I wish to thank Antonia Barke and David Chalmers for their all their effort and continuing help with the English version of this paper. I am also indebted to Antonia Barke, Andreas Bartels, David J. Chalmers, Andreas Engel, Guven Guzeldere, Peter Konig, Martin Kurthen, Lars Muckli, Eva Ruhnau and Michael Tye for valuable comments and critical discussions.
[1] I am therefore concerned with phenomenal mental models. In doing so, I am loosely taking up earlier work by Craik 1943 and Johnson-Laird 1983. Craik and Johnson-Laird have used this concept for different purposes, but it is well-suited to serving as a unified starting-point for a PDP-inspired theory of phenomenal representation; cf. Metzinger 1993, especially sections 2.2 and 2.3.1, see also Metzinger 1995b. Following this line of thought, the space of phenomenal modelling would not be identical with the space of representational modelling, rather it forms a partition of this space: the conscious model of reality. A survey of a variety of empirical findings supporting the assumption that the content of phenomenal consciousness is a model of the world, which is used to generate behaviour and which enables the system to simulate and anticipate future events, is given by Yates 1985.
[2] In non-pathological waking states there is a tiny part of the self-model which is not self-referentially opaque' in the way just indicated: Certain of its parts - higher cognitive operations, for instance 'volitional acts' and 'rational thought' - are clearly recognized as self-generated constructs by the system. Perhaps it is for this reason that philosophers have traditionally been most interested in these areas of the human mind: Only these parts enable us to express and understand the self-referential opacity of the rest of the model, because they themselves are not characterized by it. In conscious thought we are - at least in the context of a deliberately induced mental simulation - able to distance ourselves somewhat from ourselves and in this way we become, at a later stage, also conceptually able to grasp the perspectivalness of our conscious space. Cf. Nagel 1986 and Metzinger 1993, 1995b.
[3] Here I point to David Chalmers' important differentiation between the logical and the natural (i.e. nomological) supervenience of consciousness. Cf. Chalmers 1996b.
[4] I am, of coursc, well aware that this is a very vague characterization, but in this paper it fully suffices for my purposes.
[5] Mental simulations are processes in which mental models of possible worlds are activated. Mental imagery and deliberately induced acts of imagination are examples of such processes, as are daydreams and spontaneous fantasies. What we cannot mentally simulate is what we cannot imagine.
[6] The indivisibility of consciousness is a traditional topic of the modern philosophy of mind, and not only Descartes has tried to utilize it for his purposes. Soon Kant turned the synthetic achievement of the unity of consciousness as unity of apperception into a central topic of German Idealism. The phenomenological aspect of the problem has been very clearly formulated by Franz Brentano: The unity of consciousness, as we know with evidence through inner perception, consists in the fact that all mental phenomena which occur within us simultaneously such as seeing and hearing, thinking, judging and reasoning, loving and hating, desiring and shunning, etc., no matter how different they may be, all belong to one unitary reality only if they are inwardly perceived as existing together. They constitute phenomenal parts of a mental phenomenon, the elements of which are neither distinct things nor parts of distinct things but belong to a real unity. But Brentano also made clear that no ontological simplicity follows from this unity: Furthermore, it is necessary to emphasize that the unity of consciousness does not exclude either a plurality of quantitative parts or spatial extension [. . . ] It is certain that inner perception does not show us any extension; there is a difference, however, between not showing something and showing that something does not exist. Cf. Brentano 1973 [1874]: 163-4; 165-6.
[7] See also section 2.4 in the general introduction at the beginning ofthis book and footnote 25.
[8] Cf. Sellars 1963 and also 1965. Texts which I found helpful were Green 1979, Gunderson 1974, Lockwood 1993, Maxwell 1978, Richardson & Muilenburg 1982. A good account of the development of Sellars' philosophical treatment of the problem is given by Kurthen 1990.
[9] The patent carries the number 63471, the full text of the application can be found in Stich 1994: 259.
[10] Sidra Stich (1994: 78) quotes a sentence from Bachelard, in which Yves Klein enthusiastically recognized his own project: 'First there is nothing, then there is a deep nothing and finally there is a blue depth.'
[11] This indicates a second and much more radical possibility of generalizing Sellars' concept of 'homogeneity', namely by applying it to our phenomenal space as a whole: One could treat 'consciousness' (in the sense of a primitive one-place-predicate) as an invariant parameter and not as a background variable which is subject to considerable variations. Two prominent neuroscientists explicitly operating with the field-metaphor while applying it to the global quality of primitive consciousness are Marcel Kinsbourne and Benjamin Libet. Cf. Kinsbourne 1988, 1993, Libet 1994.
