This will be the reading for Saturday, December 5. A printable copy can be found here.
In Being No One, Thomas Metzinger attempts to develop an account of the first-person perspective that is both adequate to our phenomenal experience of being persons and to the empirical data, especially the neurological data. He calls this the self-model theory of subjectivity (SMT). The SMT is a constraint-satisfaction approach to phenomenal experience: it describes what properties representations in an information-processing system need in order to become phenomenal presentations, i.e. the contents of consciousness (Constraint usually implies limitation, but I think he is just as much using the word to mean condition).
In the full version of this work, there are ten constraints, but this precis covers six of them. First, there is the globality constraint, which means there must be a subset of information which is globally available for many different systems at once. This is his way of describing the fact that this information is found within a world. Second, there is the presentationality constraint; this is the window of the “now” in which everything appears as present to us. Third, the transparency constraint is the fact that we cannot get behind experience to see how it is made; we cannot experience our brain processing information. The final three constraints extend and deepen the first three, and we will leave them for part 2.
The Self-Model Theory: A First Sketch
He will start with a minimum concept of consciousness, which phenomenologically, is the presence of a world. This minimal concept is what requires the the first three constraints.
First, global information is the information we are consciously processing, the phenomenally represented information is that subset of currently active information in the system, which is globally available “for many different processing capacities at the same time, e.g, for deliberately guided attention, cognitive reference, and the selective control of action.” (2) Second, presentationality is an island of presence in the continuous flow of physical time. I always experience my conscious states now, because it is associated with a representation of temporal internality: “There is an overarching representational context governing phenomenal experience, and this context generates the experience of presence.” (2)
Third, transparency is a phenomenological concept. It deals with how consciousness appears, not with what we can know about it (which would make it an epistemological concept). It means something particular is not accessible for subjective experience. Specifically, what is not available is the fact that conscious experience is representational. What makes representation transparent is the “attentional unavailability of earlier processing stages in the brain for introspection.” (2) The instruments of representation cannot themselves be represented, and so the system making the experience, on this level and by conceptual necessity, is stuck in a naive realism – one’s phenomenal experience looks “untranscendably realistic.”
This naive realism appears on the epistemological level with the concept of “autoepistemic closure”. It’s an inbuilt blind spot, a “structurally anchored deficit in the capacity to gain knowledge about oneself.” Specially, this closure is that normally humans do not see that the content of their experience has self-structured aspects.
These three constraints come together to form a world. The presence of a world takes a global model of reality within a window of presence , and the model cannot be recognized as a model by the system generating it within itself.. This is how we get the presence of a unified, homogeneous and frozen world of the Now. He calls this “Selfless Snapshot Consciousness”.
He says selfless, because the main ontological claim of the SMT is that “No such things as selves exist in the world.” (3) For all scientific and philosophical purposes, the notion of a self can be eliminated. What we have called “the self” in the past is a form of representational content: the content of a self-model that cannot be recognized as a model by the system using it. The content of the phenomenal self-model (PSM) is the content of the conscious self – not a thing but a process. In a metaphorical sense, we can say you are the content of your PSM. Better, we are systems that confuse ourselves with the content of our PSM. The phenomenal property of a selfhood is a representational construct – “an internal and dynamic representation of the organism as a whole to which the transparency constraint applies.” (3)
The phenomenal experiences of substantiality (of being an independent entity), of having an essence (an unchangeable innermost core) and of individuality (being unique and indivisible) are forms of conscious, representational contents. Having that content is evolutionary advantageous, but not epistemically justified. That is counterintuitive; during waking and dreaming life, a self is present. We experience ourselves as being someone. So how can personal identity appear?
In answering this, the transparency-constraint is key. If all the other necessary constraints are satisfied by a representational system, the addition of transparency will lead to the emergence of a phenomenal self. Transparency “is a special form of inner darkness”, the fact that “We do not experience the contents of our self-consciousness as the contents of a representational process, but simply as ourselves, living in the world right now.” (4)
The existence of a coherent self-as-object introduces a self-world border into the system’s model of reality. It is how system-related information becomes globally available as system-related information, because the organism now has an internal image of itself as a whole. On the other hand, environment-related information is non-self: “Objectivity emerges together with subjectivity.” (4) The representing process generates both the self and the difference between object and self, because the object is a difference from the self. The partitioning between object and state is what allows for relations between self and the other.
