Tip:
Highlight text to annotate it
X
bjbjq Rikke: Today we're meeting Kristina Hook. Kristina Hook is a professor in Human-Machine
Interaction at Stockholm University. For many years Kristina has been researching Affective
Computing with her colleagues at her lab. Of course, she's the ideal person to ask about
Affective Computing. So Kristina, what is Affective Computing? Kristina: The original
vision for Affective Computing was that it would be interesting to try and recognize
what our users are experiencing when they're interacting with systems, their emotional
responses to the interaction. Then make a model of that and use that to modify the interaction
perhaps to create new and emotional experiences or to fix problems, like if the user's really
frustrated or something done that maybe the interaction needs to be adapted and changed.
So it really came out of artificial intelligence, a very strong idea of the possibility to understand
people and to bring in emotion as one of those things that wshould understand in the system.
But since then -- this is a while ago. It's really a book by Rosalind Picard in 1998 that
really established the term. There was work going on before that as well but that really
established the term and since then, of course, a lot of things have happened and there are
like several branches within this and perhaps modification to the original vision. Rikke:
Could you describe those different branches? Kristina: Yes. There's one branch that takes
perhaps a critical view on whether it's possible to know what somebody else's emotion is just
by looking at their face, their bodily gestures, or putting sensors on your body to recognize
your physical reactions to emotion. People are interesting and complex beings and maybe
we can't reduce what we experience to a few variables that we can measure. Rikke: Is that
your direction or your branch? Kristina: Yes. I think for me and some of the people I've
been working with, it's interesting to think about emotion more in terms of something that
you can add to an interaction but you don't necessarily model it and represent it in the
system, and you don't necessarily adapt the interaction in such a way that users become
exceedingly frustrated because the system does things based on perhaps like now I smile
though maybe I'm not that happy. Maybe I'm a little bit nervous because I'm in an interview
situation. If the system would, because I am smiling, start to interact with me in a
new way, not recognizing the complexity of the situation, then I would hate that system.
Rikke: Yes. Kristina: In my direction, people have been saying, "We should not reduce people
to a few variables and emotion is just really at the core of what we are and how we experience
the world and so reducing that to a facial expression, or gestures, or whatever, is not
only very difficult and that is also a stupid thing to do." Rikke: Could you put a name
on this direction? Do you have a name yet? Kristina: Yes. Phoebe Sengers and I, and some
other people talk about this as affective interaction, rather than affective computing.
Sort of putting more emphasis on that it's not the system, the computer, that thinks
that it knows what your emotion is, that it has this model inside that is computing your
emotion. But instead, it's an interaction where emotion is constructed and interpreted
and given meaning as in a dialogue between the user and the system, or between users
through a system. After a while, instead of working in opposition against Affective Computing,
I think we just discovered there's so many interesting applications that we can build
and we can use some of the same technologies and some of the same insights perhaps on what
emotion is, but build other kinds of systems. So now, I wouldn't say that it's working in
opposition. It's more of working in a constructive way, finding other kinds of applications.
On the one hand, we have the Affective Computing Group, the AI people who are really interested
in modeling emotion both for recognizing user's emotions and for generating interaction. Then
we have this effective interaction group where we are looking more at emotion as a constructed
process, something where we create meaning between ourselves and where probably the system
does not, definitely not know more about you than you do yourself. But then in general,
I would say it's stupid to say that this is a new thing because of course we've always
been designing with emotion in mind. Musicians, poets, writers and filmmakers have always
been working with the interaction between what they produce and how people react to
that emotionally. Of course, when you're trying to computerize an interaction with computers
that interaction happens over time when you can explicitly work with what the user does
and how the system responds and so on. That is new and different but a lot of people have
been doing beautiful IT designs for a long time. And I think people like Don Norman [sp]
picked up on that one. He talks about emotional design so I guess that's to put a bit of a
perspective on it and not to say, "Oh suddenly in 1998 we discovered emotion." Rikke: Yes.
Rikke: Could you tell us about the theoretical background for your school and the other schools
as well? Kristina: I think Affective Computing as it was envisioned by Roz Picard at MIT,
was very much built on cognitivism of a certain kind and on neurological a biologism, while
the perspective that Phoebe Sengers and I and others have had is more of constructivism,
phenomenology, looking at people as being in the world and that the tools are part of
us and part of our expression rather than, "This is me and the system is looking at me,
recognizing stuff in me and doing stuff on behalf of me. That's simplifying of course
but perhaps are the two paradigms and the backgrounds they have. Rikke: Can you name
some different products which we can see coming from the different directions? Kristina: In
Affective Computing, they worked with for example, Rosalind Picard s Group at MIT Media
Lab. They have been working with both learning applications and with autistic children and
they're learning of emotional communication. They also worked with health issues of various
kinds. For example, for autistic children, recognizing somebody else's facial expression
is difficult. They don't understand necessarily the emotional - what the emotion is, what
it is that they are seeing. So their idea is that by simplifying this and by putting
a recognition system on other people that the kid is interacting with, looking at their
faces and then simplifying the facial expression into, This is the happy person and this is
somebody who's getting angry. And so on. They can actually learn rules for, "Okay. Hang
on." This person is now getting angry. I didn't see that but now I understand that when this
person has this particular expression then that actually means anger. Now I need to watch
out. So this one is actually putting a camera on top. So I think what they did -- I don't
know what the system is looking like today but when I saw the system last, it was -- you
have a hat on and there's a camera pointing to the person you're interacting with. It's
doing in real time, analyzing the facial expression of that person, simplifying that into a symbol
like happy, angry, whatever and so that the kid can interact with this in real time. Rikke:
Are there any problems in this approach? Kristina: There are many interesting problems so I think
on the one hand it's fantastic because autistic children have a lot of problems because they
just don't understand what is going on. Why did that person become so angry now? They
don't see the little signs and signals that things are going in the wrong direction. On
the other hand, some of them feel like, Why do we always have to adapt to the normal people?
