15 • The Relevance Problem
At the core of what makes us intelligent is our capacity to zero in on what is relevant. Can we have a scientific theory about relevance? A deep journey into how science works suggests we cannot.
Last time we established that at the core of our intelligence is our capacity to zero in on what is relevant. We’ve been looking at various different aspects of our cognition where relevance realization is crucial:
Remember this:
All of the information available in the environment, overwhelming combinatorially explosive! You have to selectively attend to some of it. So this is doing relevance realization: selective attention.
And then you have to decide how to hold in working memory what's going to be important to you. Lynn Hasher's excellent work showing that working memory is about trying to screen off what's relevant or irrelevant information.
You're using [working memory] in your problem solving. And here is where you are trying to deal with the combinatorial explosion in the problem space.
That’s also interacting with the proliferation of side effects […] when you try to act. You're trying to select: What do I hold in mind? How do I move through the problem space? Once I start acting, what side effects do I pay attention to? Which ones do I not pay attention to?
And all of that has to do with, out of all the information in my long term memory, how do I organize it? How do I categorize it? How do I improve my ability to access it? Longterm memory organisation and access is dependent on your ability to zero in on relevant information.
[…]
This is the Relevance Problem. That's the problem of trying to determine what's relevant. It's the core of what makes you intelligent.
Why does that matter? What I'm trying to show you is how deep and profound this construct is, this idea of relevance realization is at the core of what it is to be intelligent.
And we know that this isn't just cold calculation. Your relevance realization machinery has to do with all the stuff we've been talking about:
it's about what motivates you
what arouses your energy
Relevance realization is deeply involving. It's at the guts of your intelligence, your salience landscaping, your problem solving.
If relevance realization is at the core of what makes us intelligent, then how does it relate to meaning?
Connections = relations of relevance
Here's the proposal: That what we were talking about when we talked about meaning in terms of the three orders — the nomological, the narrative, and the normative — were connections that afforded wisdom, self-transcendence.
But what connections? The connections that were lost in the meaning crisis: The connections between mind and body, the connections between mind and world, the connections between mind and mind, that connection of the mind to itself. These are all the things that are called into question. The fragmentation of the mind itself.
And we saw how this, all throughout, had to do with the relationship between salience and truth, what we find relevant in terms of how it's salient or obvious to us and how that connects up to reality. And how it connects — remember Plato — parts of us together in the right way, the optimal way.
What if what we're talking about when we're using this metaphor of meaning is we're talking about how we find things relevant to us, to each other, parts of ourselves relevant to each other, how we're relevant to the world, how the world is relevant to us? All this language of connection is not the language of largely causal connection. It's the language of establishing relations of relevance between things.
Perhaps there's a deep reason why manipulating relevance realization affords self-transcendence and wisdom and insight, precisely because relevance realization is the ability to make the connections that are at the core of meaning. Those connections that are quintessentially being threatened by the meaning crisis.
That would mean if we get an understanding of the machinery of [relevance realization], we would have a way of generating new psycho-technologies, redesigning, reappropriating older psycho-technologies and coordinating them systematically in order to regenerate these fraying connections, re-legitimate and afford the cultivation of wisdom, self-transcendence, connectedness to ourselves and to each other and to the world.
At the core of meaning is our ability to connect to the world, connect to our selves, and to connect to other people. And all these connections, which are relations of relevance, come about through our capacity for relevance realization.
If we can understand how relevance realization functions, if we can make sense of it, we have the potential to answer a lot of the questions we are facing in the meaning crisis.
So let’s get to it and figure this out… what do we need for a proper scientific theory?
How science works
Inductive generalizations
Now, of course, the philosophy of science tackles all kinds of controversial claims about what is science and how science works. But I take it that one thing that is agreed upon in science is that science works through inductive generalization, or it tries to generate inductive generalizations.
What do I mean by that? In science you study a bunch of things here and then you make predictions and claims that will be the case for all of that type of thing. Here, I study a bunch of… here is a hunk of gold, here is a hunk of gold, here is a hunk of gold… I come up with a set of features or properties. Does that generalize to all the instances of gold? And if it does, then I come up with an inductive generalization. I want to get the broadest possible inductive generalizations that I can, because that’s how science works. It’s trying to give us a powerful way of reliably predicting the world.
