I did a good bit of reading this weekend into the idea of creating interfaces which invoke or respond to emotional cues.  One of the things I cam across was the Emotion Markup Language from the HUMAINE group, a fascinating attempt to generate and record meta emotional information within various media types.

As valid and valuable as I think these efforts might be, I’ve some strong feels related to this and the role that computers might come to play as emotional mediators.  Do Androids Dream of Electric Sheep sees its first scene as one with human beings struggling with emotional control in a post apocalyptic work, using computers to dial in and determine their emotional state for the day.  Its cinematic twin, Bladerunner, likewise sees a tension between emotion and computing, as Turing Tests are employed to track down androids, those who aren’t entirely human.  But we see by the end of the movie that things aren’t so simple.

The issue I have is the categorization and quantification of emotion in these kinds of settings.  I think this is a fantastic idea by the HUMAINE group, and I think a lot of the work they’ve been doing is very valuable.  What worries me is the inevitable automation of this kind of markup will do more to alter our emotional responses than to record them.

A brief example: an image or movie is automatically tagged with emotional notation related to anger, loss, grief, etc. through its association with its uploader, a youtuber who really likes war films.  Interpreting this markup, the site itself is modified to enhance these feelings: darkening the background, saturating the reds, cooling the colours.  Further, since naturally our homes will connected, the lights will dim to enhance the feeling, the temperature control will rise a few degrees, ambient noise will be blocked out.

The movie in question might be something as emotionally sophisticated as a Thin Red Line (I love that movie) or might just be an action romp like Commando.  By automating the process of emotional tagging, the intended nuances of that media will be altered by the computers’ interpretation of past quantified data and in-media visual cues.

There’s a lot of exciting stuff here, but there needs to be a lot more in the way of research and responsible design using these methods, I’d think.