparation for action". They boil emotion down to an evaluation of events, their valence, their controllability, etc. which is a very functional approach to the concept.

 - This area ties directly to the idea of "behavioural ecology", completely lost on this author.
 - Evolutionary adaptations which start for one purpose can become appropriated - discussion is missing here!

The philosophical approach to truth about emotion is practical, but the power of cognitive models to predict is omitted completely from this article.

 - The authors should mention the practicality of cognitive science, for technology, therapy, diagnosis, etc.

2. emotions as sociocultural constructions:

Emotions are phenomena that are shaped, experienced and interpreted through social and cultural practices (p.2).

"Weak thesis" (Kemper): some emotions are universal, some acquired by socialization, & universal emotions can be altered through socializing "agents"

 - When we discussed this in class, most of us agreed that the weak sociocultural approach seemed to be most intuitive.
 - We also discussed how the "weak" vs. "strong" nomenclature is a bit biased; the author is having a hard time presenting a neutral case.

"Strong thesis": emotions are irreducibly sociocultural products, entirely learned and acculturated; purely contextual; intersubjective rather than individual experience

Harre: there is no such thing as an emotion, only multiple ways of acting & feeling in a bodily way; one "does" rather than "has" an emotion; self-assessment plays a role

Cultural anthropologists (Heelas, Lutz): strengthen the sociocultural direction by doing cross-cultural analyses & demonstrating major emotional differences in diff. groups of people

Social historians (Stearns): strengthen the sociocultural direction too, but in this case, longitudinally: demonstrating the dynamics of emotion through different eras (i.e. Victorian domination of grief)

Structuralists (such as Durkheim): examine the bonding nature of rituals "collective effervescence"; social order is not just rational, but underpinned by affective/social ties

Phenomenological Approach: a person's interpretation of bodily sensations are the emotion, NOT behavior or awareness of physiological changes

---

Picard Affective Computing: Part 1: Applications (+extend): Mark

Picard defines affective computing as: artificial intelligent systems that can detect, evaluate, and process emotions. It is a human-computer interaction that includes 'sensing and reasoning context' that is derived from emotional influences.

In the book, Picard provides examples of how we can implement affective computing in various fields such as learning, social development, entertainment, preventative medicine, and consumer products. A few examples are: embedding emotions in learning, gaming that measures performance using biofeedback, healthcare specifically helping people with autism by providing feedback using their emotional responses, and software applications such as incorporating social human emotions into digital communication. Picard also points out that no application can solely be based on affective computing. It has to be associated with other tools such as the user interface or computer vision.

Issues related to emotion recognition: -how we express emotions has a broad range and the meanings or modalities vary across cultures, gender, and age. Restrictions on accessing emotions is a concern as well for example it depends on the level of neural brain activity and biological configuration. According to Picard, being able to accurately interpret and define this level of complexity is highly unlikely.

-the other difficulty is how we interpret our emotions and the variability that exists amongst individuals. Also our emotional circumstance tends to rapidly change from one state to another and this minimal latency between the shifts will be difficult for a system to interpolate.

Picard Part 1: Potential Concerns (+ Turkle handout): Terry

Picard Part 1: Emotions are Physical and Cognitive: Daniel

Picard Part 2: Affective Signals & Systems: Veronica, Jay & Daniel

Picard is going to describe the mechanics of detecting and expressing emotions on three levels. Low-level, mid-level and high-level... this dimension pertains to the level of analysis. Whereas low level are direct sensors, mid-level are abstract combinations of sensors, and high level are concepts which power connectionist networks.

Another dimension is the emotional state of the perciever, this can affect which emotion is interpreted. Bottomup and topdown influence on perception of emotion.

Context is also important, one may have the same emotional physiology in different contexts for different reasons. Emotion is not only bottom-up (from signals to symbols) but also top down.

Bell is a physical metaphor for emotion. Curved response. Successive strikes are cumulative. Different bells respond differently (based on "temperament").

Linear, time-invariant system are easy to understand. Picard suggests that computationally modeling people would depict them as neither. Instead, parameters which change over time and person would be defined to control the basic response curve (sigmoid).

