Case 1. On a train to a new destination in a foreign country, you are lulled to sleep by the gentle rocking of the carriage. Suddenly, you are startled awake by a sudden stop and the opening of the compartment doors. You realize you must have missed your stop! You leap up, gather your things, and jump off. You have no map and no phone. Disoriented, you wonder, where am I?
Case 2. You are using your virtual reality headset to explore a high mountain ridge in the Alps. As you walk along the thin edge of a precipice, you trip over a concrete block that your business partner, a known practical joker, put on the floor of the room you are in. A rush of fear combined with a disorienting return to external reality jerks you from your VR experience back into the room.
What is your mind doing when it reorients itself?
1. The self in experience and decision
Exploring these sorts of disorientation can help us to articulate the structure of what it is to be a self. Understanding the constituent features of one’s self is highly relevant to questions of artificial intelligence and to designing a machine that could be a thinking self or that could think like a human.
A related philosophical connection is to recent work on the nature and structure of transformative decisions, experience, revisionary epistemic change, and self-change (Paul, 2014a). That work explores the deep epistemic structure of how we understand who we are, and how we re-construct ourselves through major epistemic upheavals. With transformative decision-making, the focus is on decision models for first personal decisions. A central idea involves the concept of transformative epistemic change: some decisions can lead to epistemic changes so profound that they create significant self-change. An example that brings out the idea involves a congenitally blind saxophonist.
Imagine a blind adult who makes his living playing the saxophone. One day, he is offered a one-time-only chance to have retinal surgery to become sighted. How does he assess the decision of whether to have the surgery? In this kind of case, before the epistemic change occurs, there is no way for him to imagine or represent the experiential nature of the change. What will it be like for him to become sighted? In effect, he has the chance to have a new kind of experience, an experience that he cannot assign a subjective, experiential value to. He cannot assign an experiential value because he lacks the capacity to imaginatively represent the nature of this lived experience.
The situation raises a distinctive set of decision problems, some of them associated with his inability to grasp his possible future self as a sighted adult. The problem illustrates ways we can lack the ability to imagine and model our future selves, and to assign values to possible lived experiences when our futures involve dramatically changed selves. By extension, it illustrates problems with formulating diachronic decision rules for radically incommensurable selves.
All of these ideas, at bottom, are founded on an understanding of the nature and structure of the experienced self, and on a picture where, from the first personal perspective, in many ordinary contexts, we are selves who plan, decide, and act as we evolve forward in time.
2. Constitutive features of selves
Today, I want to focus on identifying some of the foundational elements of the experiencing self. My interest in the examples of the train and the virtual Alps is in how they can be used to highlight constitutive features of thinking selves. Such features may be so basic that we don’t explicitly attend to them in ordinary contexts. The examples also help us to take a distinctive perspective: they help us take the first personal view of the self. The view of the self, a view from an immersed perspective, is different from the view on the self.
To construct a machine that can think like a human, we want to find a way to represent a first personal point of view and capture the way that a self, from its perspective, deliberates and functions in the world. Ordinarily when people think about the self, they start by thinking about how an individual recognizes itself as a thinking self, usually in terms of what that self values or desires, its intentions, and how it has self-awareness. But this builds in a lot right away.
I am starting at a more fundamental level. To understand what a self is, and how an individual knows who she is, we need to understand her immersed perspective. We need to understand the view of the self in question. This involves an exploration of the fundamental experiential structure of a first personal point of view.
So my first point is that the underpinning of an individual’s understanding of who she is, of her self-conception and self-awareness, is structured by her consciously centered, experiential point of view.
What is a consciously centered, experiential point of view? An example, couched in terms of camera angles, can help to bring the idea out. Think of the sort of view that you get with a Go Pro, a type of digital camera designed for filming action while being immersed in it. Or take an immersive computer game. An “immersed first personal viewing angle” is a distinctive and important camera angle that you get from, metaphorically, occupying the boots of your character in a computer game. The view is as though you were looking out the eyes of the character, seeing the world as it sees the world. This captures the first personal visual perspective of the player. Computer games can add a further level of cognitive immersion from an action camera if they give you a certain amount of control over the character’s visual perspective and actions.
The immersed first personal camera angle, set up as though you were looking out from the eyes of the character, gives us a visual analogue of an individual’s consciously centered, experiential point of view. Note that the analogy is only partial, because the centering it represents is largely just visual and causal. A person’s first personal perspective brings in more than this, as it is both a sensory and cognitive centering of the perspective.
