Tuesday, April 15, 2014

Experience and Relativism

Disclaimer: On a certain level, the content of this blog post is simple and obvious - but it can be harder to feel than just to think it.


Three blind people came across an elephant, but didn't know what it was. The first felt the leg, and decided it was a pillar; the second felt the tail and decided it was a rope; the third felt the tusk and decided it was a tree branch. They argued for a little while, and then the elephant let out a very elephantine trumpeting noise - and the people realized how silly their argument had been.

-- Paraphrased Parable

I won't go into the sort of conclusions people usually draw from this story, but its a useful reference point for thinking about the nature of our experience.



Experience

We can think about experience as having two parts: The first is perceptual - what we see, how we feel. The second is interpretational - how we think about our perceptions. As far as I'm concerned, there's not much to say about the first layer - I can't entertain the idea that my perceptions are anything other than what they are.

But second layer is much more interesting - the way we think about what we perceive drastically changes the quality of our experience. Just consider a few of the ways you could describe your experience of seeing this image:

  1. "There is just an assortment of colors and shapes."
  2. "I'm on the subway and there are a bunch of people across from me."
  3. "I'm in the matrix, and it's being simulated that I'm on the subway, etc."
  4. " --- ". That is, you are meditating and not interpreting your experience at all.
You can look around right now and try out interpretations like these for your current perceptions. Notice that it feels very different to think in the framework of different stories, even when your perceptions are (or at least seem to be) fixed.


Blurred lines between perception and interpretation

To me, it feels pretty easy to distinguish between perceptions - which feel like they are "out there" - and interpretations thereof - which feel like they are "in here". But I think that it's a little simplistic to assume (as I have so far) that the divide between them is so sharp. In particular, establishing a divide between perception and experience seem to presume a worldview in which there is an objective reality we perceive through our senses. But that's definitely putting the cart before the horse!

I think a more honest treatment is to toss out the notion of perception, and replace it with experience. So whereas before we broke experience into perception and interpretation thereof, we're now describing it in a self-referential way - experience is composed of itself, and our interpretation of it. 

This seems a little bit strange if for each element of my experience we imagine an infinite loop of interpreting that element, and then interpreting the interpretation and so on, because I don't feel like I'm capable of more than a few simultaneous levels of interpretation. Rather, the concept is that some part of an experience consists of interpreting other parts of the experience. For instance, right now, it feels like I'm perceiving shapes and colors in the outside world, and interpreting some of them as objects, and am also interpreting this object-interpretation as an act of interpretation. I can go up to higher levels of interpretation if I want, but then I can't also feel the lower levels at the same time. In this way, my experience is in part composed of interpretation of subsets of itself, without involving any infinite loops.

Worldviews

Let's call any general framework for generating experience-describing stories/interpretations a worldview. For example, the "English language worldview" classifies our experiences as consisting of abstract and physical objects (nouns), where objects can take actions (verbs), and where both objects and actions can have properties (adjectives, adverbs). The "physics worldview" tells the story that there are fundamental particles and forces through which they interact, and these cause us to have certain experiences through our physical brains. A common add-on to one of these worldviews would be to assign moral valences to actions or states of the world.

This post itself expresses a worldview, one we'll call the meta-worldview. Namely, it's interpreting experience by telling the story that, in general, part of our experience consists of us interpreting experience.


If we like, we can think of a worldview as a relation φ between some some set E of experiences and some model of the world (or whatever it is) M. That is 
φ matches up particular experiences with states of the model that describe those experiences. Let's think of some properties we'd like this function to satisfy:
  1. We'd like φ to be one-to-one - so that no different experiences are described in exactly the same way. 
  2. We'd usually like E to be contained in φ's domain - so that the worldview is "complete" in the sense of giving an explanation to each experience.
  3. We'd sometimes like φ to be a function - so that each experience can only be explained in one way, though maybe there other things going on in the model which we don't experience.
  4. We tend not to care if the M is contained in φ's range - there could be events in our model that we can't experience.

Relativism

Comparing Stories

One difficulty with the framework I've just laid out is that it makes it hard to choose between interpretations/stories. The most we can do to say what the different properties of stories are. For instance, the "English-language worldview" has the property that feels intuitive (to me). The "physics worldview" has the property of feeling very simple and appearing to make good predictions about data. A worldview that incorporates morality has the property that I have an easy way to interpret it as telling me how to act. 

Note that when I'm comparing worldviews, it only makes sense to describe them with properties that I can experience; I can't go inside the worldview and say things that only make sense internally. For instance, saying the words "this feels immoral" doesn't presume a worldview - rather it is just the expression of a label that I put on experiences that feel a certain way. But if I were to say "this is immoral", then I would be presuming some moral system.

