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

Tuesday, January 25, 2022

Derationalizing, Part 2: Check your Blind Spot

[Addendum to last time]

I glossed over one key point that helps make more sense of the rest:

Recall my example of walking into a room and seeing (perceiving) the chair in it. How do we do that, and what does "perception" mean exactly? The answer (my opinion, but one I'm very confident in) is that perception is the act of matching some inputs (the image hitting your retinas) with your existing model of the world. And by "model" here I mean (something equivalent to) a generative model--i.e. something that could "imagine" scenes from the world.

I.e., you don't walk into a room and see a chair in a strictly bottom-up (concrete-to-abstract) way. Rather, you walk into a room expecting (being able to imagine, top-down) various things that might be there, and expecting what they might look like, and in effect you cull this set of possible things down to the ones that match the inputs you get.

Why this matters is: People are utterly incapable of perceiving anything outside of their existing model. (People can update their models with time but this is slow and.. rare.)

It's a little harder to illustrate for something like a chair (because we all have very detailed models of 3d objects in general) although it's more applicable there than one might guess, but probably the most accessible example is if you've ever tried to learn a new language with a different phonetic base than you are used to: You will initially perceive the language in terms of your existing base, and in general won't be able to properly group and differentiate some of the sounds. And by that I mean, someone native to the language might be able to say two sounds that are clearly different to them, and to you they will sound qualitatively the same. With practice those sounds begin to emerge as distinct, and eventually you will (effortlessly) hear them as two different things (as your model slowly updates to include them). Now, someone with a more elaborate low-level model of phonetics might be able to differentiate sounds from a new language immediately -- so it's not strictly true that you need experience with a particular thing to be able to perceive it; rather, it just depends on where your particular model has what level of detail. For someone who grew up in a single accent of a single language, there is no need to have a finer-grained model than the phonemes of that dialect, and so the brain generally won't waste resources on that, and they will be the most "blind" to the nuances of a novel language.

Anyway, where this becomes more relevant is with more abstract things. Consider all the nuance of life, social interaction, politics, and so on, which are "invisible" to children, and which you learn to see more and more of in better and better detail as you get older. My point here is to reflect on what you could see a few years back and compare to what you can see now; there must be some examples of things you just naturally perceive now which you remember being completely unaware of a few years back? If you can hone in on that contrast, then realize the same applies between any two people -- there are ways that the people around you are or will be like children compared to you in terms of what you can easily see that is entirely invisible to them.

I'm being long-winded on this, but it's super important, because we naturally tend to imagine that everybody sees the same world that we do. But the reality is they don't--things can be plain as day to you (not even a judgment call, just what seems to be a direct perception) and totally invisible to them, and vice versa. If you don't account for this, it's a lot harder to understand other people's perspectives. Once you do account for this, you can generally learn the limits of other people's models--typically by tribe since people's models are, as I mentioned, mostly absorbed from the presuppositions of their social/media context. And you can be on the lookout for anywhere that others seem to be able to see something you don't, and can proactively learn to see that thing.

The example I gave before was media slant. I have now witnessed multiple people transition from being blind to it to seeing it. It is the closest real-world analogy to the "red pill" Matrix moment. In general, this only happens when they have been so personally familiar with something that they were forced to see the media version vs. the real version, especially as both evolved in real time. This serves as a training example, via the correlates between the nuances of the media and the particular ways and places they are deviating from or misrepresenting reality. For some reason, some people go through this but never develop that model. That's call Gell-Mann Amnesia. But people whose brains haven't calcified, or whatever is the issue, update their models and gain a new and permanent insight.

The same situation with media exists with academia and science, unfortunately. (I've been collecting the patterns of bias and fraud for years; thinking of assembling them into a book some day.) And the same principle applies here: Most people simply can't see it, because they've not spent the time looking closely enough at it to build a model of how it works; and without that model, they simply can't see it even when it's right in front of them. Again, I've seen people transition (red pill) and almost always it is due to some direct personal experience where they knew something so well that they could not escape seeing the contrast between what was true and what was broadly and universally accepted, supported, and defended to the death by "science". (Until years later, when science finally comes around, and declares that those who thought otherwise before were right for the wrong reasons.)

So, bottom line, don't assume the person you're talking to sees the same thing even if you're both looking right at the same thing. Study their tribe and how it models the world, and you'll eventually figure out what they will and won't be able to see; and once you know that, people make a lot more sense... (And you still won't be able to teach them anything, alas.)

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

Simon Funk / simonfunk@gmail.com