Metzinger

Good morning to everybody. I want to ask what a conscious self is and what a first person perspective is this morning. And another way to ask this question is to ask what makes a phenomenal state a subjective state. And you see the answer already on the bottom. My answer at the end will be the phenomenal model of the intentionality relationship. And this is what I want to briefly explain to you this morning. I only met Francisco two times in my life, but I will three times in this lecture point to connection points we had in our work.

But most of our life we have known each other only indirectly. So can I have the next slide? And one more, one more click.

So Francisco says, I guess I’ve had only one question all my life. Why do emergent selves, virtual identities pop up all over the place creating worlds? Whether at the mind body level, the cellular level, or the trans-organism level. And I’m a self-consciousness person too, but I claim that no such things as selves do exist in the world.

And let me talk a bit about this today. So what is a phenomenal first person perspective? There are three phenomenological target properties which we have to understand. The first property is the phenomenal property not of mindness, but the felt sense of ownership.

That is a higher order property of particular forms of representational content. And here are some examples how we speak about this form of mental content when referring to it from public space when we talk. We say things like I experience my leg subjectively as always having belonged to me. I always experience my own thoughts and my emotions as part of my own consciousness. Or voluntary acts are initiated by myself. The second target property is actually the core property. It’s the property of selfhood. Some old fashioned German philosophers call this prereflexivisatz, free reflexive self intimacy. I saw yesterday that old fashioned bad German philosophy is very popular here.

I was appalled by this. And this is actually the core property. Here are some examples how we speak about the phenomenal core property of selfhood. We say things like I am someone. I experience myself as being identical through time. The contents of my phenomenal self consciousness form a coherent whole. Before initiating an independently of any intellectual operations, I’m somehow a priori already directly acquainted with the contents of my self consciousness. The third property we have to look at is the property of perspectivalness. That is a global structural property of our conscious space as a whole. And that property is that it possesses an immovable center.

Now here is the problem. For each of you in the room it is true that you are the center yourself. To be phenomenally aware means to possess an inward perspective and to take this perspective on in your experience of the world. And the experience of your own mental state. But there is a philosophical puzzle there. I can describe things like that from the third person perspective. But the moment you do justice to your own phenomenology for each individual person it is true that you are the center yourself and you do a weird flip into the first person perspective.

And we don’t know what is happening through that flip. Okay. What I want to do today is offer you a brief analysis of these target properties on the representational and functional level of description. I want to go from the phenomenological to the representationalist level of description. And my first step will be to introduce a new theoretical entity, the PSM, or the phenomenal self-model.

At the very end of my talk I will very briefly introduce a second theoretical entity. What is a conscious self-model, a PSM? It is the instantiation basis of the phenomenal properties to be explained on the representational level. It is the set of properties you have to have to get the conscious experience of being someone. It is an episodically active representational entity, the content of which is formed by properties of the system itself. A self-model is a model of the system in which it occurs as a whole.

From a logical point of view it is very easy to define what a self-model is. There are some information processing systems that simulate parts of the world. Think about in the meteorology department a large computer that simulates weather. There are some information processing systems that also emulate other information processing systems. That is, they are non-observable internal functional properties, they are information flow. Now if you have a system, if target system and simulating emulating system are identical, you have the case of self-modeling. So what I am saying is that all of you right now are systems that simulate and emulate themselves for themselves. My background assumption of course is that that self-model processes a true neurobiological description.

For instance as a complex activation pattern in the human brain, but I will say not very much about this today. There is a conscious and unconscious self-model. The phenomenal self-model is that part of the mental self-model which is currently embedded into the highest order integrated representational structure, your global model of the world. So the content of consciousness at any one instant is the content of a global model of the world. And that partition of your self-model that is integrated into that is what you now experience as your conscious self.

The phenomenal content of the self-model supervenes locally. So that determines your conscious experience, not your interaction history, not your knowledge. But what determines your conscious experience is determined once all your internal properties and your contemporaneous properties are fixed. The phenomenal self-model in our own case is a plastic structure. It is multimodal, many different proprioceptive modalities feed into it.