Crucially, there is no experience that directly corresponds to "this worldview is true". The notion of something being true requires us to build the notion of "truth" into our model of the world. And so there is no way to call some stories true and others false, some right and others wrong - they are just different.

So how should we choose which to use? To a large extent, we don't have to choose because our minds are prejudiced towards certain ways of thinking. For example, it's pretty hard to be solipsistic, or to avoid thinking in terms of objects-with-actions-and-properties. But finer details - like whether gods or morality exist - are less constrained, at least for me.


Problems

Two things are troubling about the apparent arbitrariness of worldview choice.

The first problem occurs on a personal scale - how can I mediate between the multiple worldviews I hold? For example, I don't have a fixed worldview with respect to morality; sometimes I feel and act like an altruist, and other times, like a selfish hedonist. Since it is arbitrary which moral worldview I choose, it can be hard to get excited about imposing one interpretation over the other.


Unfortunately, I cannot avoid this conflict by just acting selfishly when I feel that way and acting altruistically when I feel that way, since my actions have long term impacts; I don't want to make decisions, which - being optimal from only one point of view - I later regret. This makes me hesitant to make long term choices in general, which is problematic. One possible resolution would be to use psychological tricks (e.g. giving myself candy when I donate to charity) to better align my hedonistic desires with altruistic outcomes. But it's not obvious that mechanisms like this exist for coalescing other not-yet-eliminated worldviews.


The second problem occurs on an interpersonal and intergroup scale - how can we mediate between agents with contradicting worldviews? It is "common wisdom" to practice tolerance towards worldviews different from one's own. But in reality, we only tolerate beliefs to which ours are sufficiently similar - when we witness something in the world which strongly clashes our worldviews (usually on the level of self-interest or of morality) we intervene. 


For example, I would intervene if I saw my neighbor torturing his children, even if we knew that - in his worldview - this was moral. I might even be compelled to violence, and still feel my actions were justified. But now consider someone whose worldview holds that non-human animals as sentient as humans (indeed, how do we know they are not?). Just as I felt justified in acts of violence against someone acting counter to my moral views, this person might feel justified in acts of violence towards meat-eaters. Given the apparent arbitrariness of worldview choice, there seems to be no way to distinguish between these two acts, although I'm much more comfortable with the former than the latter.


One solution to inter-worldview conflict would be to ensure that everyone has sufficiently similar worldviews. Indeed, we already do a lot of this sort of "brainwashing" with schools and churches and American imperialism - but it feels uncomfortable in the extreme. Another solution is to isolate ourselves into groups which only interact internally. We also already do this to an extent, by limiting our interactions with people far away from us physically, socio-economically, politically, etc. 


Neither of these solutions feel very satisfying. I hope to think about some alternatives going forward.



Summary

  1. Experience is in part composed of interpretations of subsets of itself.
  2. Different worldviews consist of different ways of mapping experiences to some model (usually a model of "what's out there").
  3. There is no non-arbitrary way to choose between worldviews, since the properties that would help choose are worldview-dependent.
  4. This leaves us with the problems of mediating between different worldviews within ourselves and between people and groups.

Thanks

To Margaret, for helping me think through some of this.
To Ben, for bugging me to lower my standards and start posting things already.

3 comments:

  1. > how can we mediate between agents with contradicting worldviews?

    As a (perhaps trivial?) observation, note that this is impossible in the general case, largely for reasons described by Eliezer Yudkowsky in "No Universally Compelling Arguments": http://lesswrong.com/lw/rn/no_universally_compelling_arguments/

    I think it's probably fruitless to look for some objectively-right-seeming way of mediating between different moral views. If you have a moral disagreement with someone, ultimately your options are either to change their view (most likely through emotional/aesthetic, not logical, appeals), engage in trade with them (i.e. strike a bargain that satisfies both of your views--"I won't eat animals if you won't beat your children"), or make them to conform to your view by use of force.

    ReplyDelete
    Replies
    1. Agreed. Definitely some worldviews of the form "It is maximally important that X" and "It is maximally important that not X" are going to come into conflict.

      I think the interesting question is more along the lines of: Given some distribution of people and their worldviews, under what minimal conditions - if any - can they coexist?

      Delete
    2. I think that liberalism argues that liberalism is the minimal conditions for different worldviews existing, though I also think there's a tendency for every moral system to argue that it's the minimum for people to "really" be able to cooperate.

      If we restrict ourselves to discourse between humans, it seems that coexistence is basically possible when nobody involved thinks that they have substantially more power than the people they disagree with -- people seem to be decently good at kind-of-not doing anything obviously horrible to each other, as long as no side can more or less unilaterally dictate and enforce upon another what they have to do.

      Delete