Emotions as a sense for stories · 11:11am Dec 27th, 2021
In vision and hearing, the objects we work most directly with aren't the things our eyes and ears pick up. Our eyes pick up photons, and our ears pressure waves, yet our conscious mind are not quite built to work with photons and pressure waves. What actually matters, the end form of our senses, are the compositions. They're things like shapes, patterns, and words, and these things don't accurately retain information about their source. The objects you see reflect your sense of photons, but you can't easily determine the photons that led to them. The objects you hear reflect your sense of pressure waves, but you can't easily determine the pressure waves that led to them.
I'm toying with the idea that the objects you feel, emotionally, reflect your sense for stories. When you learn to work with edges, spaces, and gestalt, it changes the way you see. When you learn to work with scales, arpeggios, and chords, it changes the way you hear. Can you change the way you feel by learning to work with stories?
This post is on one fragment of that question: meaning. Meaning, as in "meaningful", is possibly the richest source of intuition I have. Whenever I get the sense that something is important, I find it useful to explore what that thing means to me and why it feels important. When I succeed, it's almost inevitable that I discover some hidden structure that has a large impact on how I think about things that matter to me.
For the sake of having a meaningful meaning of "meaning", I'll take that as a defining property: things that are meaningful to me are exactly the things that shape my interpretation of things that matter to me.
Let's see some of the implications.
Meaning can be systematically decomposed
I'll take a classic example. Some people find life to be meaningful, and there's a question of why that is. (Sorry for choosing something so boring and cliche, but it's the most accessible example I have.) When inclined to come up with an unambiguous, non-asinine answer, people often describe it in terms of things like dreams, love, friendship, family, and other such things.
If your dreams matter to you, how does your understanding of life shape the interpretation of your dreams? If you see life as a thing that would "carry the torch," then that would be one thing that makes life meaningful. If instead it matters to you to be able to experience many stories, and if you see life as a conduit for those experiences, then maybe you see meaning in the fact that life grows to experience so many stories.
Of course, all of this depends on what matters to you and how you're interpreting "life," but the point is that this is one seemingly fruitful (to me, at least) way to decompose questions of meaning.
Meaning can be constructed
If you want something to be meaningful to others, figure out how to use it to give shape to something that already matters to them. Examples include:
- Writing fanfiction in a way that gives additional depth to characters that already matter to people.
- Outlining a vision that changes how people think about work that they already find worthwhile.
- Planning an activity to change how people think about themselves (hopefully for the better).
Fiction can matter
This is just a bit of closure for those, like myself, that have been haunted by that one line from Skywriter:
Skywriter: Sometimes, fiction matters.
This post started with the idea that emotions are story aspects given sensory form. In other words, emotions give shape to stories. If those emotions are distinctive and meaningful...
Things that are meaningful to me are exactly the things that shape my interpretation of things that matter to me
... then the stories that give rise to those emotions matter.
If it's too difficult to figure out why some work of fiction feels like it matters, maybe it's easier to start by figuring out why the resulting emotions felt meaningful. If the two relationships— first between emotions and stories, and second between meaning and things that matter— hold, then this should give an indirect path to a satisfying answer.
A lack of meaning implies a lack of understanding
If nothing feels meaningful about a thing you believe is worthwhile, then there is nothing that shapes your interpretation of that supposedly-worthwhile thing. This almost definitionally implies a total lack of understanding about the thing you supposedly find worthwhile. While that's not an impossible situation, it would make one wonder why you would believe so strongly in that thing in the first place.
Anyway. This post was brought to you by B_25.
B_25: Please write a blog on how meaning needs to be felt, for it to be believed.
That was a great exercise, and I learned a ton from it. Keep B'ing awesome.
This was a nice thought-provoking read.
5622238
Thanks! And thanks for reading & commenting!
I didn’t comment when I originally read this due to feeling like I needed a philosophy degree to fully understand what you’re saying. Now, I realize that some of my comprehension difficulties are due to my worldview being somewhere between hedonism and existentialism (sometimes it approaches nihilism). My flippant response to people who want me to look for meaning is that many people spend all their time searching for meaning and forget to do anything meaningful.