It’s doing other things: Very importantly it’s also trying to give us a way of explaining the world. […] This is not meant to be an exhaustive account of science. It’s meant to point to a central practice within science, but a constitutive practice nevertheless. If you can’t generate inductive generalizations in your purported endeavor, then you don’t have a science. This is why pseudo-sciences like astrology fail, precisely because they cannot do inductive generalizations.
Systematic import
What J. S. Mill pointed out is: that means that we need what’s called systematic import. And this is so relevant to what we were talking about last time, even using the word import is really relevant.
What that means is: Science has to form categories, because that’s what I’m trying to do. I’m gathering a bunch of things and saying they are the same type of thing, they’re all instances of gold, they all belong to the category of gold. Science has to form categories that support powerful — meaning as broad as possible — inductive generalizations. To be able to do that is to have systematic import.
What do I need? Think about reverse engineering this. In order to have reliable — that’s what powerful means: reliable and broad inductive generalizations — in order to have those, what do I need to be the case here? I need there to be important properties for that category.
Homogenous
One thing is: I need the category members to importantly be homogenous. There’s a sense in which all the members of the category have to share properties. […] And it’s because they share properties that I can make the inductive generalization that other instances will also have those important properties. That’s exactly what I need because if the members are heterogeneous, there’s no set of properties I can then extend in the generalization. They have to be homogenous.
Now this gets us towards something very important. This gets us towards an idea from Quine, because there’s a lot of discussion about [essence] right now in the culture. And I think the discussion is too polarized. And this has to do with an issue made by Wittgenstein. But I want to put Wittgenstein and Quine together on this (very important modern philosopher). And this is what some of the critics of essence say.
If you remember, according to Aristotle […], an essence is a set of necessary and sufficient conditions. And what Wittgenstein pointed out — and remember we did this with the example of a game — that many of our categories don’t have essences. There is no set of necessary and sufficient conditions that will pick out all and only games. There is no set of necessary and sufficient conditions that will pick out all and only tables.
Many of our categories don’t have essences. That was Wittgenstein’s point. Wittgenstein, I don’t think you could ever pin him to the claim that no categories have essences. And that’s what some people, I think, have concluded: that no categories have essences — everything is just nominal description. But that’s not right because, of course, non-controversially, for example, triangles have essences. That’s why Aristotle thought many things did. If it has three straight sides, three angles, it’s enclosed, it’s a triangle. That’s an essence to a triangle. Now, that’s mathematical.
Here’s what Quine argued, at least I think this is an interpretation of Quine that is philosophically defensible: Things like triangles, these are deductive essences, these are the essences that we can deduce. But what science discovers are inductive generalizations. And if they are powerful enough, science gives us the essence of something.
The essence of gold is the set of properties that will apply to all instances of gold — all and only instances of gold. That homogenous set that can generalize is what an inductive essence is.
What that means is a couple of ways of talking in the media, or the general culture, should not be so uncritically accepted. Essentialism isn’t bad for things that have essences. Why would it be? Essentialism is the mistake of treating a category as if it has an essence. It is a mistake for things like games and tables, precisely because they don’t have an essence. It is not a mistake for things like gold, because gold has an essence, inductively. Or triangle, because triangles have a deductive essence.
It is too simplistic to say, “everything has an essence” or “everything doesn’t have an essence”. It cuts both ways. There are many things that don’t have essences, that’s what’s right about the critique of essentialism. But it is wrong to say that Wittgenstein’s argument is not an argument, because it’s not a deductive argument that concludes that there are no essences, it only points [out] that many categories don’t have essences.
That means it is possible to do a science when we categorize things in such a way that we get [support for powerful deductive generalizations], because when we get this then we have the essential properties of the thing.
The reverse is the case. That’s what I mean by it cuts both ways. We can’t have a scientific explanation of everything. If the category is not homogenous, if it does not support powerful inductive generalizations, if it does not have an inductive essence, we cannot have a science about it. It doesn’t mean those things don’t exist. It means we cannot scientifically investigate them.
