So basically I'm saying that answers can come from direct experience or through the understanding of models. I'm further claiming that direct experience is beyond justification and even perhaps communication, that the best one can do is create a model that leads its listener to the edge of the experience as Joyce's epiphanies, works of scripture or music do. You cannot share the experience itself. If this is the case, then everything that we know collectively is in terms of models.
I'm afraid that I may be accused of re-hashing Platonic idealism. I want to draw the distinction as clearly as possible. I am not saying that it may be possible for a given person to gain enlightenment, come up out of the cave, or what have you, and gain direct experience which they can then attempt to communicate to their less fortunate brethren. It seems likely to me that everybody has about the same level of direct experience as anyone else, and that far from being a path to enlightenment it often constitutes a tremendous barrier to understanding. My point, however, is not that one kind of answer is superior to the other, but that they operate in very different domains. As far as using models to propel answers based on direct experience, perhaps an example would help to clarify what I mean.
Take solipsism, Descartes old dilemma which he solved by appeal to God, I think therefore I am, but what if nothing else is? What if it's all a hallucination produced by my mind? (this differs from Descartes specific case, but works better for the example). I can prove to my own satisfaction that the world is external to my mind, because I have had the direct experience of genuine surprise: of recognizing that something in the world operated in a way that was contrary to how my mind would have ordered it. I therefore have knowledge of the independence of the world, but I cannot communicate this knowledge. I can only share the mechanism that I used to reach the knowledge (as I have above). If someone were to come to me and say that they had never been surprised in their life, I would have no power to transfer my knowledge.
So again, models are all we have to collectively learn. Further, models are how we typically function as individuals (though whether this is learned or inherent is more than I know). But models can say anything, they can say absurd things about the king of France or non-Euclidean space. Models are independent of reality. Have I stepped into Relativism by accepting as our only shared grounding a medium which is wholly independent of Truth? Do I need to reject the possibility of the existence of Truth because of this?
Actually, I think that the existence of direct experience is far more problematic for Truth than model-based understanding. After all, people frequently claim direct experience of radically different things. Where direct experience is concerned, either most people are liars, Truth plays favorites (and is therefore no universal Truth at all), or both. Model-based understanding actually saves Truth from this damnation because, unlike direct experience, we can share our models and build consensus. It turns out that building consensus doesn't primarily involve majority vote (which would be the case in a Relativistic cosmos), but judging the models against other models and finding the ones that fit together best. The convergence of this process is the best evidence of Truth of which I know.
We seem to judge models in two distinct ways. Again, I am at difficulties labeling them, but I will use "internal power" and "extensive power" to distinguish the two. "Internal power" in a model allows one to determine what will happen within the confines of that model's domain. "Extensive power" is the ability of a model to migrate to other domains. For example, Evolution is a model with extraordinary extensive power, but very limited internal power as currently understood.
An interesting case for the consideration of these attributes of models is the rise of the heliocentered model of the solar system. If you were to accidentally allude to a belief that the sun goes around the earth at a party, you would likely be laughed out of the room. Yet, the geocentered model is not only valid, but has so much internal power that it is still used by all of us (every time you say "what time will the sun come up"). However, while there is no fault in assuming the earth to be fixed and determining the motion of all heavenly bodies around it, and in fact such a model can be sufficiently codified to be useful to agrarian engineers (i.e. a Farmers Almanac), the heliocentered model (or rather the gravitational model that it prefigured) has as much Internal Power and adds huge Extensive Power. It is a superior model. Not more true, just more powerful.
No end of contention has arisen questioning whether various models are true or false. I think we would be better off considering them strictly in terms of their power. Models are never "true" (if they were, they wouldn't be models), and attempting to use relative comparisons (more true, less true) presupposes an access to Truth that we simply don't have.
However, the fact that models have predictive power at all (or more to the point, the fact that some models have greater power than others) seems to me powerful evidence of the existence of an absolute Truth by which models are indirectly judged.
Friday, June 20, 2008
Thursday, June 19, 2008
Abandoning Epistemology
I used to think that I was developing an Epistemology (that is, an understanding of how knowledge is obtained and used). This was a very appealing thought to me, because I consider Epistemology and Ethics to be the most important branches of philosophy which have managed to escape the harvesting of science.
