Tuesday, May 31, 2016

Institutional Realities vs. "De-Biased" Realities

The limits of your language are the limits of your world. -- Wittgenstein

What is a language? A dialect with an army. -- Old joke among linguists.

Bias is an instrument of institutions to control depictions of reality.
(See Institutionalization) Bias is pervasive. We can’t escape it completely. Even our nervous system biases our perceptions at fundamental levels.
Source: https://goo.gl/images/9xnvjl
Bias is also relative. But this need not be a constant concern. We should only worry about bias when it favors or disfavors those who are making decisions for us. This is why we look to find “disinterested,” “objective,” “unprejudiced” judgments to support our decisions.

Of course, people who are unprejudiced in one situation may be highly biased in another. It is we, ourselves, who need to discriminate when this is so. That is the problem.

Every organization that persists through changes of leadership or mission becomes an “institution.” As an institution it imposes ways of perceiving and thinking on its participants. The benefits of membership, you will be told, outweigh the costs of the conformity expected of you. You may have to subordinate your own judgment to that of the traditions, practices, etc., i.e., residues of former and present leaderships. You will have to learn which “elephants in the room” are to be ignored. Institutions seldom invite the examination, much less the mention, of “de-biased realities.”

An important point: what we imagine to be our "natural" languages are highly institutionally reconstructed means of communication. As we grow out of the sole influence of family and neighborhood, especially as we go through schooling, our languages become refashioned to meet the demands of the institutions we go through. We subordinate our dialects to our institutional languages for the promise of support and protection.

Scientists, even, sometimes fall into the trap of thinking that their professional pursuits purge their working language of institutional biases, that their speech is a neutral medium of communication. This can stifle research by restricting both theory and practice within the boundaries of passing academic or corporate philosophical preferences. (See for example, The Functional Analysis of Human Behavior.)

Just as radical forms of Behaviorism and Empiricism of the last century impeded progress in psychology and economics, (see The Mathematics of Behaviorism: an informal examination), so it is that naive "natural language" notions of conceptualization and abstraction undermine progress in applicational development in AI. To examine this further, see More Related Issues

--- EGR