For example, I can’t have a science of white things. Now, are there things that are white? Of course there are. This blackboard is white. This pen, at least part of it, is white. This piece of paper is white. To say there are white things in this room is to say something true.
Notice that: there are truths that are state-able, but the category that I am using — this is J. S. Mill’s example, white things — does not support any inductive generalizations other than "the thing is white". Now don’t give me, “Well, we can have a theory about light and lightness!” We are not talking about a theory about light, we are talking about a theory about white things.
Knowing that this [sheet of paper] is white, what does it tell me? So I study this white [marker]. What do I learn about it? Nothing. Other than it’s white. Is there any other important shared [feature between the sheet of paper and the whiteboard on the wall]? No. They’re both flat, but this is vertical, this is horizontal. You see? It doesn’t generalize.
It is correct to say that there are many categories that we form for which we cannot generate a scientific theory or explanation, precisely because those categories are not homogenous, they don’t have an essence.
Notice what that doesn’t mean: The fact that I cannot have a scientific theory of it does not mean that the white things are made out of ghosts, or dead elves, or ectoplasmic goo. It licenses no metaphysical weirdness. It just says, that category functions in the sense that I can make true statements about its membership, but it does not function in so far as it supports, through systematic import, powerful inductive generalizations.
What else do I need? Let’s compare the white things — as J. S. Mill did — to horses. We depend on the fact that horses seem to have an essence. Whether or not they ultimately do, at some sort of species level or something really argued about in biology — and I’m not trying to be negligent of that, but I’m also not going to try and resolve that. What did Mills mean by his example of horses?
If I learn a lot about this horse, it will generalize to other horses. It will generalize. Horses are in really important ways homogenous. That’s why we can have a veterinary medicine and things like that. I can learn about it in terms of horses that have already been studied, and it will generalize well to horses that have not themselves yet been studied. That’s fine.
Stable
What else? And I don’t mean this to be a pun! I need the category membership to be stable. That doesn’t mean to be "horses in stables". What’s in the category, the kind of things that are in the category, should be stable, shouldn’t be constantly shifting or changing. Because if this — and this was a point made a long time ago by Plato — if what’s in [the category] is constantly shifting — now I don’t mean particular members, I mean what kind of thing is in [the category] is constantly shifting — then of course I can’t do inductive generalizations because I will get into equivocation.
The word "gravity" originally meant having to do with drawing down into the grave. It had to do with a sort of important seriousness. But now we use that term to describe a physical mode of attraction and interaction. If I don’t notice the change in what goes into my categorizations, I am not making a good inductive generalization, I am engaged in equivocation. And as I’ve tried to show you, equivocation is a way in which we make invalid, often ridiculous arguments. So it needs to be stable.
Intrinsic / inherent
We need the properties of the objects to, in some sense, be intrinsic or at least internal, inherent. This also comes from an argument by John Searle. Many objects have properties that are not intrinsic to the object but come from the object’s relationship to us, for example, and they are attributed properties.
A clear example, a sort of non-controversial example, is something being money. Is money real? A lot of my life is bent around money, so in that sense it seems to be real. But does anything intrinsically possess the property of being money? If I take out some coin or piece of paper, is it intrinsically money? No, it’s only money because we all attribute it as being money. We all treat it as money and that’s what makes it money. If we all decide to not treat it as money, it ceases to exist as money. We can’t do that with gold. Now, notice what I’m saying: we could all decide that gold is no longer valuable, no longer analogous to money, but we can’t all decide that gold no longer possesses it’s mass, atomic number. We can’t do that.
Now, the thing you have to remember is that many things that we think are intrinsic are actually attributed. This [plastic water bottle] being a bottle is attributed, because what it means to call it a bottle is the way it is relating to me and my usage of it. If there had never been human beings and this popped into existence because of some quantum event near a black hole or something, it isn’t a bottle. It is an object with a particular mass, a particular structure, but it’s not a bottle, because being a bottle is something that it gets in its relationship to me.