What I mean by "harvesting of science" is the way in which branches of philosophy have been individually isolated as scientific disciplines, such that a person can reasonably consider themselves a psychologist, physicist, biologist, lawyer, theologian, mathematician, etc. without concerning themselves with other erstwhile philosophical areas that are not a part of their discipline. What has been left for philosophy departments to teach after all of this harvesting is the historical progression of such areas up until their isolation, plus a few topics that have not been teased apart sufficiently to stand alone as disciplines. Of these few topics, most are marginalized by at least some contexts. For example, the pragmatist school of philosophy (by which I mean C. S. Pierce, William James, etc. not later thinkers such as Rorty who adopted the name without regard for the theories behind it) declared metaphysics to be an unsuitable topic for philosophy, while the importance of logic is a major point of debate between the analytic and continental schools.
However, it seems to me that you can't have a decent philosophic vision without an idea of what knowledge is (Epistemology) and how people can live better through its use (Ethics). Since writing on Ethics is a veritable cottage industry for philosophy professors while Epistemology is largely neglected, I thought that I was well served by taking the path less traveled in search of better understanding. I started reading the more accessible works that related to epistemology, most of which have been written by people working in artificial intelligence. This was pleasant and interesting but not particularly fruitful, so I moved into more theoretic territory (courtesy of Wittgenstein). I suppose that sharing Wittgenstein's frustrations and perplexity must count as some kind of partial success, but at this point I believe that my train of thought is so different from the considerations that writers have been taking, that I can't consider myself to be thinking of epistemology at all.
Let me show you what I mean.
Consider two ways of coming to the answer of a question, which I will label "knowing" and "understanding" (I use these terms with some misgiving, but I have no better at hand). In this parlance, "knowing" is meant to refer to an answer that is the result of direct experience of truth. If I experience a pain, I know it. I can respond to the question "Do you have pain?" with knowledge "Yes, I do". The knowledge, however, is intensely personal. It may not correspond with the experience of others, or even with an accepted view of what is real. If a person has lost their arm and yet they experience pain in their missing hand, their knowledge is at odds with what they know to be real. "Understanding", on the other hand, is the recognition and anticipation of the proper working of a model. For example, I have a model of the laws of physics (courtesy of Newton) that I use to predict things about the world. I do not "know" that Force = mass * acceleration, but I "understand" that it is so, and thus have access to answers based on this model.
So far so good. My terms may seem a little obtuse, but you can see what I'm getting at. Now here's the kicker: anything that can be explained or justified must be done so purely in terms of understanding, not of knowing. This makes sense in that language is a way of constructing models, and justification must be done in terms of a context (a model), so perhaps it is unsurprising that I can't explain _directly_ how I feel while listening to Beethoven's Fifth, or what emotion I have when looking at my child. I construct a model of language and convey it to you and you interpret it. Sometimes I may look into your eyes and "know" that we share the same feeling, but I cannot justify that knowledge either.
I don't want to say that "knowing" is never shared. Music seems to be a fantastic counter-example to such a claim. In literature, consider Joyce's "epiphany" stories (such as those collected in Dubliners). The point of these stories is to evoke a visceral emotion that occurs after the end of the text. In this they are like zen koans, they are meant to create "knowing" and go beyond the model constructed by their own language. Religious feeling in general seems to be based on "knowing", though many models have been built up an presented, like Joyce's stories, in order to lead the faithful to the brink to the inexpressible experience of knowledge.
However, in normal discourse and certainly in all scientific discourse, "understanding" is the mode for shared answers. One must understand answers within the model of mathematics, within the model of a history of experiments, and within a model of experimentation itself. This last assertion should give pause. Anyone who has done an experiment knows that it is a carefully constructed model, where various outcomes are looked for, others guarded against, some ignored. However, if the experience of running an experiment doesn't constitute an aspect of "knowing", then how can any reliance on ones senses?
It can't. It doesn't. Our senses provide us a model to understand, not input to knowledge. Hume's idea of consciousness as a pile of sensations is fundamentally flawed, because such sensations could not even be recognized without a model to understand them in. Seeing is not "knowing", merely "understanding". "Cogito Ergo Sum" is knowing. Nearly everything else is understood in terms of a model.