Part of why I am emphasizing immersion here is that, when thinking about these issues in the abstract, we can miss details by moving too quickly. We can shift, almost without noticing it, into a third personal approach to the self. This shift is like shifting from the immersed visual angle where I am occupying the boots of my character to a third personal viewing angle using a “follow camera” to track my character. The importance of this difference is represented in how distinctive it can feel to make this visual shift in gameplay. Moreover, the shift in perspective can change the way the player is able to solve tasks in the game, aligning the perspectival shift from first personal to third personal with a functional shift.
When reasoning about the self, if we only explore the third personal angle, we miss the difference between, for example, a self being located in time and space and the experience of being located in time and space. A self may be located in space and time, but it’s the immersive or centered experience of being located which is a constituent of the self. This is not the same thing as just having a location in space and time!
Further, if we miss the crucial difference between having a location and the immersive experience of being located, we can miss the deeper structure of how the immersive experience of being located is comprised of a sense of being here, now, along with coordination to external spatial and temporal cues.
My second main point is that these sorts of immersive experiential features are part of what make up the self. In the two examples I started with, you are disrupted along some dimension of your first personal orientation. In the train case, where you wake up and jump off the train, you are spatially disoriented, because your internal representation and monitoring of your spatial location (which involves keeping it correlated with the external facts) has been disrupted. To orient yourself, you need to recalibrate by finding your location on a map.
This disorientation highlights the immersive, first personal experience of being located which is different from knowing your location on a map. Ordinarily, I have an immersive experience of where I am created by constantly coordinating or updating my immersed first personal sense of being at a location with my third personal perspective or map view of where I am.
Time and temporal experience are the same. I engage in regular calibration and updating of my experienced temporal location by comparing my experienced sense of what’s present or now, and my sense of how much time has passed, and coordinating it with my location as understood externally, using a clock.
There is an even deeper sort of temporal coordination involving the direction of time. My personal sense of time passing, and of the deep difference between the past and the future, are fundamental structural features of my point of view, and I orient myself in the world by coordinating this internal point of view with the external world. (Consideration of time travel cases can bring this out: imagine looking out the window of the time machine and watching the world running backwards as, inside, you live forwards). My internally directed, asymmetric sense of what counts as past and what counts as future are constituents of my first personal self. I find myself balanced in the nexus between the past and the future, and direct myself towards the future. (Relatedly, this gives me an internal sense of the direction of causation).
The structure of our immersed, centered temporal and spatial experience also includes another element, a more esoteric sense of who I am. I know I am here, I know I am here now, but I also know that I am me. When I anticipate, I am thinking of my future. When I remember, I recall my past. And the same is true for you. If you lose your memories, there is an important sense in which you no longer know who you are. If you had an accident where you lost your memories, you’d be disoriented with respect to who you are. And this is temporal and causal: you need to know that your experienced memories are representations of past experiences that played a role in creating who you are now.
Also note that, to recalibrate your sense of who you are, it isn’t just a matter of thinking of past experiences. You need to recognize these thoughts as your memories. If you had the memories but somehow didn’t recognize that the first personal experiences you are recalling are your experiences (perhaps you thought they were false experiences, or experiences of someone else’s first personal perspective), you would not recover your sense of self.
So the point here is that an immersed first personal representational sense of one’s own memories are essential elements of the centered conscious experiencer (the first personal self) (Paul, 2014b). At least one crucial way I know that I am me, and how I define who I am, is that I grasp my memories (as my memories). I sync and update my current experiences as causal outgrowths of my past experiences, and I recognize my memories as my past experiences.
Similarly for the temporal character of anticipation: I have to know what counts as a memory versus what counts as an anticipation. I must distinguish my past selves from my future selves, and recognize my future selves as future.
Very briefly: this can be important for rational deliberation and action, and it comes up in the discussion of transformative experience and decision. In the case with the congenitally blind saxophonist, the trouble is that he cannot, in the ordinary way, project himself forward into the shoes of his possible future self. What he wants to be able to do is consider future ways he could live, or future ways he could be, as a sighted individual. But in an essential sense he can’t mentally project or evolve himself forward from his immersed perspective. He has to become sighted to know what it will be like to be sighted. Before the operation, he faces an epistemic wall that he can only get past by having the experience itself. The further implication is that this epistemic change, becoming sighted, will scale up into a change in who he is. The addition of sight fundamentally alters the structure of his immersed experience, and by extension alters the nature of his centered, conscious, experiencing self.
The point generalizes, especially because there are many other new kinds of experiences that can transform you, such as going to war, becoming a parent, or experiencing massive technological transformation. A profound epistemic change in the nature of an experiencer’s first personal perspective can lead to a restructuring of his values or preferences, and thus can change, in a deep way, who he is. This again connects to AI, for the building blocks of AI include a conception of what a self is, how it is structured by its values, and how it makes decisions and updates itself in response to the external world (See Paul 2014a for further discussion).