5622292
That makes sense. Hedonism and existentialism are both about understanding experiences in terms of some unified concepts, like "pleasure" or "purpose". They give you ways to interpret experiences in a way that makes them fit into a bigger picture (sometimes with the goal of demonstrating that the picture doesn't show anything).
In a sense, this blog post is about doing the exact reverse: understanding some unified concepts, like "meaning" and "significance", in terms of experiences. Hedonism and existentialism ask "how can we measure experiences using this pleasure ruler or purpose ruler?". This blog post asks "how we can piece together instances of meaning and significance using our experiences?" It's less about looking for meaning, more about making use of it when you incidentally bump into it.
Thanks for reading and commenting!
Great blog!
5622329
Thanks, B, and thanks for sharing your insight.
5622331
I just realized that insight is sight that is inward! INward SIGHT!
5622347
That's pretty. I guess taken literally, insight would be a glimpse into your inner workings.
This is a difficult post to grapple with. I think people throw around the term "meaning" very casually, and half the time in a way that is meaningless. Decades of programming and debugging trained me to automatically, subconsciously check for type violations as I think, and this quickly destroys most "rigorous" logical philosophical arguments, because philosophers don't understand strong typing.
In this case, the variable "meaning" must be instantiated by something of type "information", and I think ultimately of type "communicated message." If someone creates a sculpture, the file that tells a 3D printer how to print it conveys all of that sculpture's information, but none of its meaning, which must therefore be something inferred from the information in that file, plus the information in the viewer's brain.
A "meaning" in a story is information conveyed from the text to the reader, by giving the reader a bundle of information which, when combined with their life experience, lets them infer or consider something else. (This communication might not be from the writer to the reader, but the reader's got to be involved.)
So I decline to ever even consider what "the meaning of life" is, because that's a type error, because life is not a communicative act. I'm not being OCD. It's hard to say precisely what any word means in any given context, but it helps a lot if you rule out the logically impossible meanings.
When you're reading for entertainment, the story is a sequence of events which, when communicated to the reader, arouses emotions in them. We can say that an entertaining story communicates emotions to the reader. The meaning of the story is just the emotions it arouses.
I am inventing this theory just now, so it might be stupid. But I'll claim that what makes a story art is doing one more level of inference. You tell the reader a sequence of events, which arouses a series of emotions, which focus the reader's brain on the story's important moving parts (see Antonio Damasio's Descarte's Error), causing the reader to recognize a puzzle (usually a moral quandary) and entertain hypotheses about how to resolve or deal with it.
5622314
That is what stories-as-art (see my comment just above) do.
5782115
This is a more precise explanation of how I think about it, though I'm not sure if it clarifies or confuses things:
I'm positing that meaning– as in meaningful rather than definition– is literally a way in which we measure the stories we use to contextualize emotional experience. For vision, there's a function that takes two arguments– a stimulus and a visual perception system– and outputs an RGB value, and that function respects the underlying structure of both the stimulus and the visual perceptive system. Similarly, there's a function that takes two arguments– a story and an "emotional perception system"– and outputs a meaning value, and that function respects the underlying structure of both its arguments. Just like RGB encodes a fragment of visual qualia, meaning encodes a fragment of emotional qualia.
To be ultra-pedantic, I'd disambiguate information from data. Information is a quantifiable object that narrows a space of possibilities. Data is information + structure, and structure makes explicit the functional properties that are implicit-or-nonexistent in a chunk of information alone. Meaning would be a subtype of data, not of information.
5782144
I suppose that when you say data is information + structure, you're thinking of data structure in a programming language, in which you can regard a 2D array of integers as quantized numbers associated with a spatial structure. But most people think of information as something you get by compressing data. Data analysis is the process of extracting information from data. If you have a spreadsheet where the rows are people, and the columns are properties of people such as income, SAT score, years of education, parents' income, and number of convictions, and you want to predict income from the other properties, then the data in the spreadsheet is the set of numbers, and the information is the beta values and other results of running a multiple regression on it. Information is always relative to a purpose or an agent. There isn't, I think, really a sense of "objective" information. The "TV screen tuned to a dead channel" of random static, which you're probably too young to have ever seen, can be viewed as a whole lot of information, or as very nearly zero information.
