Saturday, October 24, 2015

Generally Speaking

You could characterize a lot of my writing by saying that it involves observation and generalization. Journalists and academics seem more prone to describe and compare than observe and generalize, and fiction writers make things up for the entertainment of their readers, which may involve observation but usually doesn't include much explicit generalization. My writing style tends to presume that through the process I will arrive at a worthwhile conclusion or a formula that more or less sums things up. I do this unconsciously because that is how my mind works. Despite not being a scientist, I may come across as an informal scientist who is always making loose observations and proposing general theories that somehow explain whatever it is that I have chosen to focus upon. For this reason, I sometimes feel as if I am not providing the kind of writing that readers are most likely to expect. I don't write according to a journalistic model, I don't write according to an academic model, I'm not writing fiction, and even though I may seem to follow a scientific model, there is no research behind most of what I suggest. A little discussion of why I generalize is in order.

I recently read an article by Robert Sapolsky in which he says that by nature not only people but other mammals operate according to mental categories. The specific example discussed is that of gender. Like us, chimpanzees divide the world into male/female; upon coming into contact with a newborn chimpanzee, the first thing they want to know is its gender. I also just happened to notice, in the process of doing genealogical research involving census records, that gender is always one of the most significant characteristics recorded for each household member. This kind of categorization is probably a form of generalization in which different sets of behaviors are associated with males and females, and on that basis alone the future behavior of an individual is predicted. The article points out that there is in fact a range of gender characteristics built into each of our brains which does not necessarily correspond to our sexual organs, so it might be said that chimpanzees would be puzzled by contemporary LGTB thinking on gender, which makes gender identification an interior process that does not necessarily have any exterior manifestations.

It is true that categorizing and generalizing can be haphazard activities that result in distorted understanding of phenomena, but they are also default analytical methods that allow us to make basic evaluations and decisions. In the case of gender, within specific species, males may realistically be expected to behave aggressively and physically attack newcomers under certain circumstances, whereas females may be expected to avoid confrontation with the exception of defending their young. By adopting models of gender-based behavior, certain kinds of life-threatening actions may be avoided most of the time, even though the models may not be appropriate in every case and may not provide an accurate picture of reality. Sapolsky's main point is that gender stereotypes among humans are for the most part intractable, with important implications for transgender people who will find that others may not understand their choice of a gender that is at odds with their bodies. I am only using this as an example of how human minds work, but my personal position is that if I had gender issues myself I wouldn't change anything and would just deal with it. I've noticed an assortment of gender characteristics in myself and others and don't think that making an exterior change to put forth your ideal gender preference has much meaning at all. There have been feminine men and masculine women for millennia, and most of them would probably give you a blank stare if you suggested that they change their gender.

Over the last few years, much research has been devoted to making AI think as well as or better than humans. A major breakthrough occurred in 1997, when the computer Deep Blue defeated Garry Kasparov, who is considered to be one of the best chess players ever. That is seen as a "brute force" method, in which the computer evaluates hundreds of millions of moves per second, which is not at all how humans play. Humans simplify their play by using strategies and insight, which require far less processing than Deep Blue and produce a high relative return for the amount of processing. Improvements in chess programs since 1997 have resulted in newer programs that can run on mobile phones and play at the grandmaster level with less analysis and hardware than Deep Blue. Some AI researchers are looking at tier-like programming that simulates human thinking in order to achieve a more human-like cognition process in AI.

As I've suggested before, I don't think that human cognitive abilities are all they're cracked up to be if you look at how we got them, their purposes, our small brains and the totality of what we don't know. Computers that already exist are capable of processing far more information than we ever have, and their capacities are likely to increase rather than level off. In this context it seems that our brains have evolved, in effect, to get more bang for the buck by using processing shortcuts. Though you can see it in much of our cognitive activity, it is particularly clear in sciences such as physics, which aim at general rules that can be expressed mathematically and have the greatest explanatory power. At the frontiers of knowledge, physics and other fields are becoming so difficult that few people are capable of advancing them on their own. I recently read about a new mathematical proof that is so complex that hardly any mathematician in the world can read it, let alone understand it. Contrary to the image of the robust human intellect that emerged from the Enlightenment and portended almost unlimited mental potential, it is beginning to look more as if we may be reaching an upper limit to our unaided intellectual capacities.

There are two points that I want to make about this. On the one hand, a case can be made that the way we generalize is inaccurate and misleading, and that there may be other more direct ways to understand reality that are beyond our inherent intellectual capacities. On the other hand, we may not be capable of escaping this kind of shortcut method of understanding the world, because it was built into us by our evolutionary development. As a thought experiment you can see how these factors might play out today in the domain of political correctness. A purely politically correct person would theoretically make no inferences about anyone based on their outward appearance. Thus, a demure, fashionably dressed woman in a gown arriving at a gala social event in a black limousine accompanied by an elegant man in a tuxedo would not be presumed to have an inner life different from a burly, unshaven man with windblown hair, sunglasses and tattoos covering his arms stopped at a red light and revving his Harley-Davidson loudly. While it is theoretically possible that the gowned woman and the motorcyclist would have similar inner lives, the odds of that are comparatively low, and a politically correct person who ignored that fact would not only have difficulty communicating with them by treating them identically, but would also be unable to function in society as it exists. In this example, pure political correctness would have the virtue of a complete lack of bias, as if using only deductive logic and never inductive logic. The downside, however, would be significant. Without making inferences about people based on their outward appearances and behavior, their actual personalities become unknowable, hence it would be impossible to apprehend their motives. All people other than oneself would become noumenal, making it impossible to categorize and group them. The human instinct to belong to a group would be stifled if it were impossible to identify groups, and it would not be possible to lead a normal human life as we currently understand it.

My thesis, then, is that generalization is what humans do, and that although it is subject to error, we cannot avoid doing it. This means that I favor empiricism and the scientific method over the view that reality is essentially unknowable. There remains the possibility that through future genetic engineering or AI some sort of trans-human model that explains reality without relying on traditional empirical methods will emerge, but in the meantime I intend to stick to what we do best, since that's all we really know.

No comments:

Post a Comment