[12] Of course this is one of the ways in which philosophers have tried to express the original grain-problem. (Compare the formulations of Meehl 1966:167 and Green 1979: 566 f.) It is interesting to note that if we were to imagine counter-examples with a continuous, flowing change of a certain type of qualitative content, i.e. cases in which we might not be able to discriminate any finite regions in phenomenal space any more, the Sellarsian ultimate homogeneity with regard to the generic trait, e.g. of colouredness, would still hold.
[13] Regarding the theoretical difficulties to which this intuitive 'intrinsicality' gives rise, cf. Joseph Levine's contribution to this volume.
[14] Cf. Lockwood 1993: 288 f.
[15] At this point I am not certain whether this really is a logically necessary condition: Could not duration on the phenomenal level be represented without the presentational aspect in question? I am grateful to Lars Muckli for critical comments on this issue.
[16] Cf. Fodor 1983, Poppel et al. 1991.
[17] Cf Nagel 1974.
[18] Rcgarding the different levels of the binding problem cf. e.g. Koch & Crick 1994: 94 f. or Poppel et al. 1991: 58 ff. See also the second section of Eva Ruhnau's contribution to this volume. It has been assumed that between these levels there may in some cases exist top-down effects. Cf. for instance Ruhnau 1992, Poppel et al. 1991. However, one must not conceive of such downward-processes as if there were a homunculus in the system, a little man pointing the beam of its already given awareness at inner states and thereby turning them into intentional objects: the phenomenal self, the centre of our inner experiential space, must itself be thought of as a naturally emerged representational object, a transient computational module, which the system uses in organizing its behaviour.
[19] Cf Barlow 1972.
[20] Cf. Hardcastle 1994: 66 f., 85.
[21] Meanwhile, a large number of publications on the binding problem and the role of oscillatory activity in the formation of representational objects have appeared. The emergence of the temporal correlations, which are so very interesting from a theoretical perspective, (especially across larger distances) is in the literature frequently associated with thc generation of oscillatory patterns of discharge. The entire EEG-research also rests on precisely this kind of long-distance oscillatory process. On the other hand, the real 'glue' in the selforganization and integration of mental structures consists in the synchronicity produced by the system. The oscillations seem to work as carrier waves or a local mechanism which itself does not play any role in the information processing. Cf. Gray 1994: 17, who provides a survey of the role of oscillatory activity within four different subsystems of the brain. For us philosophers, thejungle of empirical literature has long become impenetrable and difficult to survey. Texts which 1 found hclpful as a general introduction were: Barinaga 1990, Engel et al. 1993, Engel et al. 1992b and Singer 1989b, the classical papers of Crick & Koch 1990, 1992 and the more precise account in Koch & Crick 1994, Crick 1984, Gray et al. 1989 and Poppel 1972; concerning the general role of synchronization phenomena Engel et al. 1992a and c; for empirical evidence concerning the self-organisation of temporally coherent structures, Singer 1989b, 1993 and (containing a wealth of bibliographic references) 1994; for well-founded background information Engel 1994, and also Engel et al. 1991 a, b, c, Podvigin et al. 1992, Poppel & Logothetis 1986, Ruhnau 1992; concerning the qucstion of sensomotoric integration Singer 1995. Meanwhile, the synchronization phenomena in question have been shown in a variety of functional systems, in waking monkeys and also in human beings. Cf. Kreiter & Singer 1992, Kristeva-Feige et al. 1993, Llinas & Ribary 1993, Pfurtscheller & Neuper 1992, Roelfsema 1995. The dreaming state in humans is also accompanied by 40 Hz oscillations, which are very similar to those of the waking state. Cf Llinas & Ribary 1993, Llinas & Pare 1991. A short review of the empirical material together with the sketch of an alternative interpretation on the level of non-linear dynamics is given by Hardcastle 1994.
[22] Cf. von der Malsburg 1981, 1986. The main idea of this theory can also be found in Abeles 1982 and, in an earlier shape, in Milner 1974.
[23] l have borrowed this term from Wolf Singer. The 'escape into the temporal dimension' is in part interesting, because it dramatically increases the number of the functional states available to the system. In Llinas' and Pare's words: A totally different type of functional geometry [. . .] has emerged in which that of temporal mapping, in addition to its spatial counterpart, are important variables... Spatial mapping allows a limited number of possible representations. However, the addition of a second component (serving to form transient functional states by means of simultaneity) generates an indefinitely large number of functional states, as the categorization is accomplished by the conjunction of spatial and temporal mapping. Cf. Llinas & Pare 1991: 527.
[24] Cf Singer 1994: 237.