And possibly it evolved from a partially innate and hardwired model of the spatial properties of your body. There are some philosophers like Brian O’Shaughnessy who have spoken about the long-term body image. Of course there is a Damazio and Nelsack story.

Marcel Kinsburn has been writing about this a little bit about this later. An active self-model is not a little man in a head. It’s a sub-personal functional state. It plays a specific causal role and from a purely analytical philosophical perspective it’s a discrete set of causal relations. So if anybody here still believes in a computer metaphor of mine, just to illustrate under a classical cognitive description it’s a transient computational model that gets switched on and off if the system has to organize its integrated, regulated behavior, regulated interaction with the environment and has to achieve sensory motor integration.

So the basic idea is when you wake up in the morning, that is the moment when the system which you are switches on its PSM. Because you have to go, say, to the refrigerator or to the toilet, you have to achieve sensory motor integration and the moment where the system activates a coherent self-model on the level of global availability, that is, I would claim, the moment when you come to yourself. There is something that philosophers call teleo-functionalism, a teleo-functionalist background assumption. I would say that the development and the activation of this computational model has a purpose in our own case.

It plays a role for the system. And can we go on? Here comes a 15-year-old Andy Clarke quote that already tells you how unromantic this theory of self-consciousness is going to be. The functional self-model possesses a true evolutionary description as well. It was a weapon which was invented and optimized in the course of a cognitive arms race. There are a number of other philosophers who have in the past written along these lines and I think it will be very illustrative to treat the human self-model as something with a biological history which has been selected because it had an advantage for the system.

So one perspective is to look at it as a weapon. So the functional basis for having something like a conscious first-person perspective, for having the experience of someone, is a specific cognitive achievement and that’s the achievement to open centered representational spaces. And there must have been a time in evolution where that property occurred for the first time, where the first systems were not only able to have coherent global models of the world, but centered models of the world, centered around a self-model.

I will give you two low-level examples of what I mean by a self-model. For astronauts, if they are in space for some time, often get difficulties in feeling where up and down is, where their body access is. This is very unpleasant when they have to go to the toilet, when they eat. Every astronaut knows how to help his body if he has that problem. You just hit them on the head and immediately it locks in and you have the feeling this is up and this is down again. Every astronaut also knows how to annoy, sorry, you hit them on the, I mixed the joke up. You hit them on the sole of your foot, everybody, every astronaut knows how to annoy his body.

They hit them on the head right afterwards and they flip again. What that shows, I just got confused, what that shows is that the human self-model, the bodily model, is a virtual model. And it’s highly context-sensitive, if it is under-constrained by external input, it becomes very malleable. So we see that already in that situation, we see that it is something that depicts a possibility. You experience it as a reality, but it’s actually only the best hypothesis the system has about its own current state.

Many of you may have heard about the experiments of Haramachandran in San Diego. Probably everybody knows what a phantom limb is after amputations. You get phantom limbs, well, some patients do not have the normal rehabilitation history.

Normally, if your arm would be amputated after the surgery, you might feel a hand and an arm, you’re able to clutch a fist, which is not there. And within a couple of weeks, you will have a so-called telescoping effect. That phantom limb will vanish, at the end you will be able to make a fist in your stump, and then it will disappear. You may have some remapping to the contralateral cheek, even we’ve heard something about that yesterday. Now there are some people, especially after traumatic amputations, which have a paralyzed phantom limb which never goes away.

And sometimes it even hurts for 12 years or so. How do you treat a patient who feels pain in a non-existent limb? What these researchers have found out is that the first thing you have to do is you have to bring the paralyzed phantom limb under voluntary control, else it will never disappear. So how do you bring a paralyzed phantom limb under control?