You could say "information" is overloaded, having 2 senses which are nearly opposite: one is that information is a raw bitstring; the other, that it is bits of information that reduce the uncertainty about a set of facts or possible futures of interest to an embodied agent. You're using the "raw bitstring" meaning, but in my experience, no one uses information theory with that meaning. The applications in which the raw bitstring meaning is useful are things like specifying how much information a computer's RAM can hold, and you don't need to understand information theory to do that. You only need information theory when you have in hand a prediction or judgement you want to make, and then the amount of information is relative to that application and the agent's pre-existing knowledge.
What does "measure" mean? It must be a function with a numeric result.
5782186
You're right, I did interpret "information" as in "information theory". With the colloquial interpretation, it does make sense to say that meaning is a subtype of information.
Re: raw bitstring. Within the information theory lens, I'd disambiguate again between message, signal, and information.
The way I'm using the word, data could be like in software, but you can attach structure to any information. A point in a space is just an element of a set (unstructured), but if you attach a neighborhood and metric tensor to it, you can call it a location and start reasoning about nearby points (structured). A data structure (software) makes the information-structure pairing explicit.
Thanks for the complement
. Unfortunately, I have seen that, and I do remember the (later) godawful satellite days where I couldn't watch television because of a storm.
5782198
I don't mean it in a numerical sense. I mean some combination of observe, derive something from, and encode. So categorical and qualitative "measurements" are within scope.
This is a more precise restatement of the part you quoted. Again, I don't know if this will clarify or confuse things.
stories contextualize emotional experiencemeans there's a class of functions{ f: story -> emotional experience }, which represents how we interpret stories in terms of emotions. Let's call that classF(capital F). In each of these functions, emotional experience is used to measure (observe, derive something from, encode) a story.meaning is a way we measure the stories we use to contextualize emotional experiencemeans there exists a functiong: F -> meaning. Here, meaning is used to measure F.gis coherent with respect to changes on both sides of the functions inF(the story side and the emotional experience side).5782186
I wrote,
I think Shannon's noisy-channel coding theorem is a counterexample to that claim.
5782280
I don't call those measurements, because there really is a qualitative difference between using integers and using a continuum. It made the difference between the dark age and the Enlightenment. The only thing AFAIK that disappeared completely in the dark age was numerical measurement other than counting. In the middle ages, until about 1300, there were no measuring tapes, no measurement of time, and counting eg baskets of apples. They had scales, but I don't think they were marked; they told you if one thing weighed more or less than another. Tailors didn't take measurements; they had an unmarked string, held it up to your body, gripped one point with each hand, then held it up against a cloth and cut the cloth. If anyone wanted to measure land, they would do it with a cowhide, counting or guessing how many times it would have to be laid down to cover the land. They hardly ever measured land, because they usually cared more about what was on it than how big it was. (The exception was in assigning plots of land to peasants, and they did this in strips, each strip being the same width and length, so no measurement was needed to make sure they had the same area.) They demarcated land claims by landmarks. The only people who may have taken measurements on a continuum AFAIK were stonemasons, who had to do so to build cathedrals from precise architectural designs, starting around 1140 AD; they had developed a rational set of units which gave them a way of measuring that basically produced a binary string (but not a unique binary string; the same length might have multiple representations). But they kept that sacred information so secret that no records of how they used it were ever found, and we can't be sure that they didn't just use integers and ratios. Possibly the Freemasons know.
Partly this was because the Egyptian & Greek systems for using fractions were nearly impossible to do arithmetic with. Measurement returned about 100 years after the introduction of Hindu-Arabic numerals ~1200 AD.
