In the previous post, I put forth the question of whether Husserl’s phenomenology could be of use to AI, weak or strong. This is a genuine question that I put out there to discuss…I have no thesis to support. Just curious to hear what you think.
In writing this post, I realized I’d have to break this down into several segments. From now on, I’ll be using Husserl for the most part, not Heidegger, to explain aspects of phenomenology…although I do like Heidegger’s readiness-to-hand and presence-at-hand distinction. But I prefer the bracketing methodology of Husserl for these purposes. I could see Maurice Merleau-Ponty coming into the picture, especially on the issue of AI embodiment, but I haven’t read him. (Perhaps those of you who have can weigh in. I’d love that.)
I might stray from Husserl too, setting out on my own. In other words, not everything here will be a lesson on Husserl. I don’t want to be encumbered by referring back to his works to verify what I’m saying, because that would make what should be a simple blog post an academic enterprise. I’m not feeling that game right now.
Conditions of experience
Phenomenology allows us to describe experience as it’s actually experienced. In doing so, we look for conditions that make experience possible—the constitution of meaning. These “rules” are not likely to be revelatory in describing what happens inside a biological brain. However, phenomenology could run parallel to neuroscience. After all, in order to know what’s going on in the brain, we must know what brain states correspond to—the so-called “subjective” experience, i.e. 1st person accounts. One might argue that 1st person accounts tend to miss the mark, fall into error, but we can’t allow all 1st person accounts to err on a grand scale. There must be a back and forth here, perhaps only a preliminary one at the outset. There is no mapping of the brain without knowing what it is we’re mapping.
Why should we care about a philosophy that sounds very much like navel-gazing? Well, this navel-gazing isn’t about the stuff we ordinarily think of as “subjective”: our favorite ice cream, the personal feelings we get when we listen to music…that stuff we generally agree is “a matter of taste.” Husserl’s direction is actually scientific (like, Wissenschaft scientific, “the sciences” scientific) in the sense that we are looking for elements of experience that are essential to it.
For example, those of you familiar with Kant’s Critique of Pure Reason may remember that space is the a priori outer form of experience, and time, the inner form. Causality was explained in this way too; everything we experience will be shaped by the categories because these are necessarily presupposed. (Kant also believed there were inexperience-able things “out there”—noumena—which causes phenomena. Let’s leave this aside.) Husserl goes further than Kant by setting forth a philosophy that seeks to ground the content of experiences individually, on a case by case basis. We’ll see how this works in later posts. Let’s just say for now that Husserl’s phenomenology is a lot more detailed and specific.
The very fact that phenomenology seeks out “rules” makes me wonder if it could apply to AI in some capacity, especially in areas that have to do with perception and learning. It might actually be preferable to bracket the “natural world”: “objective” reality, Kantian “things in themselves.” In a way, we’re looking at our own experience as if it were virtual reality. Like a computer.
However, phenomenologically speaking, we live in an environment that is not closed, which seems to imply that computers just aren’t like us. It seems that AI would have to progress significantly to allow for open-ended possibilities if we want to achieve those hard-to-accomplish tasks that for us seem basic. Does that which allows for creativity and learning in us preclude algorithmic AI? Maybe, maybe not. I’m not well-informed in this area, but it seems at the very least we’d have to know what makes our experience what it is in order to answer the question. Do we really take in new information just as it comes to us, spontaneously, or do we have to synthesize that information onto pre-existing charts? I suspect the latter, and I suspect if we could “crack the code” that allows us to understand our own learning methods, we’d be better able to do the same for AI (even if only in weak AI, or for certain specific goals).
In my last post I told you I’d explain how phenomenology operates by exploring Husserl’s intentionality. Let’s do phenomenology.
Husserl’s Intentionality is at the heart of his phenomenology. Intentionality is our directed-ness toward things, and it’s basically this: Consciousness is always consciousness about or of something. Pause here for a moment. Really stop and give this consideration. Much of phenomenology is reflection on experience. If you don’t do it, if you read articles on phenomenology and look for ways to summarize the logic, to relate to it only on the level of mere verbal cohesiveness, you’re missing a crucial aspect of it. The process is intuitive. You analyze the veracity of such statements as “consciousness is always consciousness about or of something” via intuition, reflection on your own experience.
Try not to think about anything. You might think you’ve experienced something like this once: a dreamless sleep, a coma perhaps. But were you conscious? No. So right now do this: Really try not to perceive something, to be aware of. You can close your eyes, close the windows, block out the sound, but time goes by. What happens? Well, if you’re like me, perhaps even more happens in your consciousness now that the senses are closed off. Ideas, daydreams, random thoughts…these are included as content, “about-ness.”
Those of you who meditate may raise objections, and these will be well taken. I, for my part, have never found myself to be conscious while being conscious of nothing, absolutely nothing.
It is the nature of our experience to be directed towards things or about things. (What I’m loosely calling “things” are not just objects of sense perception, but can include thoughts, ideas, memories, etc.) Intentionality is always there. In other words, it plays a pervasive role in every kind of experience: perceiving, judging, remembering, dreaming, screwing up, etc.
