Can you build subjectivity into a model?
Sat, 4 Dec 1993 10:07:20 EST
I'm kind of dubious about one's ability to "model" their own biases.
I can recognize all the inherent prejudices that result from things such as
my own class position, ethnic background, national origin, cultural ethos,
educationa and training, gender, race, religion, sexual orientation, and
personal experiences and historical situation. It might even be possible
to quantify all of these prejudices into Bayesian decision-making factors.
But those things do not constitute all of my own subjectivity. It would
be tough to throw into this model such things as the activity of the
unconscious, what I had to eat for breakfast this morning, or my own
idiosyncratic aspects to which I cannot find credible systematic origins...
See, I see one of the fundamental aspects of subjectivity is that it
humans don't like having their behavior predicted... and go out of their way
to avoid behaving as someone else predicts they will. Especially if what is
predicted of them is something that is embarassing or shameful. You can put
this into your models too, but it would be tough.
There's another problem, that of infinite regress. Any model of my own
subjectivity must include my current cognitive act of examining my own
subjectivity which must include... I think you see my point. At least I