Karl Leidlmair, University of Innsbruck
0. Preliminary Remarks
Critics of the AI approach very often emphasize that AI technology is based on several assumptions which give a distorted picture how human knowledge really works. In the field of expert systems above all I want to mention three of them here: representationalism, atomism of meaning and computationalism.
Representationalism is the assumption that it is possible, at least in principle, to elicit knowledge from an expert in the form of explicit rules and to transfer and implement this knowledge on a computer. It is the task of the so-called knowledge engineer to question the expert about his various intelligent tasks in such a way that his expert knowledge can be modeled into a representational system and - given the transitivity of this process of modeling - can in turn be implemented on a computer. In other words, according to representationalism, every knowing-how can be converted into a knowing-that.
The other two assumptions, atomism of meaning and computationalism, just to sum them up, maintain that 'knowledge' can be deconstructed into 'chunks' of determinate and context-free units which in turn can be manipulated according to explicit rules.
In regard to the representational account we can object that it mistakes a model about our use of knowledge for the knowledge itself. We have to make a clear distinction between an explanatory theory as to the activity in our brain and the activity itself - the former can never exhaustively substitute the latter. This difference becomes apparent in situations in which reflection on our skills leads to a drastic decrease in our command of the very same skills.
In regard to the second assumption, i.e. atomism, one can contend that context-free 'chunks' of knowledge emerge only against a background of tacit and transparent know-how with which we must be endowed from the very beginning. Such atomic elements or knowledge-pieces are therefore only theoretical constructs. They owe their existence to a detached mode of contemplation in which the primordial background knowledge has been artificially faded out. For the same reason, it is impossible to reconstruct this primordial knowledge simply by assembling atomic elements which have been formerly deprived of their context.
It is not the aim of this article to belabor all the various aspects of this criticism. Instead I want to outline an alternative view in place of representationalism. First of all, it must be clearly understood that any criticism of representationalism stands and falls with the positive evaluation of the background knowledge which cannot be substituted with explicit rules. Put in other words, we must have an alternative in order to criticize successfully the representational account.
It was Hubert Dreyfus who has given a clear idea of such an alternative view in his commentary on Being and Time. Please note that, although the following ideas are drawn from Dreyfus commentary on Being on Time, they are depicted here in a specific way according to my own preference. So I will take responsibility for all possible misunderstandings.
Following Dreyfus´ criticism of representationalism, two basic questions have to be answered: 1) How can the tacit and transparent background knowledge be phenomenologically revealed? 2) If it can be shown that the representational account is grounded in that background knowledge how could we ever take it into our heads to try and explain knowledge in reverse order? How could it ever occur to us to explain the original from the derived? In short, how can the representational account be explained? In the following I shall answer these two questions, but in reverse order.
1. The Emergence of the Representational Account
Before proceeding with the second question, one preliminary remark may be in order. In our skillful coping we normally do not take notice of the equipment we are dealing with. We are not aware of the availableness of the equipment, instead it has - Dreyfus treats this point carefully in his commentary - a tendency to 'disappear'.
This invisibility of equipment can be made clear by elaborating the distinction of the periphery and the center - as psychology of perception has pointed out. Let me give three short examples. The first is the often-mentioned example of the blind man’s cane. At the beginning - when we hand the man a cane - it is the cane itself with its special properties (e.g. whether it is curved, how heavy it is) which is in the center of his perception. But after a while of getting used to it the stick will disappear from perception and become part of the man’s peripheral perception - like his eyes or his finger tips. This means that the medium must be invisible and inconspicuous in order to fulfil its functional role: in order to reveal the objects which are touched by the cane it ceases to be the center of attention. A second rather old example of the psychology of perception may illustrate that shift of a medium from the center of attention to the periphery of perception: Erismann and Kohler´s example of inverted eyeglasses from some decades ago. If we wear such glasses for several days the perceived surroundings flip into their original position. So after a period of habituation these lenses themselves become part of the inconspicuous periphery. One further and perhaps more familiar example may illuminate this circumstance. When we learn to drive a car we initially focus our attention to the medium by which we drive, e.g. the gearshift (if there is any). But as soon as we become familiar with the medium we will no longer take notice of it, instead we will pay attention to the road, the traffic, the next exit and so on.