Did I just show you that everything is an illusion? No. The fact that there are many things that are genuinely relational, genuinely attributed, doesn’t mean that I’ve shown you that everything‘s false. I’ve just shown you that you can’t do science unless the members of your category are homogenous, stable, intrinsic, or at least inherent, because that’s what you need to have powerful inductive generalization.
Let’s see something that fails this, all of these tests: Things that happen on Tuesday, events that happen on Tuesday, Tuesday events.
Are there events that happen on Tuesday? Yeah. And there are even events that can happen on multiple Tuesdays. We categorize things in terms of the days, we categorize events in terms of the days.
Are all the events on a Tuesday homogenous? No. Are all the events on many different Tuesdays homogenous? No. They are very, very different and widely varying.
Is it stable? The things that happen on Tuesdays, is it the same every Tuesday? No, that’s Groundhog Day or some kind of horrible Nietzschean hell.
And what about “Tuesday-ness”? Being Tuesday, is that inherent? I mean is there Tuesday in the room when it’s Tuesday? It can’t be because there was a time when we didn’t even have calendars. But notice how hard it is to realize that: there’s no “Tuesday-ness”.
Can I make true statements? “Last Tuesday I went to a movie.” Is it true? Yes.
Can I do a science of events that happen on Tuesday? No, I can’t, because it doesn’t satisfy these criteria [— homogenous, stable and inherent].
Does that mean that Tuesday is made out of ectoplasmic goo? Tuesday events actually take place in a different dimension? No, it doesn’t. None of that!
You have to be careful on — and this is what we learned from Wittgenstein — we have to be very careful about the grammar of our thoughts, how we are regulating our cognition.
Why there cannot be a theory of relevance
Now, what I want to try and show you is that relevance does not have systematic import. Relevant events are like Tuesday events. Here let me show you:
The things that I find relevant, other than me finding them relevant, what do they share in common? I might find this pen relevant. I might find my knee relevant. I might find this air relevant. I might find the fact that it’s a particular day in May relevant. Do you see what I’m showing you? The class of things we find relevant is not homogenous. Other than we finding them relevant, there is nothing that they share. It’s exactly like the class of white things.
What about stable? When I find something relevant, do I always find it relevant? This [pen] is relevant to me now. Will it forever be relevant to me? […] No! Things are not stably relevant — relevant one minute, irrelevant the next.
“There’s things that are always relevant to you!” Always? Don’t know, very hard to find them. Maybe oxygen? But that’s only relevant to me if I want to keep living. A person who commits suicide — and some people commit suicide this way — they suffocate themselves to death, because that was more important to them than oxygen. It’s not stable.
Is relevance — and here’s where I think we will get into some difficulties, I suppose, with some people — but is relevance internal or intrinsic to the object? Is there a way, if there had never been human beings or sentient beings, could this [plastic bottle] have relevance? It doesn’t seem that that’s, at all, a plausible intuition. Relevance always seems to be relevant to someone or something.
And that, of course, I think is going to be bound up, that relevance, ultimately, has to be relevant to an auto-poetic thing. Only things that have needs, only things that are self-organized, so that they have the constitutive goal of preserving their self-organization — that’s what it is to need. I need food because I am self-organized to preserve my own self-organization, which means I need food. Food literally matters to me. It’s hard to see how things could be relevant unless they are in relationship to an auto-poetic thing.
Relevance is not something for which we can have a scientific theory. I want you to notice what has come along the way: Relevance is not intrinsic to something. There can be no essence to relevance; nothing is essentially relevant. That’s the whole point about talking about the problem of essentialism. And relevance is not stable, it’s constantly changing.
Bummer! To the best of our current knowledge, we cannot have a scientific theory about relevance. :(
As you may suspect, however, that doesn’t mean the story ends here…
Mirror of the Self is a weekly newsletter series trying to explain the connection between creators and their creations, and analyze the process of crafting beautiful objects, products, and art. Using recent works of cognitive scientist John Vervaeke and design theorist Christopher Alexander, we embark on a journey to find out what enables us to create meaningful things that inspire awe and wonder in the people that know, use, and love them.
If you are new to this series, start here: A secular definition of sacredness.
For an overview and synopsis of the first 13 articles, see: Previously… — A Recap.