Full-blown self-consciousness always involves a relation between the self and an object-component; “The content of a perceptual state is not a part of the environment, but a relation holding to this part.” (5) The ongoing, episodic subject-object relation is the “phenomenal model of the intentionality relation” (PMIR). A functioning PMIR says, “I am someone, who is currently visually attending to the colour of the book in my hands.” The main characteristic of the PMIR is that they depict a relationship between the system as a transparent whole and an object. The PMIR allows a system to experience itself as being a part of the world, but also being immersed in a network of causal, perceptual, and cognitive relations.
The globality constraint is a more refined version of global availability. The globality constraint means that individual phenomenal events are always bound into a global situational context, always part of a world-model. On a personal level of description, we can say that if a person is conscious, a world exists for them, and can make the fact of living in a world available for herself, cognitively and as an agent.
On the phenomenological level of analysis, we see that the contents of conscious experience are characterized by my ability to react to them with a variety of mental and bodily capacities, e.g. directing my attention to a colour. Sometimes I can form thoughts about the colour, or maybe a categorical representation (which means it is available for phenomenal cognition, that is, to consciously think about) or associate it with a previous colour (which makes it available for autobiographical memory), or I can talk to other people about it (that is, it is available for speech control or communicative applicability). I can touch them (it is available for control of action). It is my flexibility and autonomy in dealing with the content of conscious experience.
As a representational property, globality is what makes intentional content (“I am thinking about X”) available “for concept formation, for metacognition and verbal report, for planning or for motor simulations with immediate behavioural consequences.” (6)
On the functional level, globality offers the the generation of an inner world as a computational strategy. In the brain, there is no final phase of processing, but the generation of a coherent world is a strategy to reduce ambiguity in the confusion of the external world [so Kantian]. It also leads to a reduction in data—the amount of information available to the system is minimized, and so computational load is reduced. There is a high-level integration function; it dynamically unifies a large number of small causations in a distinct causal role. It produces a context-sensitive, flexible behavioural profile. The more information is conscious, the more challenges from the environment can be reacted to in a fine-grained way. Only if you have the subjective experience of a world can you come up with the idea of a single reality—or even the appearance/reality distinction.
There is no current detailed theory, especially of the minimally sufficient correlates, of global integration. But there are interesting speculative hypotheses. One intuition has been to study anesthetics—that is, the conditions under which phenomenal experience as a whole disappears or re-emerges. It is also important that the globality constraint applies both to dreams and to waking states. In both, the system operates under a single more-or-less coherent world-model, though its global functional properties differ greatly. A good strategy might be to subtract some global properties of the waking world-model from the dreaming world-model to arrive at a common neurophysiological denominator. The intuition behind that research program is philosophical: what we call waking life is a form of online dreaming; so, “If there is a common functional core to both global state-classes, then conscious waking would then be just like a dreamlike state that is currently modulated by the constraints produced by specific sensory input.” (7)
The world appears by being present; this is the central aspect that cannot be bracketed out. On the phenomenological level, we can see that to have conscious experience is to be in a present. This is a matter of processing information “in repeatedly and continuously integrating individual events (already represented as such) into larger temporal Gestalts, into one singular psychological moment.” (9) What is that moment? The phenomenal experience of time is constituted by a series of achievements: temporal identity (or simultaneity), temporal difference (or non-simultaneity), seriality and unidirectionality (the succession of events), temporal wholeness (the “specious” phenomenal Now), and the representation of temporal permanence (duration). The key step towards a phenomenal representation of time occurs when event representations are integrated into psychological moments. The difficult conceptual point is how can experience a full-blown present as embedded in a uni-directional flow, i.e. the experience of duration. There are Nows, but only against a uni-directional background flow.
At the representationalist level of description, we see that there is not only spatial but temporal internality; a specific now character of phenomenal content. From the third person perspective, there are no tenses. A complete description of the universe would not involve any information about “now” or an analysis of time as a unidirectional phenomena. The conscious experience of time “possesses an indexical component in the temporal domain.” (10) “Indexicals” are words whose meanings change with the context, like here and now.