Now, in epistemological discussion, great importance is put on understanding such sentences as "The present king of France is bald". The weird thing is supposed to be that there _is_ no present king of France, and yet we know him to be bald. By my way of thinking, the language merely constructs a model which is like any other model. The fact that it may be considered nonsensical by some doesn't enter into its validity as a model. It is "understood", because an audience would be able to answer questions within the context of the model. It doesn't provide "knowledge", but models never do. Can this be epistemology?
Since it's a theory about how we attain knowledge, it must be. Furthermore it's hardly a revolutionary idea. Hegel made a similar distinction in his Encyclopedia between "direct" and "derived" ideas. The idea of the model being distinct from the thing is a fundamental tenet of modern science. Put these two concepts together and one arrives very close to my view.
But why spend so much effort on the logical inconsistencies of models placed into alternate contexts? It seems that much of this work is not geared to understanding how we think at all, but toward the very different question of how we can make a machine think. There are those who say we cannot claim to have accomplished the former without fulfilling the latter and those who point out that we will never be sure whether we have fulfilled the latter until we accomplish the former. Fortunately for me, I don't care about this chicken and egg argument. The creation of a thinking machine seems such a remote event that I'm tempted to consider it a modern impossibility and recognize the remote hope that I'll be pleasantly surprised. This means that I ignore the latter question, and the formers relation to it, as moot.
It also means that I'm in the uncomfortable position of abandoning the study of modern epistemology.
What I mean by "harvesting of science" is the way in which branches of philosophy have been individually isolated as scientific disciplines, such that a person can reasonably consider themselves a psychologist, physicist, biologist, lawyer, theologian, mathematician, etc. without concerning themselves with other erstwhile philosophical areas that are not a part of their discipline. What has been left for philosophy departments to teach after all of this harvesting is the historical progression of such areas up until their isolation, plus a few topics that have not been teased apart sufficiently to stand alone as disciplines. Of these few topics, most are marginalized by at least some contexts. For example, the pragmatist school of philosophy (by which I mean C. S. Pierce, William James, etc. not later thinkers such as Rorty who adopted the name without regard for the theories behind it) declared metaphysics to be an unsuitable topic for philosophy, while the importance of logic is a major point of debate between the analytic and continental schools.
However, it seems to me that you can't have a decent philosophic vision without an idea of what knowledge is (Epistemology) and how people can live better through its use (Ethics). Since writing on Ethics is a veritable cottage industry for philosophy professors while Epistemology is largely neglected, I thought that I was well served by taking the path less traveled in search of better understanding. I started reading the more accessible works that related to epistemology, most of which have been written by people working in artificial intelligence. This was pleasant and interesting but not particularly fruitful, so I moved into more theoretic territory (courtesy of Wittgenstein). I suppose that sharing Wittgenstein's frustrations and perplexity must count as some kind of partial success, but at this point I believe that my train of thought is so different from the considerations that writers have been taking, that I can't consider myself to be thinking of epistemology at all.
Let me show you what I mean.
Consider two ways of coming to the answer of a question, which I will label "knowing" and "understanding" (I use these terms with some misgiving, but I have no better at hand). In this parlance, "knowing" is meant to refer to an answer that is the result of direct experience of truth. If I experience a pain, I know it. I can respond to the question "Do you have pain?" with knowledge "Yes, I do". The knowledge, however, is intensely personal. It may not correspond with the experience of others, or even with an accepted view of what is real. If a person has lost their arm and yet they experience pain in their missing hand, their knowledge is at odds with what they know to be real. "Understanding", on the other hand, is the recognition and anticipation of the proper working of a model. For example, I have a model of the laws of physics (courtesy of Newton) that I use to predict things about the world. I do not "know" that Force = mass * acceleration, but I "understand" that it is so, and thus have access to answers based on this model.
So far so good. My terms may seem a little obtuse, but you can see what I'm getting at. Now here's the kicker: anything that can be explained or justified must be done so purely in terms of understanding, not of knowing. This makes sense in that language is a way of constructing models, and justification must be done in terms of a context (a model), so perhaps it is unsurprising that I can't explain _directly_ how I feel while listening to Beethoven's Fifth, or what emotion I have when looking at my child. I construct a model of language and convey it to you and you interpret it. Sometimes I may look into your eyes and "know" that we share the same feeling, but I cannot justify that knowledge either.
I don't want to say that "knowing" is never shared. Music seems to be a fantastic counter-example to such a claim. In literature, consider Joyce's "epiphany" stories (such as those collected in Dubliners). The point of these stories is to evoke a visceral emotion that occurs after the end of the text. In this they are like zen koans, they are meant to create "knowing" and go beyond the model constructed by their own language. Religious feeling in general seems to be based on "knowing", though many models have been built up an presented, like Joyce's stories, in order to lead the faithful to the brink to the inexpressible experience of knowledge.
However, in normal discourse and certainly in all scientific discourse, "understanding" is the mode for shared answers. One must understand answers within the model of mathematics, within the model of a history of experiments, and within a model of experimentation itself. This last assertion should give pause. Anyone who has done an experiment knows that it is a carefully constructed model, where various outcomes are looked for, others guarded against, some ignored. However, if the experience of running an experiment doesn't constitute an aspect of "knowing", then how can any reliance on ones senses?
It can't. It doesn't. Our senses provide us a model to understand, not input to knowledge. Hume's idea of consciousness as a pile of sensations is fundamentally flawed, because such sensations could not even be recognized without a model to understand them in. Seeing is not "knowing", merely "understanding". "Cogito Ergo Sum" is knowing. Nearly everything else is understood in terms of a model.
Now, in epistemological discussion, great importance is put on understanding such sentences as "The present king of France is bald". The weird thing is supposed to be that there _is_ no present king of France, and yet we know him to be bald. By my way of thinking, the language merely constructs a model which is like any other model. The fact that it may be considered nonsensical by some doesn't enter into its validity as a model. It is "understood", because an audience would be able to answer questions within the context of the model. It doesn't provide "knowledge", but models never do. Can this be epistemology?
Since it's a theory about how we attain knowledge, it must be. Furthermore it's hardly a revolutionary idea. Hegel made a similar distinction in his Encyclopedia between "direct" and "derived" ideas. The idea of the model being distinct from the thing is a fundamental tenet of modern science. Put these two concepts together and one arrives very close to my view.
But why spend so much effort on the logical inconsistencies of models placed into alternate contexts? It seems that much of this work is not geared to understanding how we think at all, but toward the very different question of how we can make a machine think. There are those who say we cannot claim to have accomplished the former without fulfilling the latter and those who point out that we will never be sure whether we have fulfilled the latter until we accomplish the former. Fortunately for me, I don't care about this chicken and egg argument. The creation of a thinking machine seems such a remote event that I'm tempted to consider it a modern impossibility and recognize the remote hope that I'll be pleasantly surprised. This means that I ignore the latter question, and the formers relation to it, as moot.
It also means that I'm in the uncomfortable position of abandoning the study of modern epistemology.
Purpose
Some time ago, I became concerned with some questions about the nature of self and other. Over the last ten years I have canvassed students and professors of philosophy, psychology, artificial intelligence and the like. Many of these discussions have been very helpful, but no one has been able to point me to a specific thread in these disciplines that would help advance my train of thought.
My intent is to write some of these concerns in blog format, in order to better order my own thinking and (eventually) solicit help from others. I have had too many conversations where misunderstanding or incomplete description determined the content of the discussion. Perhaps I will be able to marshal these ideas into a cohesive picture that will enable me to better express myself in the future.
I make no claims to originality. Most of what I'm going to say here seems to me to be generally accepted, but not well understood (least of all by myself). I am seeking to ask clear questions, I welcome direction to answers.
My intent is to write some of these concerns in blog format, in order to better order my own thinking and (eventually) solicit help from others. I have had too many conversations where misunderstanding or incomplete description determined the content of the discussion. Perhaps I will be able to marshal these ideas into a cohesive picture that will enable me to better express myself in the future.
I make no claims to originality. Most of what I'm going to say here seems to me to be generally accepted, but not well understood (least of all by myself). I am seeking to ask clear questions, I welcome direction to answers.
Subscribe to:
Posts (Atom)