[25] It is interesting to note that the concept ofa 'time-window' also plays an important role in a completely different theoretical model which, however, calls upon very similar and partially identical empirical evidence such as the correlation theory. The hypotheses developed by Eva Ruhnau and Ernst Poppel, that atemporal zones in the system are generated through phase-locked oscillation processes on a very fundamental level, i.e. states of the system governed by a principle of 'simultaneity', is aimed at a different and more special problem than the correlation theory of feature binding which 1 have just sketched. The opening of time-windows - according to this second interpretation - enables a system to generate an operational time for itself: By quantizing its information processing, it, so to speak, 'swallows up' the flow of physical time on a very fundamental level of its inner representation of the world. It distances itself from its own processuality by carrying out a very interesting form of data reduction. The physical time interval itself still exists, but the content of the respective states of the system loses all or a part of its internal temporal properties: For the system itself representational atoms emerge. These atoms have also been called 'elementary integration units' (EIUs).
Amongst other things, this provides us with an idea about how, on a higher level of processing, the transition from parallelity to seriality could be possible - the implementation of a Dennettian Joycean Machine on a massively parallel system (which, on a theoretical level, is one of the many more recent versions of the mind-body problem). For reasons of space I am not able here to pursue the highly interesting question, of how subjective time experience and, most of all, a phenomenal present can emerge from objective time structures. In the present context, the most important thought of this theoretical approach seems to consist in the idea that a naturally emerged representational system can generate atemporal zones for itself in its depiction of the world. I believe that this theory is very interesting because it can help us to achieve a better understanding of what the element of 'phenomenal presence' described in the first section of this contribution actually is (see also section 2.4 of the general introduction). For this reason, this theory provides a valuable complement to the correlation theory, which on the other hand supplies us with conceptual ingredients for a theory of 'phenomenal wholes'. However, it is presently more than an open question whether the two theoretical models can be projected on to each other at all in an interesting way. Cf. Ruhnau & Poppel 1991, Gornitz et al. 1992 and Eva Ruhnau's contribution to this volume.
[26] This is primarily a question in need of an empirical answer. I have, however, already named the fundamental characteristics of the intended class of systems: They are systems which work in a massively parallel fashion, according to the principle of coarse coding, which activate distributed representations and which can be described as plastic, self-organizing networks.
[27] It thus has to be shown that they are not artefacts or epiphenomena as they always emerge within complex systems. What I have called a 'representational Holon' must also act as a transient functional module in the inner ecology of the system. Meanwhile, a great number of empirical findings demonstrate the functional relevance of phenomena of synchronisation. Cf. e.g. Roelfsema 1995 and footnote 21.
[28] A content property in the sense of Richardson & Muilenburg 1982: 177 f. Cf. also Lycan 1987: 85. It may be that the analysis of homogeneity as a higher-order property could become questionable in the face of the new empirical material. This will be a topic for future investigation.
[29] I am grateful to Andreas Engel for this thought - as for many other valuable and stimulating comments.
[30] This means: a naturalistic solution of the Dilemma sketched by Sellars (1963: 36) is potentially available.
[31] By 'format' I mean a set of abstract properties, discoverable with regard to an active data structure -for example topological properties. Cf. Metzinger 1993; Churchland 1986.
[32] Thus it is not globally available for control. Cf. Block 1995 and Chalmers 1996a.
[33] Cf. Eva Ruhnau's contribution to this volume.
[34] Cf. Singer 1995.
[35] Cf. Metzinger 1993.
[36] In the words of R. Melzack: In essence, I postulate that the brain contains a neuromatrix, or network of neurons, that, in addition to responding to sensory stimulation, continuously generates a characteristic pattern of impulses indicating that the body is intact and unequivocally one's own. I call this pattern a neurosignature. If such a matrix operated in the absence of sensory inputs from the periphery of the body, it would create the impression of having a limb even when that limb has been removed. Cf. Melzack 1992: 93 and also Melzack 1989.
[37] Cf. Conant & Ashby 1970.
[38] Andy Clark has coined this very illustrative and unromantic metaphor, cf. Clark 1989: 62.
[39] This concept of 'confusing' - just like the 'time-window metaphor' - must only be understood as a first illustration! if one takes it too seriously on the conceptual level, one may easily create another transcendental homunculus or an objective self which forms the subject of this ego-illusion. It is also interesting to note that a complex and stable ego-illusion of this kind may have been biologically very successful, because it inevitably makes the system egoistic. However, what makes the emergence of the phenomenal self through entirely natural processes so fascinating, is the very fact of this ego-illusion being no-one's illusion. Cf. Metzinger 1996.
[40] Cf. Baars 1988 and Baars & Newman 1994.
[41] One of the most interesting contributions to the clarification of this problem which I know of is (although I cannot agree with all his points) Chalmers 1996b.
References