Picard models emotional response as a sigmoid. As events happen the response strengthens but it hits a ceiling. The threshold, gain and ceiling are all dependent on context, personality, mood and other factors. So the sigmoid is the proximal emotional response, while the contextual properties and dimensions of the sigmoid are controlled my mood, among other things.

Context is plausible in the proximal, but long-term moods still remain fairly nebulous. Picard suggests some constant monitor of every event and input may be able to simulate it and affect the sigmoid.

Picard acknowledges that valence and intensity is a simplisitic model. She suggests that multiple parallel sigmoids could represent a wide range of emotional dimensions which may or may not affect eachother.

The sigmoid based system represents response decay, repeated strikes, temperatment, nonlinearity, time-invariance, activation, saturdation, feedback, mood.

Picard then discusses physiological signals and how they correlate to emotion. She is defining a gap between these correlations and models for emotion, and depicts them in a way which implies that it could eventually be filled.

Picard Part 2: Recognizing and Expressing Affect: Veronica, Jay & Daniel

This chapter aims to shed light onto affect recognition (as a pattern recognition problem) and affect expression (as pattern synthesis).

Setting expectation to the capability of recognition of the system is relevant. It'd be satisfying if it can recognized as well as a person; if wearable clothing (physiology) is used the recognition should be higher.

Representation of emotions can be divided into discrete emotions (i.e. basic emotions) or continuous (i.e. arousal-valence). Both models have uses and limitations, and are concise.

Another distinction is universal and person-specific (same physiological response versus difference of emotion defined, elicit, and expressed). Affective computing is first person-specif, and second universal (examining individual data to see common patterns).

Picard highlights the tension between pure or mixed emotions, she uses the metaphors of microwave (pure states alternating) and hot tub (new state from the mix). Picard also notices that when mixing emotions inhibiting can occur (i.e.: lying might inhibit love).

Picard presents low and mid-level models for pattern expression and recognition. The "best" model depends on the task.

• Facial expressions: the most visible and easily controlled. Discrete and pure model. Underlying theory is Facial Action Coding System (FACS) by Ekman and colleagues, which relates emotions to muscular movements. A second variant in facial expression is by identifying motion energy. The relationship to temperament is still not investigated.
• Vocal intonation: how we say things; cues on what’s important and our affective state. Some aspects considered speech rate, pitch, voice quality. These can be consciously controlled but less than facial expressions. Affect editor is a program to ‘read’ sentences with affect. Recognizing is more difficult, even for people (60% of reliability), recognizing arousal better than valence. Context is very powerful.
• Combination of facial and voice seems to be promising since arousal is easier recognized with voice and valence with facial expression. But it’s challenging facial recognition with talking because of the mouth movement.
• Physiological patterns: computer monitors multiple signals such as EMG, GSR, BVP, respiration. Deciding which features of the signals to analysis is an art, according to Picard. It is also an open question which combination of signals are the best indicators for affective change.
• Models for affective behaviour: expressing emotions through combining modalities (i.e.: voice, facial, HR, context), it also aims to adaptation to new theories on mixed, blocks or granularity of constructed emotions; a higher step is on predicting what emotional state comes next.
• Models and learning: the influence of emotion in memory and performance. Backpropagation as the emotional energy being attached to an association, its roots are in Freud’s cathexis to objects.

Picard Part 2: Emotional Synthesis: Veronica, Jay & Daniel

This chapters moves up from low-level signal processing and considers the role of pattern analysis in emotion detection.

Again, as in Chapter 4, Picard suggests we can't directly observe emotions from the outside... Green would argue (part-whole)

Picard looks through a wide range of emotional theories which suggest a few basic emotions, and some theories which suggest "dimensions" of emotion (Arousal (calm vs. excited), Valence (negative vs. positive)).

Picard acknowledges different theories of emotion, but suggests that these models will be useful for representation : even if they turn out to be an oversimplification.

Patternmatching lies in the chasm between individual person-specific emotions and universal patterns.