This type of content is simulational, in that it is not an epistemically justified form of content; it is not actual knowledge about the world. While we experience ourselves as being in direct and immediate contact with the ‘Now’, “all empirical data tell us that, strictly speaking, all conscious experience is a form of memory.” (10) Representations by models of reality are always presented to the subject as actual information; this form of temporal internality is a simulational fiction from the third-person perspective.
As a functional property, the window of presence allows us to get around the fact that a purely data-driven world-model does not permit explicit predictions in time: “Only additional, recurrent networks will allow for the generation of time-dependent states.” The “Now” becomes the simplest form of time representation, which is a set of recurrent loops along with a decay function. Hence, short-term and working memory are at the heart of any cognitivist or functionalist account of presence.
A broad definition of transparency—only the contents of conscious mental representations are available for introspection, not its “vehicle properties.” But that’s too broad, since sometimes non-intentional and vehicle properties are accessible to introspection. Metzinger says that “Transparency results from a structural/functional property of the neural information-processing going on in our brains, which makes earlier processing stages attentional unavailable.” (11)
With the self-model theory, we are concerned with phenomenal transparency, so unconscious representations are neither transparent nor opaque. “Transparency is a property of active mental representations already satisfying the minimally sufficient constraints for conscious experience to occur.” (11) They are always activated in a window of presence and functionally integrated into a world model.
On the phenomenological level, “What is inaccessible to conscious experience is the simple fact of this experience taking place in a medium.” (11) It leads to the subjective experience of immediacy. Many bad arguments are based on an equivocation between epistemic and phenomenal immediacy, like the fact that the colour of an object appears to be directly given. Individual phenomena can have degrees of transparency, and so can world-models. After a traffic accident, the world can appear dreamlike; this happens in certain stress situations and psychiatric syndromes. The best example is the lucid dream.
On the phenomenal level, these phenomenal representations are transparent, because they appear fixed; an object will remain that object, the paper in your hands will remain the paper. To clarify transparency, we need to differentiate between the vehicle and the content of a representation; “The representational carrier of experience is a certain process in the brain.” The process has nothing paper-like about it, and it is not consciously experienced; it is transparent, you look through it. What you look onto is the representational content, “the existence of a paper, here and now, as given through your sensory organs. This content, therefore, is an abstract property of the concrete representational state in your brain.” (13)
There are two kinds of content. The intentional content is an epistemic state, dependent upon the paper actually existing and the representational carrier actually functioning; when these two requirements are met, you can look through the carrier onto the paper.
The phenomenal content is not dependent upon the existence of the paper. It is solely determined by the internal properties of the nervous system. A hallucination is when you are no longer looking through the representational vehicle, but at it, without this information being available.
As a computational strategy, the transparency of internal data-structures is an advantage for a biosystem which has to operate with limited temporal and neuronal resources. It minimizes computational load since it is synonymous with missing information on that level of processing. Our representation architecture only allows for a very limited introspective access to the dynamics of individual neural events out of which our phenomenal world finally emerges in a seemingly effortless manner.
As a functional property, systems operating under a transparent world model live in a reality which for them cannot be transcended. On a functional level, they are realists. This does not mean they must form certain beliefs; it means that the implicit assumption of an actual world is causally effective. The transparent world-model allows a system to treat information as factual information; it enables the internal representation of facticity.
As soon as a certain degree of opacity is available, a second functional advantage emerges: the appearance-reality distinction can be represented, that is, it becomes an element of reality. The fact that some elements of the flow of experience are representational contents and may be false becomes globally available.
The next step is a differentiated consciousness. Recall how the first three constraints—globality, presentationality, and transparency, gave us the most basic kind of consciousness—the presence of a world, which is the activation of a coherent, global model of reality (globality) within a virtual window of presence (presentationality) neither of which can be recognized as a model generating itself (transparency). But we don’t have a first person perspective yet. It is just an internal model within an organism, and it remains undifferentiated in its representation of causality, space, and time. It would be an eternal now devoid of individuated structures.
So we need to add three more constraints. The fourth, convolved holism, will allow for scene-segmentation, or a nested hierarchy of contents. Dynamicity adds temporal structure. Perspectivalness will create the appearance of a world from the first person perspective.