What lesson can we draw from this transparency of a medium to equipment itself? Take the example of the blind man’s cane mentioned above: The stick disappears from perception in that very moment when it takes over its functional role, namely to work as an additional sense of direction. Taking over its functional role essentially means to become something available (something ´in-order-to´). Availableness - the ontological character of the available - and invisibility are two related and inseparable aspects of the available. In order to understand this invisibility of equipment thoroughly we must take into consideration that the (unnoticed) periphery through which one thing comes to the center of perception is not an isolable equipment, instead it is a whole peripheral system of available things. For something to function as equipment, there must be a whole context of other equipment, its interrelations and human purposes: a world. A good way to understand this notion is to consider an expression which is commonly used in contemporary German. When we are in a situation in which we have difficulty mastering a task (when we change our job, for example) we sometimes say: This is not my world! With this figure of speech we simply want to say that we are in the wrong context. Similarly, available things are always encountered in a context and it is this context which constitutes the availableness of the equipment. Children, for example, have their own world. The availableness of the things they deal with is quite different from that of the adult. So the world which makes up the availableness of the equipment is by no means a totality of occurrent facts.
Now it is this essential characteristic of the available - belonging to a world - which normally remains invisible and inconspicuous. We begin to reflect on characteristics of equipment only in situations in which it becomes conspicuous. Becoming conspicuous in turn requires that the equipment ceases to be available. The gearshift, for example, moves into the focus of attention only if it doesn’t fulfil its functional role any more. So we take notice of the being of the available - its availableness (which in turn is grounded in its belonging to a world) - only when there is some kind of disturbance in our daily coping with the available. In case of disturbance a new mode of encountering things emerges: When the coping activity is disrupted the equipment is deprived of its availableness and, as a consequence, the sheer occurrentness of the available comes to the fore.
This analysis of disturbance sheds light on two aspects: On the one hand it shows that the ontological character of equipment - namely its availableness - must remain in the background (recognizing it results in the elimination of the same). On the other hand it explains the special situations in which we encounter decontextualized objects.
Dreyfus uses such situations of breakdown to explain where the idea of a detached 'subject' and an isolable 'object' comes in and how this leads to the mistakes of the representational account. Dreyfus' analysis of breakdown shows in particular that encountering objects 'outside of the mind' is by no means the way we deal with things under normal conditions. He writes: "We shall see that there are subjects and objects but that the tradition has introduced them too early in the analysis and, moreover, has mischaracterized them so as to give them a foundational function they cannot perform." Unlike Heidegger, who does not explicitly connect breakdown with the emergence of the representational account, Dreyfus elaborates in minutious detail how different situations of breakdown, juxtaposed as increasingly serious disturbances of skillful coping, bring the subject/object relation to the fore.
In order to correctly understand the significance of such situations of breakdown for the traditional account of intentionality two remarks may be in order: 1) The shift from an involved way of encountering entities to detached reflection must be grounded in additional theoretical considerations independent of experiences we might gather when our skillful coping is disrupted. To encounter entities in situations of breakdown which seem to be only occurrent is one thing, to interpret the directedness of human being towards entities as a relationship between a detached subject and an independent object is another thing. But in order to achieve such a theory we must in a first step encounter entities deprived of their availableness. 2) This deprivation is not an event which already has been terminated; instead the transition of involved skillful coping to detached deliberation is an ongoing process. For the same reason the availableness does not simply disappear in situations of breakdown, it only temporarily takes its leave. So availableness does not vanish completely, it only gets lost during the time of transition when the breakdown takes place.
2. How the Phenomenon of World is Revealed
Now that I have shown how Dreyfus explains the emergence of the representational account we can switch to the second basic question I mentioned at the beginning: How can the whole of the context, on the basis of which every available thing is conceivable, be made explicit? As Dreyfus has shown along with Heidegger - I mentioned this issue above - it is an intrinsic feature of availableness (the ontological characteristic of the available) to be transparent and invisible. So, how can availableness, whose mode of being is grounded in this invisibleness, ever be revealed?