A key question is whether emotions are mixed or pure. How is a state of multiple emotions represented? This is important as we are not likely to get pure emotions from users of affective computers.

Facial recogntion is a prototypical example of pattern analysis being used to model emotions (ekman, etc.). Another example is the synthesis and recognition of affective vocal intonation. Picard reviews work in both these domains in detail.

Roy and Pentland are studying whether vocal utterances are generally approving vs. disapproving using audio measure (mean frequency, harmonic ratios, etc.) - high accuracy was found.

Given the promise in all this work, a future direction is the combination of face/voice for more conclusive recognition. The McGurk effect shows why this is a powerful process and similar to how humans perform it.

Picard used a series of physio sensors to read emotions over time and found significant predictable patterns for emotions

Hidden Markov Model estimates probability of emotional transition based on observation... what are the probabilities of moving from joy to sadness vs. interest to sadness? Here we're goin up a level in analysis: you could better identify ambiguous physiological state by knowing previous states.

Interesting that Werbos came up with backpropogataion in connectionist networks based on an emotional idea from freud: freud saw emotions flowing back through cause-effect relationships (if waiting in line leads to a packed club which is happiness, then waiting in line gets associated with happiness too... the emotion back-propagates).

Picard Part 2: Affective Wearables (+extend): Suk Kyoung, Mark

Picard’s ‘wearables’ are computational devices worn as an article of clothing or jewelry that are affectively aware of their host and surrounding environment (in a friendly, attentive, or assistive way), are non-invasive, comfortable, and voluntary in their deployment. Affectiveness is described as the identification and reproduction of patterns in biometric data that are interpreted as indicative of human emotional involvement with stimulus and context. Picard reminds us that bad interface design “will not be easy to tolerate with a system that you cannot get up and walk away from.” Picard clarifies that research experiments in Emotion Theory to date have a certain “internal validity” (repeatable results in the lab) but that most cannot claim to have any “external validity”, that is, “validity to real life experience”. Picard refers to one study suggesting that some theorists believe that real-life observation of emotion theory is impossible (Wallbott and Scherer, 1989). It is noted that we are beginning to see the influx of mobile and ubiquitous computing wearable computing in various commercial applications (FedEx deliveries, service industries, utility companies…not to mention the smart phone). Picard suggests that because music is the ‘most socially accepted form of mood manipulation’ that a personal DJ would be one suitable application of affective technology, although the association of the validity of this idea with claims by the MUZAC corporation for effective affective response from shoppers and workers may require further substantiation. The system would monitor biometric data, user choice history and associated context, to build affective profiles of their music preferences and suggest current programming. Potential issues are addressed with acknowledgment that some manual input (key words) are likely to be required, and that contextual pattern recognition (techniques known as collaborative filtering) may be used to improve affective response accuracy. Picard proposes that it may be possible to augment human memory through a combination of technologies of constant surveillance and relational affective patterning, coupling ideas of Steve Mann’s WearCam with affective computing. Eidetic (‘photographic’) memory is linked with emotionally significant events in the theory of “Flashbulb memory” (Brown and Kulik, 1977), and Picard draws from this the idea of ‘affectively indexed’ memory banks, extending Mann’s WearCam to include “not just video imagery, but simultaneously with it, to gather biometric signals”. The suggestion is that a request to remember a moment would initiate an analysis of the “corresponding psychological patterns and context, and analyze these to learn how they can be used to predict things that an individual wants to remember”. However, “computers are currently limited in their ability to recognize image content” (p. 231) or to recognize the wearer’s affective expressions (p. 228).1 It will be necessary to develop algorithms that learn continuously, and that are capable of distinguishing between biometric data, the environment, and the patterns that are observable within and between those interacting worlds. Picard proposes that the best way to do this is by continuous measurement of day to day routines in order to develop a machine that perceives context, and can learn and reason about relationships between that context and the wearer’s situation and affective response. Picard states that the remaining obstacles to this development are “engineering problems, and conducting human studies as opposed to having to overcome any known theoretical limitations” although it seems odd to suggest that the development of a technology that remediates our relationship to memory (reality?) would not lead to emergent and perhaps even unexpected theoretical problems.

