Memetics as an analogy and its implicit connotations

post by Rachel Shu (wearsshoes) · 2024-06-25T05:13:12.746Z · LW · GW · 0 comments

Contents

No comments

(Standalone, but mostly written to be referenced by a future post on sandboxing frameworks and ideologies.)

Humans are bad at sandboxing Fake Frameworks [LW · GW], and thinking with them tends to create side effects. That is to say, when we make analogies, we transfer more characteristics than the comparison justifies, and bundle in assumptions that are undeserved. 

Memetics (the analogy between ideas and genes) is such a framework, so I want to point out some of the assumptions that this particular analogy from ideas to genetics might unconsciously crystallize. Daniel Filan writes that meme theory "considers the setting where memes mutate, reproduce, and are under selection pressure." [AF · GW]I think this is a good characterization of memetics. Let's go through some of the implications of this definition and the strength of various assumptions made:

  1. Ideas are transmissible from mind to mind. This is required by the analogy; it's a natural implication and how we normally think of ideas anyways. The analogy isn't generally taken so far as to claim that ideas are heritable, though. If only! So much wasted time spent educating the youth to the level of their parents.
  2. Ideas are copied when transmitted. Another basic assumption of the analogy, and a reasonably sturdy one too; it would be weird if, as soon as I told you an idea, I no longer had a copy of it.
  3. Ideas are like infectious particles. This is already beyond a strict gene-to-idea analogy, but is a consequence of points 1 and 2, and is captured in the idea of a meme going 'viral'. This is also a common notion that predates the genetic analogy, as found in the older concept of an "earworm".
  4. Ideas are possessions of minds. There is no idea which exists and is not possessed by an mind, just as genes that no organisms possess do not exist. We can still talk about hypothetical genes and memes. As a corollary ideas can become extinct. This assumption seems natural, but its corollary is weaker: ideas can be easily revived, whereas genetic extinction has until recently been more or less permanent.
  5. Instances of ideas are possessions of discrete minds. In other words, minds are to ideas as cells are to genes. You have one token of an idea, and by telling it to me you transmit it to me and reproduce it in my mind. This is questionable; there is a very common phenomenon of a concept that contains lots of ideas, like a business plan or a language, being only partially understood by each of the individuals who share it, but who in total understand it completely. Actually, the common understanding of memes has already grown beyond the metaphor here. There is an existing term, memeplex, for this sort of phenomenon, that does not have a matching term "geneplex" in genetics from which to draw an analogy (the nearest term, gene complex, is a bad match).
  6. Ideas are functional. They produce behavior in minds or populations of minds as genes produce traits in cells or organisms. People seem to have a lot of dysfunctional and useless ideas, which the analogy still sustains, as many genomes contain maladaptive genes and junk genes.
  7. Ideas are under selection pressure and have a fitness relative to their environment. Useful ideas survive, maladaptive ideas eventually die out. Ideas sometimes survive because they are individually competitive (good jokes, catchy tunes), and some because they help the agents who possess them survive (how to light a fire). A place where this breaks down is when considering ideas, like how to play the piano like Mozart, which are really valuable but totally intransmissible.
  8. All ideas evolve from other ideas. In genetics, spontaneous generation is entirely absent. This doesn't hold well for ideas; ideas often come into minds that aren't transmitted there, and aren't permutations of older ideas; we call this creativity. You can strengthen the analogy by admitting memetic recombination within minds., but I think it's fairly metaphor-breaking that ideas are frequently created or recreated without transmission. Also, see below about "bad ideas".

By contrast, some common notions about ideas which fall outside the analogy:

  1. No agent possesses two instances of the same idea. I probably don't have two identical notions of apples in my mind. Maybe a sufficiently disorganized mind can have multiple identical notions of apples. A genome can, and usually does, contain multiple copies of the same gene.
  2. Having a "bad idea", as in "touching the hot stove was a bad idea". It seems like there's a whole range of mental motions which can be considered ideas in this sense, and we can assume reinforcement learning acts on them, but they are not really the sort of thing that we transmit.
  3. Ideas require comprehension. People can understand some ideas, others they struggle with. You can model this within memetics as some ideas being more infectious than others, where as other ideas you are immune to, but fit minds are better at receiving ideas than poor minds (well, also better at rejecting bad ones), so the analogy is hard to sustain.

(Not all these notions of "idea" are the same, but "meme" doesn't really cover a coherent notion of "idea" either. Are earworms and dance moves the same sort of thing as some unit of tacit knowledge about the world?)

Analogies are intentionally weak theories; or in an ontologically more proper sense, theories are unusually strong analogies. Like all analogies, some parts of the meme metaphor are useful, and others more questionable, even though frequently deployed. Some parts add new confusions. We're generally pretty flexible with metaphors, adopting them when useful, discarding them when they lose predictive value. Still, it helps to consciously trace the boundaries of the metaphors and frameworks we use, noticing where they fail to adhere well, and especially where they make it easy to overstep.

(Thanks to Sinclair Chen for reading a draft of this post.)

0 comments

Comments sorted by top scores.