There are enough comments now that I'm having a hard time keeping track of what we're talking about. Going back to the beginning, you wrote,
I believe you can change the way you feel by working with, or just reading, stories. I did so myself, deliberately, though not really believing at the time that it could work. Sometime in 2012 I realized that I was crippled as a writer by my early focus on SF and stories about ideas. When I tried to write a story, what came to me were abstract propositions which might be demonstrated in a story. Not characters, not scenes, not denouements. Themes without feelings. I had stopped reading fiction years earlier, partly because I was bored with that kind of story but didn't realize it. I stopped writing idea stories, and tried to read and write only character stories. I remember after writing "Loyalty" (late 2013) that I was stunned after reading the story--not that it was great, but that I'd actually written a pure character story, with no new ideas in it, and it didn't suck. I hadn't really believed that I would ever be able to do that.
I think that going thru that time of focusing only on characters changed me, not just as a writer, but as a person. To write about characters I had to think about characters and their personal problems; and that made me notice them more in everyday life. I think I became a more caring and sensitive person. I'm sure I found myself crying at sad movies more. It is hard to disentangle; in 2013 I was laid off from my last job just in time, maybe, to avoid killing myself. My career had left me so dead inside that sometimes I wanted to write hurtful, vicious stories, just to see if I could feel, or at least make someone else feel, even if it was only pain. I don't think I really differentiated between hurting myself and hurting others. I wanted some kind of connection with emotions, and pain was the easiest to grasp. So some of what happened was, I think, that in trying to write things that would hurt people, I had to think deeply about what hurt people, which made me more sensitive and empathetic. That seems counter-intuitive, and it probably wouldn't work for anyone who wasn't so analytic that, beginning with the goal of hurting others, they would analyze the emotions of others to the point of developing more empathy for them.
Then you wrote,
(BTW, David Chapman has written a bunch about the distinction between "meaning" and "meaningfulness" which you might find interesting, though I think he makes several critical wrong assumptions.)
The fact that "meaning", as you're trying to use it here, is perhaps the most-important concept in our lives, and yet is (I think) totally non-applicable to the thing we're trying to apply it to, is a huge red flag which says that philosophy is entirely broken--that all our philosophical ontologies and epistemologies are worse than useless to us, and only send us wandering in circles. I think you mostly agree with me about that, and also that, completely accidentally, AI based on deep learning solved all the mysteries of epistemology and is obviously the right tool to use to understand human language and thought and what this "meaning" stuff is.
(This is where Chapman went so disastrously wrong. He was in AI, but instead of turning from symbolic AI to distributed representations, he turned to that classic West Coast failure mode, Buddhism.)
You wrote,
Right; you are advocating the empirical approach of nominalism, in which we define concepts operationally, by how they are grounded in sensory experience, rather than the rationalist approach of essentialism, in which we try to think real hard and define words solely in terms of other words. But I think "meaning" and "significance" are both very misleading terms, and we should avoid them. They're both purely linguistic. "Meaning" and "signifying" are both communicative acts, and the "feeling of what happens" (Damasio's term) is not a communicative act. "Meaning" and "significance" lead back to the endless recursion of words defining words defining words. We need instead to bottom out in pixels, sine waves, and sense-receptor grids. We're looking for qualia, and they must be there somewhere between words and sense-receptors. Maybe information theory will be useful in characterizing the structures of qualia, but I think the term "pleasure" is what we're looking for, not "meaning".
The word "pleasure" also misleads, because of Buddhists, Platonists, Stoics, Christians, and to some extend Hindus, who've spent 2000+ years trying to make people forget that "pleasure" includes love, friendship, loyalty, a sense of accomplishment, and empathy with the pleasure of others. But there is a core meaning to it, and if we can strip "pleasure" of the connotations of "hedonism", I think it's the right term where "meaning" is the wrong term.