Of particular interest: The privacy and value of personal data become increasingly at issue with affective wearable recording/networking devices. Picard points out the obvious, but under acknowledged, problem of ownership of data, its manipulation, commodification and exploitation, but seems to hold (unfounded?) hope that ‘secure systems’ will be developed2. Biometric-based encryption is suggested as a ‘new intriguing possibility’, but this seems to be a deceptive slide into a cart and horse problem. Biometric data is complex and ever changing, and context is possibly infinite. A categorical system that is, by design, constrained by quantitative limits may lead to universalization and normatization, and this will become even more dangerous if the normatization becomes affective. One is reminded of the dispensing of mood altering Soma in Huxley’s Brave New World. It is interesting that Steve Mann calls his video version of a heads-up reality-overlaid display of information “mediated reality”, expressing more clearly, perhaps, aspects of his research in this field that has led to the concept of Sousveillance and its closely related practice, Inverse surveillance. Picard stresses the idea that affective systems should only be worn voluntarily, but one may wonder if future societal prejudices could place those not wearing them into marginalized conditions of anti-normalization. In an embodied affective world, the media is the machine.

1. for other research in this area, see also "The Simulation of Smiles (SIMS) model: Embodied simulation and the meaning of facial expression" [3] 2. Picard cites “A Security Architecture for Multi-Agent Matchmaking” (Foner, 1996) for reference to issues involving personal data that is given to networked agents.

---

Niedenthal: The Simulation of Smiles (Barnes' facial expression): Veronica

Application of embodied cognition to interpret emotions on facial expressions. Mimicry as embodied simulation, useful for human development(i.e. learning by observation) rooted in survival. It also occurs over long periods of time, mimicking nuances of expressions. Areas of the brain engaged and neurotransmitters involved in rewards systems have correlations on facial mimicry.

---

Springgay: Fantastical Body: D'Arcy

---

Noe: Out of Our Heads: Lonnie

---

Farquhar: Beyond the Body Proper, Introduction: Gromala

In this introduction to a collection of essays, Farquhar notes the immense increase in academic research concerning what some academics call "the materialist turn" (now) which follows "the linguistic turn" (roughly, from WWII to 1990s) in the Humanities. Thus, terms like "the lived body" are used to emphasize the dynamic aspect of "living" bodies that are inescapably interconnected to and continuously affected by physical, material and biological forces (was "nature") as well as social and cultural forces (was "nurture"). Although clearly not a scientist, Farquhar draws upon the "fascinating empirical studies" that enlarge our "vision of human capacities" (p.1).

Of particular interest: Farquar articulates how theorists have challenged the commonsense, "proper," naturalized and normative ideas of a body (as a fixed, skin-bounded, biomechanical container [science] & unit of individuality [humanities/social science]). The focus is on trying to perceive our bodies as dynamic, intersubjective and plural (context matters). To do this, it is often necessary to re-examine habits and "common sense," because what we often take for granted is really the ways particular social/cultural ideas become "naturalized." To more fully understand how social & cultural contexts affect dynamic bodies, refer to the examples of genetics (genetic code vs. genetic predisposition: habits & behavior can "switch on") and especially (historical & cultural comparisons of) medicine.

Check out important ideas of how technologies have changed our operational ideas of bodies: "cyborg" (Haraway) and "posthuman" (humans aren't at the centre of the universe; Hayles & esp. Wolfe), p.3-4.

---

Massumi: Parables...Movement, Affect, Sensation: Tyler

Like Farquhar, Massumi is not a scientist, but isn't allergic to science. Tho neither goes as far as many in SIAT might hope for, it is important to keep in mind the multiplicities of what it means to be interdisciplinary: from experts who "quote" or use ideas from another field to dual-experts (academics who regularly publish & are recognized in 2 fields, either closely aligned or divergent).