The experiment, Haramachandran and colleagues in San Diego developed as follow. You take the patient and you have something like a box, you don’t even need the box like in this slide with holes, and you say, okay put your good arm in there and you find phantom limb with a stump in there, and the task is make butterfly-like movements with your arm. Now what is your conscious experience? And then the patient says, oh doctor, what is my conscious experience? My good arm moves, my phantom limb is paralyzed. Then you put a mirror down in the middle and say, okay can you please do the same thing? Do the same task and look at the mirror from the right side. And instantly the patient will exclaim, doctor, doctor, I can move my phantom limb for the first time in 12 years.

You say close your eyes or take the mirror out and the patient is very disappointed and says, oh it’s frozen again, I can’t do anything. What moves in that experiment is what I call the phenomenal self-model. So technically speaking what you have done is you’ve installed a virtual source of visual feedback in exactly that part of state space where the system is sending its motor commands all the time and is not able to move its arm. But then you install visual feedback through the symmetric mirror image and then the virtual model becomes controllable as well. Okay, those were two examples. Now how can we analyze mindness, the sense of ownership?

The story is simple. All representational states which are embedded into the current, the active self-model, gain this higher order property of mindness. This non-conceptual sense of ownership which also a non-linguistic creature like an animal could have. We’ve heard some examples.

We’ve seen some slides with these maravita and uraki stuff and you talked about if you have monkeys. And you give them a rake. We all know this expert skiers extend their body image to the tip of their ski. A good car racer doesn’t have to look if he fits through there because he just feels it. Whatever we integrate into our self-model is what we perceive as being belonging to ourselves. So if this integration process into the self-model is disturbed, different neuropsychological syndromes or altered states of consciousness result.

Let me give you a number of examples again. So in Floyd’s schizophrenia you have the well-known problem that consciously experienced thoughts are not the patient’s own thoughts anymore. The natural hypothesis is he cannot integrate its own cognitive processing into self-model anymore. My own leg is not my own leg anymore. You have that in unilateral hemineglect.

People are not able to experience their leg as their own leg. People will fall out of bed. The nurse will come after surgery.

They will ask what happened. I said I woke up and I found a dead leg in my bed and I thought what is this? The medicine students have made a particularly bad joke or what is it? I tried to throw the leg out of bed and I realized it was grown to my hip and I fell out of bed and then the nurse asked, so if that was true you must have three legs. Can you please count your legs? And the patient says I don’t understand it.

The only thing I know is that this is not my leg. So also on the spatial model of the self you can have this integration process disturbed. Alien hands in Rome in which patients perform bold directed actions without conscious control is particularly interesting. You will have a situation for instance after limited tumors in the corpus callosum only where somebody will button down his shirt with one hand and the other hand button that closes it again.

Often with a sense of inter-manual conflict the patient will try to receive a telephone call and the other hand will terminate it. The interesting thing about this is that the movements, the motor patterns are clearly goal directed and still the conscious experience is I did not initiate this. Everyday people turn themselves in at emergency wards in psychiatric hospitals and say things like this, I am a robot, I am transformed into a mechanical puppet, volitional acts are not my own volitional acts anymore. You have this in severe depersonalization and you can lose what Karyaspas called executive consciousness for its sub-stavos sign. The experience of not only initiating your motor patterns but of carrying them through. On the other hand you have also situations where the self-model so to speak dissolves and it’s pants to the boundaries of the world model. You have phenomenal state classes in which people say things like I am the whole world, all events in the whole world are controlled by my own volitional acts.

A couple of years ago I gave a lecture in the Psychiatric Institute in Nidalaag in Frankfurt and they had two patients and one of them was standing by the window all day making the sun move. It was taking charge and this has to be taken seriously. The conscious experience is one of constantly causing and sustaining this movement.

The other patient was looking down into traffic turning the lights on and off making the puppets walk and letting the cars drive. So these phenomenal states and a good theory of consciousness has to explain these states are one in which any event that happens is caused by myself. Everything is like in myself. Now what is selfhood? It is the existence of a single coherent and temporarily stable self representation forming the center of the overall representational state. Again, if this module is damaged, if it disintegrates or if multiple structures alternate within the system we find different neuropsychological syndromes or altered states.