We must avoid thinking of pleasure using the rationalist essentialist ontology, in which "pleasure" has an essence, and the only thing you can do with it is add it up. E.g., that pleasure is made of "utilons", and the only difference between pleasures is how many utilons they have. Essentialists don't believe that structure is important, in exactly the same way they believe that structure of information has no role in the meaning of, e.g., texts or paintings. I wrote a bunch about this in Worst of Bad Horse, which I think you have a copy of. (There's a section comparing the meanings of different paintings which is relevant, and should refer back to any earlier relevant sections. It's currently in the afterword to "Self Image", but I think it was somewhere else in the 1st edition. BTW, whenever you want to read WOBH, send me a message & I'll send you the current version. The 1st edition has many problems.)
Pleasure sensations can presumably be structured and combined to create vast, incomprehensibly complex pleasures in the same way that bits can be structured and combined to create incomprehensibly complex meanings. (By "incomprehensibly complex" I mean not that we can't perceive these meanings, but that we can't analyze them in our heads, in a reductionist manner, down to the pixels and sine waves they're made up of; we would have to write it all down somewhere and program a computer to compute answers to our questions about it. These large structures "point" to other large structures, and the grounded meaning of all these structures is included in any computations we do with those structures, but indirectly, implicitly, which may get the right result, but doesn't help us understand the structures themselves.)
I'm splitting this response into two comments. This first comment is for the parts I wanted to respond to. The second comment is for the parts I didn't want to leave hanging.
5782492
Huh. I hadn't considered the math regressions that might have happened during the Dark Ages. That's pretty interesting.
5782496
There are several interesting things about this. First, that increased empathy might be learnable in a fairly straightforward way. Second, that increasing empathy is a thing people can do by accident. Third, my security brain says, that you can probably infect people with empathy without their consent. I feel like that should be a lot less horrifying than it feels.
Will check it! Though honestly, unpacking a theory of meaning is lower on my list of priorities at the moment. I'll come back to it at some point, but I'm focused on some more mundane problems for now.
Sort of. I'm trying to tie together meaningfulness and experiences so I can pull from experiences to understand meaningfulness better. It does result in grounding each instance of meaningfulness in experience, but that's not the same as grounding the definition of meaningfulness in experience. Trying to ground a definition in experiences feels like a type error even in cases where the underlying concept is derived solely from experiences. Definitions should be conveyable, and experiences are not conveyable. Maybe this is what you intended to say.
If you're familiar with FiO– which I hope you are– I think the word you're looking for is satisfaction
.
So that's why they call you Bad Horse
.
I didn't know that happened after you joined pony. Your case was actually part of the reason why I was okay with leaving my job to do my own thing a few years ago. I found out a few weeks ago that some pony AI work I did ended up inspiring yet another person to do the same.
I'm assuming it's the same for you, but I gained a much broader perspective on work after I left my last job. A salaried job isn't necessary for meaning or productivity, and it can easily run counter to it. A job is also just a job, and there are a lot of other facets to life, many of which can easily be important enough to make it okay to significantly deprioritize, or even give up, a job. I think a lot of people naturally have this perspective, but for those of us that are prone to tying our identity to our work, it needs to be learned.
I was planning to read the afterword, and possibly Self Image after seeing your post. Can you send the current version?
Comment #2: The parts I didn't want to leave hanging.
5782492
I agree. In the specific case of integers, they can be embedded within a continuum, so you don't lose anything by choosing to note down continuum values where you'd normally use integers. However, not all data types can be (usefully
**) embedded in a continuum.** In all of these cases, you can always make an arbitrary choice of how to represent things in a continuum, but then you need to specify the translation tables/functions for how to convert observations into continuum values, you need to engineer the functions to be self-consistent when deducing properties from those observations, and those tables/functions are probably going to cost a lot of effort while providing zero new insight. Why not just record those observed values directly without all of the unnecessary conversions?5782496
Why non-applicable? The connection seems pretty concrete, though I didn't write it all in one place. Maybe I'm misunderstanding what you're referring to.
Honestly, I think that's just because we (as a civilization) suck at science. There are plenty of mathematical tools for systematically converting implicit knowledge into explicit knowledge, even without the aid of giant neural networks.