Fleshing out Definitions (re-ordered, and edited down to the highlights):

Affect: "an ability to affect and a susceptibility to be affected. An emotion or feeling is a recognized affect, an identified intensity as reinjected into stimulus-response paths, into action-reaction circuits of infolding and externalization—in short, into subject-object relations. Emotion is a contamination of empirical space by affect, which belongs to the body without an image." Of most importance here, is that for Massumi affect precedes emotion, affects are preconscious, registered changes that lead to cognitive events, such as emotion.

Virtual—"defined as that which is maximally abstract yet real, whose reality is that of potential—pure relationality, the interval of change, the in-itself of transformation." The virtual, the potential, is real, if not actual.

Event—"The time of the event does not belong per se to the body in movement-vision or even to the body without an image. They incur it. It occurs to them. As time-form it belongs to the virtual....the event is superempirical: it is the crystallization, out the far side of quasi corporeality, of already actualized spatial perspectives and emplacements into a time-form from which the passing present is excluded and which, for that very reason, is as future as it is past, looping directly from one to the other. It is the immediate proximity of before and after." The key terms here, I think, are "superempirical" and "actualized". The event occurs, it is not virtual, or a potentiality, it is happening (probably right now).

quasi-corporeality—"the superposition of the sum total of the relative perspectives in which the body has been implicated, as object or subject, plus the passages between them: in other words, as an interlocking of overlaid perspectives that nevertheless remain distinct." The potential connections which one body (or one's body) has made, for Massumi, like Deleuze & Guattari, potential is virtual, and the virtual is real, if not actual.

the body without an image—"an accumulation of relative perspectives and the passages between them, an additive space of utter receptivity retaining and combining past movements, in intensity, extracted from their actual terms. It is less a space in the empirical sense than a gap in space that is also a suspension of the normal unfolding of time." I understand this as the overlaying of affects upon an individual body.

Proprioception-"Proprioceptive memory is where the infolded limits of the body meet the mind's externalized responses and where both rejoin the quasi corporeal and the event. As infolding, the faculty of proprioception operates as a corporeal transformer of tactility into quasi corporeality...[proprioception] can be said to be the mode of perception proper to the spatiality of the body without an image because it opens exclusively onto that space and registers qualities directly and continuously as movement." Tactile sensations of a moving body in (through) space provide a moving, embodied, perspective of an affective body.

Viscerality (interoception): "Visceral sensibility immediately registers excitations gathered by the five "exteroceptive" senses even before they are fully processed by the brain....The immediacy of visceral perception is so radical that it can be said without exaggeration to precede the exteroceptive sense perception. It anticipates the translation of the sight or sound or touch perception into something recognizable associated with an identifiable object...Call that "something recognizable" a quality (or property)...It registers intensity....Viscerality is the perception of suspense. The space into which it jolts the flesh is one of an inability to act or reflect, a spasmodic passivity, so taut a receptivity that the body is paralyzed until it is jolted back into action-reaction by recognition. Call it the space of passion." Pre-cognitive affects pass through the visceral system before cognitively processed in the brain: our gut signals us to flee before we know that we are scared.

---

Haraway: Cyborg Manifesto: Terry

This was a standard reading since its publication.

---

Hayles: How We Became Postmodern, Ch.3 (Contesting for the Body of Information: the Macy Conferences on Cybernetics)

---

Wolfe:

Wolfe combines animal studies with systems theory.

---

Fraser & Greco: The Body: A Reader: Intro: Tyler

Fraser & Greco: The Body: A Reader: Bodies, Disabilities: Terry

---

Heller-Roazen: The Inner Touch: Murratia... (Ch. 1) Jay

Heller-Roazen: The Inner Touch: Awakening... (Ch 7) Suk Kyoung

Heller-Roazen: The Inner Touch: The Unnamed King (CH 14) D'Arcy & Mark

Heller-Roazen: The Inner Touch: Thorns (Ch 19) Suk Kyoung

Heller-Roazen: The Inner Touch: The Anaesthetic Animal (CH 24) Tyler

---

Wilson: Psychosomatic: Brain in the Gut: D'Arcy

---

Wallace: The Taboo of Subjectivity: Scientific Paradigms: Intro & Chapter 1: Suk Kyoung & Lonnie