Sunday 29 September 2013

Emotional Computing

With the coming of virtual reality comes the possibility for collecting enormous amounts of data about human responses in very realistic situations. With that comes the real possibility of creating remarkably rich corpuses of data about normative human behaviour. This, together with increased efficiencies in the searching of these corpuses, might change the world.

Could a computer identify how a person feels? If it could, I think the way that feelings are structured as different 'levels of expectation' is an important prerequisite for any algorithm. Expectations may be structural at a deep level (a yearning for something is a kind of deep structural expectation), or immediate (for example, an instinctive reaction to something). Immediate expectations can mask deep level expectations. For example, the demand for instant gratification almost always masks a deep seated fear (a deep expectation of something dire). Whatever feeling might be present at any particular moment, an individual produces signs in their behaviour. If an instant of behaviour is capturable and comparable to a corpus of an individual's emergent behaviour, and if the corpus of emergent behaviour is comparable to a corpus of normative behaviour, then why might computer not be able to identify how a person feels?

Different moments of behaviour have different structural properties. Whatever corpus techniques are used, it is necessary to be able to inspect a corpus at different 'depths' - probably with different analytical 'grains'. These might be thought of as at different levels of recursion. So the prerequisites for an emotion-identifying machine would be:

  1. a normative schema of patterns of emotion presented at many levels of recursion
  2. a mechanism for analysing at different levels of recursion signs of behaviour
If this really worked, what would be the political implications? How would the world be different? 

One would expect the first people to jump on it would be the advertising executives and politicians. That's not a good start! But this kind of technology is not necessarily a one-way street. Just as a politician might be able to see how a potential voter might feel, so a potential voter might see how a politician feels. The slick manipulation by spin-doctors become exposed for its vacuous corruption of the public sphere. One hopes.

Educationally, emotional recognition like this might have an enormous impact. Most learning difficulties (at all levels) can be analysed in terms of emotional difficulties. Might it be that if emotional blockages can be spotted and dealt with, then most things become learnable by most people? Who controls this technology? Where does this leave institutions?

When we get to the level of emotions, the question we ask are questions about the our relationship to each other. Feelings as expectations are expectations about  the behaviour of those we love. Fear may be fear of a loss of control over other peoples feelings (that they might cease to love us). Excitement is confident anticipation of other people's feelings. Shifting expectations and shifting feelings go hand-in-hand with shifting attachments.

Institutions are instruments of attachment. Our current institutional crises stem from the fact that the trust placed in institutions to maintain the attachments we thought they were there for is betrayed. Jimmy Savile represents this most powerfully. After something like that, institutions are in deep trouble. They may well react to the trouble they are in by making things worse. Austerity is an example of this: the institution of government is no different from any other.

Does technology "want" to replace institutions?; do institutions (like Universities for example) fight back by 'taming' technology?  Will "emotional computing" be the moment when the technology wins? What then?

No comments: