[<< | Prev | Index | Next | >>]

Saturday, January 03, 2009

Observationalism

[A compact summary of some random aspects of my personal core philosophy/epistemology. I wrote this 10 years ago, but it still pretty well applies.]

• Epistemology: How do we know what we know?

  Whether the universe is objective or subjective, whether there is a physical reality, whether consciousness is a palpable process or an etherial one -- these are not questions to be debated at the root of philosophy, as they are rightfully conclusions drawn only later from observation.

  As a starting point, nothing is self-evident but our own existence (conscious existence, not physical), and our perceptions (what we sense, regardless of what caused the sensations). In effect, we, as conscious entities, exist within a universe defined entirely by our perceptions, and the full of our experience on which to base all knowledge can be expressed by our trajectory through this perceptual universe over time. That is, what we begin with is only and exactly the sum of our perceptions, with no distinction made between introspective perception vs. external observation. A mood, a colored dot, a sound, a thought, a feeling, a craving, a fear, a smile seen on someone's face, a smile imagined, a glass touched and felt, a three-eyed gremlin hallucinated -- each of these is a fundamental perception, with no logical preference given to any above the others.

  Over time, we integrate the statistics of this perceptual universe, learning that certain perceptions tend to occur together. As a baby, we learn that two dots near eachother in our vision will tend to be the same color more often than not. As a child, we learn that sight and touch correlate in very predictable ways, which we eventually come to call the physical world. We learn that a ball which rolls out of sight will nonetheless continue to effect the future of our perceptions, and hence that the ball "exists" beyond the immediate scope of our awareness. As an adult, we reflect on our ever growing set of perceptions and observe that this concept formation of recurring patterns is an automated feature of our consciousness, and so we apply mathematics and logic to return and validate (and sometimes correct) these things we have taken for granted. Ultimately, we question the origin of our own consciousness and discover through observations that even we are manifestations of the physical world, and thus we, knowledge-gathering processes that we are, have gathered knowledge about ourselves.

  Note that from an adult perspective, the high-level perceptions we are faced with, from the "trivial" visual identification of a chair, to the intuitive perception of someone else's mood, are as uncontestible and axiomatic as our most primitive perceptions. This is not to say the chair or the mood neccessarily exists--merely that our perception of it does. It is an observational integration of all of these percepts over time that leads us to infer the complex relationships amongst them, and ultimately to validate our model of the objective, physical universe which we inferred automatically as children.

  In sum, our task in acquiring knowledge is to reverse-engineer the universe of our perceptions, to find a statistical model which encodes as best we can the complex and abstract interactions evident therein.

  And there it is, epistemology in a nutshell.

• Human Epistemology

  While epistemology, in a purely analytical sense, reduces to fairly simple and essentially mathematical goal, our biologically evolved implementation of it is rather more pragmatic. Speed of computation takes ultimate precidence over accuracy, and the majority of what we do amounts to accepting likely but unproven hypotheses. Given this handicap, it is necessary to well understand these mechanisms if we wish to evaluate our own epistemological objectivity.

  • The Role of Emotion

    The core of our consciousness is essentially a goal-oriented search algorithm which bases its decisions at each moment on our aggregate emotional state. That is, emotion is defined to be our internal representation of cognitive valuation.

    Our adult introspective perception of emotions, which includes a sense of their implications, is learned through observation and is largely inferred from a multitude of indirect clues as opposed to being direct perception as with the external senses. Consider, for instance, the cranky child who notices no subjective difference in himself whatsoever -- it is often only in retrospect that we can objectively evaluate our state, and from those observations slowly improve our ability to detect it from secondary clues as it is occurring.

    Most relevant is the emotional perception of belief. Fundamentally, belief is the valuation that tells our search process it is not necessary to explore a particular construct further--in effect, belief signals the leaf nodes of our cognitive search tree. Generally, the sensation of belief correlates with real-world validity, since the innate mechanisms which evoke belief are designed to predict just that, and as a consequence we come to consciously associate the feeling of belief with the notion of truth. However, a little honest observation reveals that the emotional sense of truth is at best an approximation of actual truth, better in some people, and worse in others, as well as changing with time, circumstance, and mood. This limitted correlation with reality applies to all emotions and their associated implications, which leads us to:

    Emotions are not primaries.

    That is, while the overt sensations of emotions are undeniable perceptions, the corresponding implications may or may not correlate with reality.

  • The Origins of Emotion

    The mechanisms that evoke emotions are many and complex, ranging from the highly innate, such as the aversion to walking off cliffs, to the purely learned or programmed, such as the many flavors of religious guilt. Our basic epistemelogical mode of creating predictions from recorded correlations applies equally to emotions, which gives us the ability, with sufficient experience, to "feel" the ultimate consequences of things up front. In the extreme, overt cases we call this "intuition", or a "sixth sense", but the basic mechanism is fundamental to our thinking and is used and taken for granted constantly.

    Correspondingly, our emotional responses remain malleable to our experience, which means we can tune them to our liking if we so choose. More, since our directions of thought are ultimately determined by our emotional valuations, tuning our emotional response is the only control we have over our ultimate objectivity. Note that this is not to be confused with controlling your emotions--once an emotion is evoked, it has had its influence. Rather, emotional response is tuned simply by mentally exploring the related issues with a rational mindset, which will naturally provide a bath of new (and presumably better) correlations to be automatically integrated. The most likely barriers to such exploration are certainty and fear, both of which it is prudent to learn to identify and to override.

    In sum, when your rational beliefs and emotions collide, explore the origins of both and you will eventually find and correct the descrepancies. That the two domains may never meet eye-to-eye is a myth (one that is unfortunately self-vindicating to those who choose to adopt it).

  Lastly, note that the majority of the above observations, aside from some of the mechanisms of consciousness, are readily available if one thinks to look for them. The descrepancies between emotional evaluation and reality, the impact of emotion on cognition, and the long-term malleability of emotional response can all be tested empirically with sufficient introspective effort. The mechanisms of consciousness, and in particular the totality of the connection between emotion and cognitive choice, are by their nature impossible to observe directly, so I shall address these later in much greater depth.

• Ethics and Purpose

  With knowledge being a summary of perceptual experience, and consciousness being a knowledge acquiring, emotionally steered, goal oriented search algorithm, the obvious question remains: What is the goal?

  • What is the Meaning to Life?

    There is no universal meaning to life any more than there is a meaning to weather. Both are processes, exhibitting characteristic properties emergent from specific organizations of their constituent elements. An individual life could be likenned to an individual cloud, spawned from appropriate conditions, interacting with its environment, changing over time, and eventually changing enough to lose the basic properties associated with being a life or a cloud, at which point it is gone.

    A tenable approach to satiating the desire for meaning is to approach that goal at face value: to satiate the desire. I.e., you cannot find a universal meaning to life, but you can introspect about the desire itself and find out what it really is that you are looking for. Generally, this sensation emerges from long-range application of our cognitive search, when we have the time and luxury to pop up enough levels of our goal hierarchy that we hit the top and suddenly find ourselves without a goal. Religion solved this problem by providing a goal with no completion: to be a certain way (good, faithful, altruistic, whatever) your whole life. Because that goal has no end, you are never left wonderring "and then what do I do after that?". Of course, the sufficiently skeptical merely have to ask "why?" and then you're riding the top again.

