Emotions in Humans and Artifacts is a difficult but truly rewarding collection of 12 essays on emotions
by leading researchers in areas as varied as cognitive science, computer
science, artificial intelligence, neuroscience, philosophy, as well as software
and game development. I say ‘difficult’ because the sophistication and the
quality of the contributions demand time and concentration. In other words,
this is not the sort of book you’re advised to read in public transport or
during a ten-minute break. Also, without some familiarity with standard
terminology, key debates, and mainstream arguments in the above disciplines it
is likely that the lay reader will find it harder to engage with the ideas
being presented and appreciate the deeper philosophical points being made.
Which is not to say that the essays are impenetrable. Far from it. Any advanced
undergraduate or postgraduate with an active interest in both minds and
computers ought to find the chapters accessible, even if at times challenging.

This said, what makes Emotions in Humans and Artifacts such a
rewarding book to read is precisely the fact that it offers a truly
multidisciplinary overview of current research on emotions. Here you can find
papers covering topics such as the evolution and brain architecture of emotions
(e.g. Ch. 2), the components of human emotions and how they may be best modeled
in a computational architecture (e.g. Ch. 3 and 4), as well as various
questions relating emotions to machines, games, virtual worlds and music
composition (e.g., Ch. 5, 7, 9, 10, 11 and 12). More unusually — and also more
interestingly, I think — the contributions also cover concerns such as whether
artificial agents ought to have emotional responses and how such responses
could be made believable, as well as whether artificial agents ought to
interpret the emotions of human users whilst displaying none of their own (e.g.,
Ch. 6 and 8). These later questions in particular tend to be ignored in the
majority of self-entitled multidisciplinary literature on emotions, mostly (I
suspect) because scientists and philosophers either think that machine-emotions
are a pipe-dream ("how can you make a machine that really feels
love, fear, etc.?"), or simply irrelevant to the study of human
emotions ("even if machines ‘have’ emotions, what’s to say they ‘feel’
anything like ours?"). Unfortunately for the rest of us, the result of such
a hard and fast dismissal of machine-emotion is that it also blinds us to a
whole range of potentially valuable insights coming from areas where ‘modeling
emotions’ is very much a practical exercise (as opposed to one of theoretical
exposition).  For notice that whether or
not we are willing to attribute ‘real’ emotions to computers, three sorts of
considerations make it worth our while to consider the role of emotional
behavior in machine intelligence. The first (following mainly from Damasio’s
work) is that emotions are a requisite for rationality. Accordingly, if we want
to build a machine that deals effectively with tasks that involve real-time
decision making in an unpredictable environment and with limited resources (and
there are many contexts in which such machines would be very useful to have),
we will have to endow it with some of the capabilities that emotions are
presumed to give humans. The second is that it may well be useful — or simply
entertaining — for humans to have computers that recognize emotions, and
possibly even (at least seem to) display some of their own. Maybe a computer
that gets ‘bored’ with the assignments you give to it doesn’t sound like an
exciting prospect, but the same can’t so easily be said for a computer that
recognizes, say, the user’s ‘frustration’ with a learning task and
‘sympathetically’ suggests a different action-goal pairing, or one that
‘enjoys’ playing with you. The third is that both the Art-world and the
Entertainment Industry are already exploring the creation of artificial agents
that display (and are meant to invoke) emotional responses (in humans). The
various ‘engineering solutions’ being tried out in these areas to make such
agents credible and life-like in our eyes are (at least) an invaluable opportunity
to test and compare the relative merits of different approaches, as well as
different working hypothesis regarding emotions and their role. By (also)
including a treatment of the above questions, Emotions in Humans and
Artifacts stands out among its peers not only for its comprehensive mapping
of current approaches to emotions, but also for its daring treatment of their
relative merits and advantages. Informative and thought provoking. 

© 2005 Isabel Gois

Isabel Gois is
a PhD student at King’s College London working on Consciousness. Her research
interests include Philosophy of Mind, Neuropsychology, and Mental Disorder. She
has articles published on Emotions, Computationalism, and Consciousness.                 

Categories: Psychology, Philosophical