First of all the context whole becomes visible only when our everyday concernful coping is disrupted. There must be a disturbance in the coping activity in order to illuminate the context whole, on the basis of which all our coping takes place. Please note that ´disturbance´ does not inevitably imply that the available has to be destroyed - it simply indicates a rupture in our transparent coping with things. This circumstance may be illuminated when we recall the aforementioned examples of the psychology of perception: Simply by refocussing the medium it loses its transparency which in turn is tantamount to a loss of its functional role. But this access to the background of our practical activity can create serious problems when we compare it with the previously made remarks about the emergence of the traditional philosophy of mind. The question arises: How can the same event which brings decontextualization to the fore simultaneously reveal the context whole? In order to understand this point we have to recall what is going on when we encounter situations of breakdown. Such situations are not events which have already been terminated; instead they are states of transition. So the context whole is illuminated at the very moment when it takes its leave. Taking its leave means still being there - but in the mode of withdrawal.
Let me now sum up the basic point of my arguments. If we accept the last hypothesis concerning breakdown we are confronted with the strange phenomenon that we can become aware of the context whole only at the heavy price of being deprived of it (because the context whole is revealed only by disturbance). Formulated in an oversubtilized way we can elaborate the underlying idea like this: The revelation of the context whole is always accompanied by, and coupled with, a break in that whole which in turn is the starting point of decontextualizing. Metaphorically speaking, we can ask: Must human society, in order to become aware of its 'soundness', first fall into 'sickness'? Does the emergence of peril simultaneously indicate the possibility of its own reparation?
One possible explanation for this ambivalent role of disturbance we can find in the theory of media. It was Havelock, for example, who maintained that the cultural feats of the ancient Greeks are a direct result of the radical change which took place during that one transition period in which their oral tradition was superseded by literacy. When Havelock describes this revolution he makes a comment which is remarkable if we read it in the light of the analysis of breakdown given above. Only in that one short period of radical change when the oral tradition was in its decline did the Greeks become aware of it. It was the very moment of this decline which enabled them to elaborate their oral tradition. Considering the above given examples of the psychology of perception we can say: There must be a revolution of media in order to free hidden peripheries from their transparency and to draw our attention to them. Only in a retrospective view - after literacy has endowed human society with new peripheries - the reflection on (and - at the same time - decontextualization of) the primordial peripheries of the oral tradition could begin. So the breakdown of orality contains two elements - gains and losses.
There is no doubt that the new electronic media too can be interpreted as such a revolution of media in which new peripheries are reframed. It would be a courageous speculation to ask whether this revolution might evoke a new cultural stage of mankind. What could be the benefits and losses of a radical change like that? What new awareness of our Being-in-the-World might we gain and what new kinds of decontextualization might take place?
Hubert L. Dreyfus, Being-in-the-World, A commentary on Heidegger´s Being and Time, Division I, Cambridge, MA: MIT Press 1991
Hubert L. Dreyfus & Stuart E. Dreyfus, Mind over Machine. The Power of Human Intuition and Expertise in the Era of the Computer, New York: The Free Press 1986
Hubert L. Dreyfus, What Computers Can’t Do, New York: Harper & Row 1972; revised edition 1979; 3d edition: What Computers Still Can’t Do, Cambridge, MA: MIT Press 1993
Eric, A. Havelock, Preface to Plato, Cambridge, MA: Harvard University Press 1963
Eric A. Havelock, The Muse Learns To Write: Reflections on Orality and Literacy from Antiquity to the Present, New Haven: Yale University Press 1986
Martin Heidegger, Being and Time, New York: Harper & Row 1962
Ivo Kohler: Über Aufbau und Wandlungen der Wahrnehmungswelt, Sitzungsberichte der Österreichischen Akademie der Wissenschaften 227, Wien 1951.