The first example is anosognosia and anosodiathoria. You all know that after certain kinds of brain deficits, patients lose insight into the deficit. They lose the higher order insight, for instance, in blindness denial. Many of you have probably heard of Anton syndrome, where people are practically blind, they don’t see, they will never see, and they claim to be seeing persons. They show all functional signs of blindness.

They bump into furniture, they fall. If you ask them how many fingers are these, they start to confabulate and say, I don’t care about your experiments. Kind of dark in here, and she takes keys and shakes them and says, what is this?

They say, yes, keys. So they start to confabulate and they keep on claiming for minutes, hours, sometimes weeks that they are still seeing persons. And then I don’t want to go into this deeply, but you probably have all heard about what was formerly called multiple personality disorder, dissociative identity disorder. I just want to sketch that under this approach, you can of course understand situations where a system creates some kind of emotional division of labor. A system could use different self models in different social contexts. And to a certain degree, this may also be healthy. But it may be also a process that gets out of control, like in dissociative identity disorder, where you suddenly have multiple personalities popping up with amnesia in between in these patients. This can be nicely representationally analyzed in the framework I’m proposing. So here’s this German word, Ichstörung, which does not nicely translate into English.

Identity disorders like delusional misidentification, there’s a large class of psychiatric disturbances, which are characterized by a dramatic deviating form of your conscious experience of your own identity. Can we go on? So what is perspectivalness under this approach I’m proposing? This is the existence of a single and stable, temporally stable coherent model of reality, which is centered around or on a single stable phenomenal subject that is around a model of the system as currently experiencing as being directed.

More about this later. This model is what I call a PMIR, the phenomenal model of the intentionality relationship, a model of the arrow pointing from subject to target object. Now if we have this structural feature of the global representational space, then we will have a temporally extended and non-conceptual first-person perspective. Any animal or artificial system that would have that would have a consciously experienced first-person perspective.

I’ll say a little bit more about this later. Again, if this representational property is lost, the phenomenology changes in different neuropsychological syndrome are altered stage result. So what are the state classes, phenomenal state classes in which we don’t have a consciously experienced first-person perspective? The first big cluster is complete depersonalization. You can lose the conscious first-person perspective and this can be accompanied by dysphoric states and functional deficits. There are some psychiatrists in Zurich who have run a gigantic statistical battery and found out etiology independent, what the clusters of the phenomenological types are, no matter how you induce a disturbance. And one very statistically significant cluster is what they call Drenfu ego dissolution, Angst volle Ich-Auf-Lösung. Now it’s important to note that there’s a second set of global phenomenal states in which the first-person perspective is lost, namely mystical experiences. Of course, there are selfless and non-centered global states which are being experienced and described as non-pathological and non-threatening.

And these people who statistically evaluated all the material called that cluster, oceanic boundary loss appealing to Romain Roland, maybe some of you know this, oceanische selbst und Grenzeln. So these are real possibilities if you lose the perspective. Now what is the central theoretical problem on the functional level of description?

And here is where I want to skip a bit in the interest of time. The first important question you have to ask, okay, okay, there is something like a phenomenal self model, but how does it differ from all the other models active in the system? Why is not this cup the self? Which functional property characteristically marks it out? How precisely does it become the center of the phenomenal space of representation? And I will very briefly give you my answer to this functional level question now. The self model in the human case is the only representational structure which is anchored in the brain by a persistent functional link, that is by a continuous source of internally generated input. That is what anchors it in a continuous causal chain.

And I will, let’s just move on. My proposal is that whenever conscious experience exists at all, that is whenever you have a stable integrated model of reality, this internal source of continuous proprioceptive input exists as well. And there are two major candidates for this right now. There’s a third one in managing. The first empirical candidate is the hardwired part of the matrix of your body image.