    Our human implementation of planning, or predictive search, relies heavily on a hierarchy of abstractions, such that longer range steps can be evaluated in progressively larger chunks rather than in time-consuming detail. This has the interesting side effect of creating a hierarchy of progressively more abstract goals, even when the originating goal is something quite concrete. Consider, for instance, if my goal is simply to have sufficient food to eat each day. In planning for this, I am led to construct a very elaborate hierarchy of goals, including such things as getting a job, acquiring sufficient skills, and so on. But each of these abstract goals implies a whole subtree terminating in a myriad of concrete subgoals, which means that the bulk of causation is from the top down. I.e., most of the things we do, we do to satisfy a more abstract goal, which in turn is to satisfy a still more abstract one, and so on. This creates the illusion that when a goal is satisfied we need to look up to the next level of abstraction to find out what to do next. But recall that in my example the entire hierarchy was actually spawned from the very concrete goal of having food each day. Consider how easy it would be to forget that, to get caught up in this huge tree of abstract goals, and then one day to hit the top: I have enough skill, I have a job, now what? The answer is to eat your food.

  • Value

    Like meaning, value is also a relative term--it does not exist as a universal. For something to be of value it must be of value to something. A tree, for instance, can have value to a bird, or to a human, or to the process of liquid condensation on its leaves, but it cannot just have value.

• More Epistemology

  • Subjective vs. Objective Truth

    Concepts, by their nature, relate things. Sometimes when thinking about or discussing a concept, we treat some of those things as being implicit in the context, in which case the concept can be said to be "subjective" with respect to that context. For example: "overcooked peas taste terrible" is a subjective concept because many of its parameters are unspecified: Who do they taste terrible to? When? And so on. However, any subjective concept taken together with its sufficient context becomes an objective concept, and as such concepts can be transported reliably between contexts by simply filling in the empty slots as a concept passes out of a given context, and removing or appropriately altering them as they enter a new context. For instance, if the above concept exists in my mind, I can give it to you by saying "overcooked peas taste terrible to me". And I can transcribe the concept for ten years in the future by saying "overcooked peas tasted terrible to me ten years ago". And so on. This leads to an important observation:

    Rational people need not agree to disagree. If you like peas, and I do not, this is not a disagreement (unless you are claiming that I like peas). Subjective tastes taken with their contexts are objective truths.

    There is the question of whether the reality in which you exist must be the same as the one in which I exist, which is once again determinable with sufficient observation. For example, if I throw a ball up and then catch it, but "in your universe" the ball never comes down, then I am left holding a ball before your eyes which you claim is not there. Observation shows, however, that if I repeatedly throw the same ball at your forehead, you will eventually complain, which demonstrates that in the end you were merely mentally incompetent but alas subject to the same reality as me.

    All conflicts can be reduced in this manner to perceptual primatives on which both parties agree, at which point the conflict is necessarily resolved assuming a concensus on the process of logic and inference itself. This latter point sounds sticky, if one reduces it to such issues as the identity and logic of quantum particles. However, such reduction is generally unnecessary, as the mutually acknowledged perceptual primatives are often of discrete and countable in nature, for which the basic rules of combination and logic are rarely contested. E.g., we need not come to a concensus on the nature of quantum physics in order to agree empirically that if we take one ball and then another ball, we will have two balls. The vast majority of knowledge can be constructed from comparably uncontested basics.

  • Certainty

    While knowledge can be said to be a summary of all prior perception, the value of knowledge is in its predictive capabilities. As such, certainty in knowledge is a measure of the contrast amongst the predicted odds of the possible outcomes, and is dependant both on the quantity of related historical perception, and on the actual odds. Thus uncertainty is a common and natural state of affairs, even for a computationally perfect knowledge integrator, and in fact excessive certainty represents less knowledge then does appropriate ambiguity.

    Hence the proper state of mental confidence is to associate approximated likelihoods with each possible hypothesis, as opposed to choosing one thing as definately true vs. another. Note, however, that many hypotheses are supported by such a preponderance of observational evidence that their odds are astronomically good, in which case it is fair to label them as "certain" with the caveat understood that nothing, beyond our own conscious existence and perceptions, is truly, absolutely, 100% certain.

[<< | Prev | Index | Next | >>]