I'm like this. And so I think Roz has been doing a really good and interesting argument
about how do we deal with that? Should we always make people who have disabilities adjust
to the normal way of being? Rikke: Their interest in you. Kristina: Yes. And also of course
simplifying facial expressions like that. It's because a lot of the time we are behaving
in ways that are socially acceptable so you don't show your frustration or anger or you
try to hide it, to mask it. And so the question is, "What is really going on?" If you're masking
a frustration or anger, on top of that, you have a happy face and then there's anger underneath.
What is it that the kids should be learning from this? Rikke: So in your direction, you're
trying to design something, which is not simplifying things as much, I guess? Yes, but can we see
that in any products now? Kristina: In our group we built several systems where we tried
to work more with emotion as an expression. It's not something for you to recognize but
I'm expressing something and I'm explicitly trying to communicate something to you. We've
worked with bodily expressions for example, because emotion is really interesting. It's
like sitting in between our social interaction and in our bodies and in the whole interaction
back and forth. So if I want to express that I'm interested in talking to you in this interview,
you see I lean forward, I open my eyes, I smile. I actively construct for an interaction.
Kristina: And you respond to that. You smile back at me. Rikke: I don't lean forward because
then now... Kristina: It would be in front of the camera. Rikke: Like this. So that's
why I am holding back. Kristina: But it's contagious isn't it? Rikke: Yes. Kristina:
It's like I do this and you feel like, Oh. I should be responding to this. Rikke: Definitely.
Kristina: Now between ourselves, we are constructing an experience so one of our systems we built
for mobile phone was that instead of sending text messages only, you have a pen that has
sensors in it so you write your text message perhaps saying, I'm so happy for you that
you have this wonderful interview today, or something. And then you do a gesture. You
can do like a happy gesture or you can do a sad, depressed, horrid, inwards, gesture.
Or you can do angry gestures or whatever. Kristina: And that changes the background
of the text message. Rikke: Automatically? Kristina: Yes. So there will be colors and
shapes and animations. So if I do the happy, there will be champagne, bubbly, nice, red
colors. If I do the angry, there will be ragged, little, angry things jumping around. And this
is what I send to you. Okay, so it's not that I'm expressing necessarily what I'm feeling
right now, or the truth of what I'm experiencing. I'm actually constructing a message for you,
I want to influence you perhaps or I want to express something of my own. So if I'm
saying, Good for you that you're doing the interview today, happy gestures. I might at
the same time be hung-over from the conference dinner yesterday and I don't want to communicate
that. So this is what we do in social communication, We construct what we want to convey. And so
with this pen you actually construct. At the same time, you get into the experience. So
if I do a happy gesture and I'm writing a happy message, and the colors are happy and
there is champagne, bubbly things, that actually creates one experience in the moment. An affective
loop if you want to perhaps, frame it like that. That makes the experience of what you're
trying to express stronger. Rikke: Yes. Kristina: So that's an example of the system that we've
got. Rikke: Personally, I don't know that product. Are there any products on the market,
which are really popular or well known? Kristina: Not that many and perhaps that's not so strange
given that the field is fairly young. But there are a lot of projects that have these
kinds of insights in there anyway, without having learned from researchers so maybe that
is fairly common thing. For example, the remote that you use when you play games, for some
of the games, the gestures are really there and they're designed to pull you into particular
experiences so I guess that's it, a good example. Rikke: Thank you so much for this first part
of the interview. Kristina: Thank you. Rikke: Let's head on for the next one. Kristina:
Okay. Rikke: If you want to get some direct guidelines about how you can apply it, you
should watch our second video with Kristina and if you want to know even more, you could
also read her chapter which you can find at www.interaction-design.org. Here you can also
find more chapters and videos like this one with other thought leaders and inventors.
Thank you so much for watching, hope you enjoyed it. Introduction to Affective Computing and
http://www.youtube.com/watch?v=JMzilFvwNXE Affective Interaction (Affective Computing)
PAGE PAGE zvlflfvlfl[lf &`#$ [Content_Types].xml Iw}, $yi} _rels/.rels theme/theme/themeManager.xml
sQ}# theme/theme/theme1.xml w toc'v )I`n 3Vq%'#q :\TZaG L+M2 e\O* $*c? )6-r IqbJ#x ,AGm T[XF64
E)`# R>QD =(K& =al- 4vfa 0%M0 theme/theme/_rels/themeManager.xml.rels 6?$Q K(M&$R(.1 [Content_Types].xmlPK _rels/.relsPK
theme/theme/themeManager.xmlPK theme/theme/theme1.xmlPK theme/theme/_rels/themeManager.xml.relsPK
Jared Normal Jared Microsoft
Office Word Title Microsoft Office Word 97-2003 Document MSWordDoc Word.Document.8