Can we just move on? There is the first empirical hypothesis results from new results, comes from new results from McGill University, Ronald Meltzack concerning pain experience infantum lens. And these people think there is a genetically determined neural matrix, the activity pattern of which could be the basis of the more invariant part of your body’s self consciousness. They call this the phylamatrix of the body image. So the idea is that there is a genetically pre-specified part of your spatial body image which is incessantly active, has autonomous activity. And right now while you’re listening to me, this would be this part of your body feeling which is so invariant that it’s almost unconscious.

That part which does not depend on motion, that part which was not acquired after birth and shaped through your experiences. This is under discussion, this is disprooted. I’m just reporting this to you.

So let’s just move on. The second empirical hypothesis is a story most of you will know. I call it the emotional embodiment story and it is of course Tony DiMaggio’s theory about the role of homeodynamics as regulated by upper brainstem and hypothalamus.

Here is where we skip along the Meltzack quote, but here is something I just want to present to you. In Peter Brogast’s lab in Sury Switzerland they have this patient. This person is 36 years old. She has never had arms or legs. She was born without arms or legs. She’s an academic, cognitively very usage and they have asked her she has phantom limbs. Well she has never had a source of input ever. Ask her to rate how real her phantom limbs are.

That is in the torso and the stumps. You have a 7 points care from 7 to 0. 0 is no awareness.

7 is as realistic as the body that exists. And you see there’s a great variance. For instance here in the hand you will have a value of 5.0.

Here you have 4.3. That is so to speak the phenomenological realness that the patient reports. In the toes you have 1.1, 2.3, 2. So I will not take a position if there is an innate nucleus of the human self model. This is just to illustrate an issue for you that’s in the background.

Is all of us acquired after birth or is there something like a nucleus around which the spatial model grows maybe through body babbling according to Andy. Here’s the Damaggio story and we skip that as well. And I will now come to the end of my talk. I see I have 15 minutes still. And ask what is actually the central theoretical problem on the representationalist level of description. I think the major philosophical problem is this. There seems to be no necessary connection from the functional and representational basis properties of the phenomenal to the phenomenal target properties of ownership, selfhood and perspectivalness. So the question is somebody like Dave Thomas could come along and say all this could conceivably take place without the emergence of a genuine phenomenal self or a subjectively experienced first person perspective.

One can imagine biological information processing systems which develop and use such centered representational spaces but without the emergence of a true self. Somebody could say come on Thomas you’re cheating with this world self model. A self model is just a model of the organism as a whole. It’s not a self but only an internal representation of the system itself actually it’s only a system model. So the question is how does one get from the functional property of centeredness of being anchored in continuous internal input and the representational property of having a coherent self model to the phenomenal property of being someone of selfhood.

Here is my answer. Transparency of the data structures used by the brain. Now here to the experts this is a notion of phenomenal transparency. This has nothing to do with referential transparency and the intentional fallacy and analytical philosophy of mind. It also has nothing directly to do with the notion of a semantically transparent system in connectionist theory. It is a phenomenological property.

So what is transparency? Only content properties of the representational structures used by the brain are introspectively available. The vehicles, the structures in the brain employed by the system are transparent that is they do not represent the fact that the representations on the level of their content and I claim this is the situation in which all of you are now. Like this thing is little pocket computer representation is transparent for you because try as hard as you may you will not be able to experience this as a representational state in your visual system.

It’s so fast it’s so flexible it’s such has such a high temporal resolution you just get the end result this little black thing. So you cannot experience it as a representation and to become an e-realist you actually think you’re directly in contact with your environment right now. This has a long story in philosophy there’s a paper by Moore 1903. The general idea is that we are a system that looks through their own representational structures as if they were in direct and immediate contact with their content. So my first empirical hypothesis is that the data structures are being activated so fast and are so reliable that the system cannot recognize them as such anymore and on a very small time window because of the temporal resolution of meta representational functions looking on it. That’s what I’m saying is that for us earlier processing stages are not available to attention you don’t have to be able to think and to speak for that animals can have that too.