3. Modality
What are some other features of the self? We can tease further elements out with more examples. What happens when you wake up from an intense dream, in an unfamiliar room? You are disoriented until you recall where you are and why you are there. Your immersed qualitative experience distinguishes between different realities, distinguishing what it takes to be real versus what it sees as merely the experience of the dream. The immersed self, then, wants to distinguish between the real world and other worlds, and needs to know what’s real to know which features of its experience are part of who it is.
Now we’ve got several distinctive features that characterize what a self is: one’s spatial and temporal immersive sense of being here and now, paired with regular updating to external spatial and temporal cues, and a directed difference between the past and the future. These blend with causal experience and the sense of having one’s own memory, to give us a located, centered, and directed point of view that makes an implicit distinction between what’s real and what’s not. In addition, the updating and monitoring of location and other elements of my centered conscious experience seem to be an internal way of tracking and modeling myself, and in this way knowing myself: as I think of it, it forms part of my intuitive self. I use it to control, create, and know who I am.
It also defines a boundary between who I am and the rest of the world. (In the following sense: when I’m mentally coordinating my immersed perspective with external cues, I’m defining myself in juxtaposition to the rest of the world. Finding myself on a map, coordinating my sense of time’s passing with the movement of the hands of the clock, and distinguishing reality from the dream world all help me know where I end and the rest of the world begins).
There are surely additional features of the self to explore. One important feature of the centered conscious self involves the nature and character of its experienced sensory information. Another very important one involves the self’s relations to other people. Once we have the basic structure of a focused and centered first personal perspective in place, I’m inclined to think that another constituent of what a fully realized self is involves its relations to other people and things. It may be that a distinctive element of the self’s relations to other selves is its representation of those other selves as selves or as conscious beings.
Now that we have all this in play, I’d like to go back to the virtual reality example. The case of the virtual Alps, where you stumble over a concrete block, is a case where you are disoriented because your immersed representation of the world, a virtual world defined by your visual immersion, has been disrupted.
You know where you are in the virtual reality, but you also need to know where you are in the external reality of the room. The concrete block disrupted your orientation in your visual (virtual) reality. To re-orient yourself, and to avoid tripping over the concrete block again, you have to recalibrate and coordinate your immersed visual perspective of your virtual reality with your tactile perspective on your external reality.
The example brings out how immersed experience in a kind of reality is a feature of the centered conscious experiencer: in the virtual Alps case, we can contrast your immersed visual experience of the virtual world with your immersed tactile experience of the external or real world. A VR user might even need to exploit her understanding of the contrasting modalities (virtual and real) for problem solving: imagine she has to find her way around a virtual boulder in her Alpine VR experience, but to do this she has to open a closed door in the external reality of the room she is in. Opening the door will move her around the virtual boulder.
What is the mind doing when it solves the task of the virtual boulder and closed door?
To successfully perform this task, she needs to clearly distinguish the two modalities she is working in: virtual and external, and she needs to coordinate between her visual VR’s spatially, temporally, causally, immersed point of view and her tactile external spatially, temporally, causally, immersed point of view. Finally she must manage the interpretation between them.
This brings out how a centered understanding of the features of an agent’s point of view can frame her actions and define her problem solving. Let’s make the story a tiny bit more complicated: imagine that the virtual boulder has to be pushed aside with the help of other VR users, and the door is too heavy to be opened without their help as well. Together, you have to move the boulder by opening the door. You are the team leader.
What is your mind doing when you solve this joint action problem?
Here, part of what you need to solve the action puzzle is to clearly distinguish and represent the first personal features of your virtual and external realities, as well as represent the different virtual and external perspectives of others, and then coordinate between their represented modalities and your own modalities.
Again, computer gameplay has an analogue: in a multiplayer game where you can see the points of view of your teammates, you have your own first personal view in addition to the first personal views of others embedded into your screen. The structure of joint action can also be applied to cases where you are negotiating with or understanding yourself at different times: your self at a past time, at a future time, or even at a merely possible time (a merely possible location or situation) can be treated like another agent with its own first personal view. If so, the complicated virtual reality joint action case isn’t just for acting and making decisions with other people. We do something similar when we are making decisions for our future selves or our merely possible selves. That is, we sometimes need to be able to represent and understand the points of view of our future selves, our past selves, and our merely possible selves in order to act rationally. (The example of transformative experience I discussed above, where the saxophonist must rationally assess his possible future self, is another example). These are just some of the ways in which understanding the self connects deep philosophical issues to exciting questions in artificial intelligence.
References
Paul, L.A. (2014a). Transformative Experience. Oxford: Oxford University Press.
Paul, L.A. (2014b). “Experience and the Arrow”. In Asymmetries of Chance and Time, 174-193, edited by Alistair Wilson. Oxford: Oxford University Press.