23 April, 2016

Narrative Morality

Epistemic Status: Plausible Crackpottery


Man-with-a-hammer syndrome. It's horrible, horrible affliction affecting intellectuals and can subvert their critical thinking in a certain area for weeks at time.

I used to think I was immune. Then I went through a spell where I was utterly fascinated with the idea of languages and ontology, sparked by a lesswrong post which said something along the lines of 'every language implies a certain ontological framework' that isn't a literal quote, but it captured the insight which then occupied my thinking for quite bit. Ultimately, it produced Translating Alien Languages, although I think it might have produced another, if I were more productive.

My next (as far as I remember, which shouldn't inspire confidence) fascination was narratives. The cognitive dynamics it inspired are mostly a kind of taste, which is quite hard to communicate.

At best it can be summed up in my tweet: "You are a story. Your brain is the story-teller. Sometimes there are translation errors."

Like a true case of man-with-a-hammer, I applied this idea to everything under the sun. Arguments, ideology, culture, popular science articles, etc.

As luck would have it, I've moved on from them. I might be contracting a rather minor case with my new, as-yet-unnamed Big Idea(tm), but that's not here, nor there.


I have this post tagged crackpot for a reason. Some months ago, again back when I had a fascination with narratives, I had a literal shower thought, which I planned on elaborating on ever since.

You see, human culture, really, is just a hand of Cthulh, a blind market place of memes. Fads emerge, and die. Jokes, fan theories, mutualist pairs of polarizing memes, insight porn, tribal rallying flagssignals, countersignals, et cetera, are all traded and shared. Our identities emerge and are shaped by the tides of media, stable systems in an ocean of information.

My big idea is that peoples, persons, are cultural constructs, much more than they are biological constructs.

But it's probably better to start small.


There's an idea, pretty simple, and probably more or less well-known, that humans can be considered a bag of sub-agents. The particularly visceral example of this I encountered in orthonormal's sequence on dissolving the confusions around Mary's Room and qualia in general. There's an idea integral to orthonormal's explanation, of very simple sub-agents, who simply process bits of information (like 'DETECT RED OBJECT') or preform actions (like speaking), and knit together into 'people'.

This is one of those mind-expanding insights that deeply shape your thinking for a bit. Interestingly enough, orthonormal attributes the idea to Marvin Minky's Society of Mind, a book which correspondingly went from my 'I'll get to it someday' list, to my 'I need to read this immediately' list. Naturally, this means I'll get to reading it in a week or two.

The idea is particularly intriguing, especially when meshed with the memetic perspective outlined in the last section.

Jokes, then, are a species of mental sub-agent which compels it's host to re-tell its joke whenever a joke-agent is activated on the mental graph.

This can be extended to most memes (especially since, colloquially 'memes' tend to refer to a kind of internet joke). Examples: "All your x belong to us", "x fills you with determination", etc.

Eliezer Yudkowsky elaborated an intuition for logic and morality in "Created Already in Motion". An epistemic agent is one who has logical dynamic where when  it believes 'X' and 'X=>Y' and thus instinctively adds 'Y'to its belief pool. A moral agent has a active dynamic compelling it preform 'A' whenever 'A is good' is in its belief pool.

Sub-agents, particularly memetic sub-agents, are therefore a kind of dynamic. A person hosting a certain meme has a dynamic compelling to carry out certain actions whenever the context arises.

This can be spelled out explicitly.

A := ( c & ~a) →◻ a

A memetic agent A is then be the statement in deontic logic that some context c necessitates action a. This could simply a pattern matching cached thought, a social ritual, a kind of in-joke, or a moral imperative.

In other words, this captures the fact that if a social interaction enters state c, and no one has done a yet, this means you gotta do it, man. Instance: you need to remind a friend of something.

This captures this whole dynamic business, albeit inside another dynamic.


When I was in secondary school, I sometimes wondered if I was the chosen one, or the main character of reality or something. I had read too much fiction, of course, and I realize the error of that thinking, ostensibly hypothetical as it was.

But like most ideas, increasing intelligence just means an increase in the sophistication of the arguments that reach the same conclusion.

I stumbled upon yet another argument for that bucket, based on how you could justify existing. I mean, how you could justify you, in particular, being the person that you are, rather than someone else. I the trait I latched onto was my curiosity, one of my few distinguishing characteristics. The meat of the argument was that if you're curious, you'll have more thoughts, and (perhaps) therefore more observer-moments, and therefore a larger measure of the space of people, broadly construed.

The idea is amplified, in my mind, by a particular quale that there isn't much common knowledge about. Occasionally, some odd thought will be triggered in my mind that typically isn't, and a memory of the last time I experienced it will accompany it, giving me a particular sense of time passing. It feels like bits of my awareness fall silent, only to re-join the gestalt again sometime later, like a cyclic ship of Theseus. Perhaps clearer: that when this does happen, it feels as though that fragment of 'me' has been completely unaware that whole time, giving a certain jolt of awareness.

An similar anecdote was mentioned in one of David Chapman's essays, "Can We Hunt p-zombies with fMRI?"