So you cannot experience this as a representation because you can direct your attention as hard as you want to the construction process is so fast that you cannot experience the construction process you only have the content okay. The second empirical hypothesis works on a much larger time window. I would propose that there has been no evolutionary selection pressure on the relevant parts of our functional architecture. Naive realism has been functionally adequate background assumption for systems like ourselves. It was just not necessary to pay the additional metabolic price to go to the effort to also know that this is a representation occur in you. For the evolutionary purposes it was enough to experience the fact that there is this object out there.

So we are naturally born naive realists. Now the important the core point of my talk is that you have applied this insight to the human self models. I’m saying we are systems which are not able to recognize their own sub symbolic self model as a model. Therefore to coin a metaphor we operate under the condition of what I call a naive realistic self misunderstanding. We necessarily experience ourselves as being in direct and immediate epistemic contact with ourselves. Now notice if this point I’m trying to make today is true then I’m committed to a certain prediction and the prediction is that if everything would stay the same and your self model would become fully opaque if your attention or capabilities would become so good that you could recognize that this is an internal model then the phenomenal property of self would by necessity disappear.

This is an empirical prediction I have to make under this model and here comes my second connection to Francisco. Quoting photo of Arayla Thompson and Rush first write in intellectual history everything happens twice first as philosophy and then as cognitive science and they go on to say it is our contention that the rediscovery of Asian philosophy particularly of the Buddhist tradition is a second renaissance in the cultural history of the West with the potential to be equally important as the rediscovery of Greek thought in the European renaissance. Okay we are almost there three more slides at the beginning I asked what is subject to the experience. I will now introduce a second theoretical entity but I will not explain it at length. The second theoretical entity is what I call the PMIR in the beginning we had the PSM the phenomenal self model the PMIR is the phenomenal model of the intentionality relationship. I claim there is a point many theoreticians have completely overlooked there is not only the theoretical problem of intentionality since Franz Frantano 1874 which has a long history in philosophy there is an equivalent in conscious experience itself in non-pathological standard states we experience ourselves as being directed to object components to a reaching target in the world to some cognitive content to another human being. There is actually a very dynamic representation going on in which there are constant couplings on the level of phenomenal representation between the self model and whatever it attends to what it targets with its motor behavior or what it grasps in thought so the actual the core content of consciousness is not never a thing but a relation it’s a dynamic as Damasi would say self in the act of knowing and the PMIR is the representation of this a dynamical and transparent model of a self in the act of knowing a continuously changing inner representation of ongoing subject-object relations that is for instance the relationship between subject and perceptual object or between subject and internally represented action goal these are two examples so let’s move on to the next slide what I would claim is that if the phenomenal model of one’s own perceptual states contains a transparent representation of their causal history then convolved global states result the content of which can only by the system itself be described in the following manner and we go to the very last slide I myself it’s the only thing the system can say about its conscious state that is the content of the currently active transparent self model and seeing this object that is the content of a transparent object representation and I’m seeing it right now all this is embedded into a virtual window of presence it’s a content of short-term memory it’s in a phenomenal now and I see it with my own eyes and this with my own eyes is of course totally false you don’t see with your own eyes right that is just a simple story about direct sensory perception which suffers for the evolutionary purposes of the brain so the general idea is that the phenomenal self is something like an avatar or something like a virtual model and that model doesn’t know it has a visual cortex it sees with its eyes because that’s the story it tells to itself on the level of global availability and that avatar doesn’t know it has a premotor cortex or something or an sma it just acts with its hands and that is how the story is dynamically described on the level of conscious experience so the answer in the nutshell then is to have a phenomenal first person perspective is to have a transparent phenomenal model dynamically changing all the intentionality relationship and let me close with a last thing I learned from Francisco when I last met him he said Thomas this is all very nice and I agree with most of your ideas but you will never get anywhere as long as you cling to this old-fashioned representationist nonsense thank you