Imagine an omniscient camera (or recorder of some sort) that captures the infinity of experiences, all sense data, equally, without any directness toward things, without signifying any particular experience. We are not even a time-limited “subjective” version of such a camera. We can speak of this omniscient experience just as we can speak of a square circle, but we can’t really picture it. That’s because, in an a-logical—non-logical—way, it is nonsense. Through intuition we know that in such a world, there would be no objects. No objects, no intentionality. No intentionality, no objects.*
You might’ve guessed by now that intentionality is broader than what we mean when we say, “I intend to fix this,” but includes such statements and meanings. Plus, intentionality is not attention, necessarily, but includes attention.
What intentionality does is acknowledge that there is always a foreground and background to experience. The background is a vague summation of the world. This world may not be the world of science, may not include the world ‘in itself’ (or it may, phenomenologically, but let’s not get too complicated here). Let’s say for now that, at a minimum, it’s a world that’s available for us, and therefore it coheres in a loose sense—it must. This background is what Husserl calls the “horizon.” It can be thought of as a potential experience, past or future, which has not yet shown itself or is not now in view. The horizon is also infinite (more on this later.)
Intentionality is mostly passive as we go about our everyday lives, and on philosophical-phenomenological reflection we can “see” it operating, to some extent.
We quickly disregard what isn’t relevant to us at the moment while simultaneously knowing that those things that are currently irrelevant or out-of-focus—on the horizon—are possible experiences that could come into the foreground. Those background possibilities constitute our foreground experiences. We know what’s behind us in a loose sense. We have expectations about what’s behind us and those inform our foreground experiences.
I repeat, these foreground experiences are not necessarily “paying attention.” More often than not, we’re not trying to focus.
We grasp content in its context, leaping ahead to the most likely meaning or its totality, its unity, often unaware of other possible meanings or interpretations of the content, although further investigation may warrant a change. This is all done in a flash due to the intentional nature of our experience. The horizon, the background, is operating at the same time that we make the leap. The meaning of words/objects are constituted in time and situation, and this constitution is holistic, yet adaptable and subject to constraints.
Furthermore, the object or content of the experience is the way we look at it. Here’s a good example found in this article:
Consider the plight of poor Oedipus Rex. Oedipus despised the man he killed on the road from Delphi although he did not despise his own father; he desired to marry the Queen although he did not desire to marry his mother; and he loathed the murderer of King Laius before he came to loathe himself. But of course the man he killed was his father, the Queen was his mother, and he himself was the King’s murderer. How shall we describe the intentionality of such acts? Oedipus’ desire, for example, seems to have been directed toward Queen Jocasta, but not toward his mother. But Queen Jocasta and Oedipus’ mother were the very same person…Oedipus’ desire was therefore not simply “for” Jocasta: it was for Jocasta as conceived in a particular way. And the same sort of thing is true, not only of Oedipus’ mental states, but of everyone else’s as well…The intentionality of an act depends not just on which object the act represents but on a certain conception of the object represented.
The intentional conception of an X is not just an imposition of our minds on “facts” and therefore subject to error. (Remember, intentionality is always there, and it doesn’t always err. Error is just a clear way of showing the difference between fact and intention.) The example above demonstrates how meaning is constituted, but also how new conceptions can arise from new evidence. The meaning of Oedipus Rex would be entirely lost on us if we did not understand Oedipus’ intentions and the context which guided those intentions.
*Here I’m combining “object” and “content” for the sake of avoiding pedantry. We’ve established we’re not talking about noumena, so I hope you’ll excuse my sloppy language.
Let’s look at our intentionality, our guiding mental behavior, linguistically.
Consider the sentence: The pig is in the pen.
I would be incredulous if you interpreted this sentence to mean, “There is a pig that is inside a writing instrument.” (Unless you happened to look down at the picture first, and you probably did because images tend to command attention. And there’s another topic for discussion…but anyway. Pretend you didn’t.)
The truth about the world, the background—that pigs don’t fit in writing instruments—informs your foreground interpretation. Yet you did not (I hope) have to analyze the sentence and determine all possible meanings of the word “pen” in order to arrive at your interpretation. You probably didn’t even think of writing instruments.
Consider the sentence: The pig is in the pen. Then imagine someone pointing to this while saying the sentence:
You might laugh and say, “Well, the pig is on the pen, or maybe the pig’s relationship to the pen is something about which we don’t wish to speculate.” Whatever the case may be, the sentence now has a different meaning constitution. You might wonder…why would the speaker say, “The pig is in the pen?” Does this person speak English? Is this person having a prepositional brain fart?
And the best question: Would you have considered “pen” in this case as signifying “an enclosure for animals”? Probably not in this situation.
Or maybe the speaker of the sentence is a moderately funny, punny person who has this whole theory about truth and language and you two have discussed this pig in the pen example on many occasions.* In this case, you might grasp both meanings of “pen” simultaneously to get the joke. You might only get the joke because you know this person makes this sort of joke on a regular basis.
As you can see, the holistic interpretation is adaptable and situational; even as it “runs ahead of itself,” it is subject to all sorts of constraints. In other words, intentionality is not just some willy-nilly imagining of the world, some sort of act of creation from nothing.
Also, this “leaping ahead” applies to all experience of objects, not just language interpretation. In my next post, I’ll go into further detail on this topic. Be on the lookout for eidos…
Ha ha. (Okay, not funny. But you’ll “see” what I mean later.)
*This is my husband’s example, which he used in a different context in his unpublished book on language and generosity (Donald Davidson’s “charity”).