Specifically, this:

If the intensity of experiential vividness is variable, is there any reason it couldn’t go to zero? Could you still function if it did?

I once suddenly “came to” in the middle of a computer science class. Not unusual—except that I was the one teaching it, and had not been asleep! As far as I could tell, I had been a p-zombie for the past couple minutes. There was no conscious experience during that period.

This could be highly embarrassing! Perhaps I had started drooling and groaning “BRAAAAAINS”?—but my students seemed as bored as ever. There was no sign they noticed anything odd happening. The stuff I had written on the whiteboard was no more incoherent than usual.

This is anecdotal evidence only. My belief at the time was that I had lost consciousness, and continued teaching more-or-less normally; but that may have been wrong. For example, I might just have suddenly lost the memory of having been conscious. Still, this makes me more sympathetic to the possibility of total zombies.


But I digress.

This is relevant to the thesis, because it illustrates the agent-y nature of the memes in question. A joke or ritual embedded in  your mind is like my consciousness-fragments re-surfacing. a dormant agent that wakes up to a certain stimulus, and proceeds to compel a certain behavior.

Anyone reading this is no doubt familiar with the replication behaviors of memes. But from our mental agent perspective, the replication is a bit more interesting.

I hated the third President of a company I once worked for.  He was a former Xerox executive who fired a close friend of mine (the first President).  I plotted to get him sacked, and eventually succeeded.  A year later I was writing in my journal, and caught myself using the word, “Boom!” in the same odd way he did.  I had picked it up from him.  And although I tried to stop, a couple of years after that I was still saying “Boom!” in that way.  A bit of his personality entered me, and stuck around.  Like a tune stuck in my head.  Like a software virus.

Memes are wetware viruses. The many books we're devoured, fiction consumed, friendships made and maintained, all served to create the wonderful little almagram you call 'I'.

I'm quite fond of Yudkowsky's metaphor for knowledge in "Truly a Part of You". He likens knowledge to water, and the shallow, regurgitating, password-guessing brand of knowledge is akin to having a vessels filled with it, and liable to be knowcked over and spilled. True, deep, model-building, understanding is akin to a foundation (or well), continually producing more and water, re-filling all spilled containers.

Personality, as well, is a fountain, continually producing your own brand of memes, jokes, mannerism, and general social/emotional dynamics. Those bits and pieces of yourself drift and free and float on the wind like motes from a flame, or, better yet, like leaves on a tree.

Let's think about that metaphor for a bit. You are a tree, your memetic phenotype is all the leaves. If you pluck a leaf from the tree, you can carry it about, study it, show it off maybe, but within a day or two it will decay, turn to dust and crumble. Some of the leaves are eaten by other creatures for nourishment. Sometimes acorns fall from the branches and some of them grow into other trees.

But when that tree is felled, stripping it of the leaves can never re-produce that tree. Even the acorns it shed won't grow into a clone of that tree. The leaves could drift, in and out of the bellies of creatures, cross pollinating and ending up on other trees, but even the most successful tree will never have it's leaves re-combine into a twin.


Now that I think I've bridged the inferential distances enough to further elaborate the main thesis, let's pick that thread back up.

We've established human minds are something like hosts to ichneumonid memetic agents. So far, this has been a rather sincere, perhaps even unobjectionable post. But it's tagged 'crackpot'. So where is the nonsense?

As I said earlier, I had a literal shower thought and have been itching to explain it for a hot minute. But in the process it mutated into a less objectionable (but still undeniably crackpot) form. I've mentioned that various memetic dynamics can be considered as agents, even raised the possibility that they could be rendered as observer-moments in the Nick Bostrom sense.

The next step is clear. Can they be considered sentient, be considered persons? Clearly, some of them can, since, in aggregate, some are writing this post, and some are reading it. Mayhaps they should have less weight, since when reducing agents, at some point you've got to get something that is not, itself, an agent. Humans are morally significant. Neurons are . . . less? It's natural to think destroying a random neuron is less bad than destroying a random person, but it's not irrelevantly bad. Destroying random neurons has an unwanted effect on a persons behavior, and thus bad.

Can we weight the wants and desires of mental sub-agents, to the extent they are well-defined, against normal everyday things in our moral calculus?

I don't know.

But here are some thoughts.
  • it provides a avenue for justifying (limited) freedom of speech, contractually. Giving any policy control over speech allows them to restrict the class of sub-agents, of memes, that get to exist and replicate. People generally don't support arbitrary restriction on who gets to reproduce, or who gets to exist.
  • Naturally this must be balanced against other concerns. E.g. racist sub-agents should be exorcised, and prevented from spreading.
  • The protection of knowledge from various memory holes, both motivated and inadvertent. Destroying a book, or website, or generally any artwork destroys all sub-agents which could emerge from them.
  • Produces an interesting dilemma: A philosophical terrorist has an innocent at gunpoint, and a detonator in hand. S/he wants you to choose between saving the innocent, or saving the library.
  • Provides something of an contractual imperative to increase the diversity of consciousness, reasoning contractually.

No comments:

Post a Comment