Материал: искусственный интеллект

10 M. Bitbol

approaches (such as physicochemical examination of cellular components), and global approaches, such as those concerned with homeostasis, behavior and intentions. If these two approaches are incompatible, then it becomes absurd to pretend to absorb one of the objects into the other, as reductionism seeks to do by excluding the categories of homeostasis and intentionality in favor of molecular concepts alone.

•The third region, finally, is one where one is interested in symbols capable of guiding not only the process of knowledge, but also life in general. This is the region of art with its formative figurations, of religion with its representations and archetypal narratives, but also of the many social practices in which an institutional symbol (fiduciary money, constitution, bill of rights) is treated as if it were an autonomous reality.

At a higher level of reflection, however, the very existence of a multiplicity of regions of knowledge is su cient to express the impossibility to separate knowledge from the processes it uses. Heisenberg thus pointed out that “It may not be possible, in a complete description of the connections of a region, to disregard the fact that we ourselves are part of these connections.” It is true that the examples of this impossibility of ignoring our commitment in the (law-like) connections which constitute a region of knowledge are found by Heisenberg in atomic physics and psychology (in his second region of knowledge). But the wording of his sentence implies than our commitment can no longer be ignored when we seek the completeness of any description whatsoever. Our commitment becomes obvious as soon as we strive to reach the boundaries of a region of knowledge, no matter which region we explore.

Many historical cases illustrate this self-revealing power of the exploration of confines. Thus, a search for completeness of the classical description of atoms and light, at the turn of the nineteenth century and the twentieth century, has led us to run up against a fundamental limitation of objectification (in the sense of complete “detachment” of an object with intrinsic properties): the limitation manifested in quantum theory. Earlier in history, a search for completeness in classical science, a science elaborated from the point of view of a detached spectator, made us stumble on the enigma of free will. As suggested by Kant, this enigma that is formulated from the point of view of the spectator could be sorted out only from the point of view of the agent.

This repeated figure of human beings meeting their own limitations at the end of a quest of knowledge has been expressed by Eddington in a celebrated sentence: “We have found a strange footprint on the shores of the unknown. We have devised profound theories, one after another, to account for its origins. At last, we have succeeded in reconstructing the creature that made the footprint. And lo! It is our own” [10].

Heisenberg made another suggestion that may help the development of a detailed relationship between quantum mechanics and the human sciences. His additional suggestion is that the impossibility of completely ignoring our own contribution can be recognized in the mathematical formalism of quantum

physics itself. To understand how this is possible, we must first clarify the process by which the search for objectification overcomes obstacles. This is usually done by incorporating this obstacle into the newly defined objects and turning it into an advantage. As Heisenberg pointed out, “(...) even if a state of a airs can not be objectified in the standard sense, it remains that this fact itself can (...) be objectified in its turn and explored in its connection with other facts” [17]. In other terms, the limits of objectification are (reflectively) objectifiable. This remark illuminates the status of the mathematical formalism of quantum theory. The mathematical formalism of quantum theory is a second-order objectification of the impossibility of any first-order objectification. The first-order objectification consists in extracting spatio-temporal invariants such as classical material corpuscles with a trajectory in space and in time. It thus directly organizes the individual experimental phenomena observed in the space and time of the laboratory. But the second-order objectification is elaborated out of the statistical distributions of these experimental phenomena. It extracts an invariant predictive tool such as the state vector in a Hilbert space, thus organizing phenomena indirectly.

What is connected through the basic law of quantum mechanics, namely the Schr¨odinger equation, is then the state vector, rather than the individual experimental phenomena. This further expresses: (i) the impossibility of a first-order objectification (that is to say the impossibility to detach experimental phenomena from their experimental context), and (ii) the e ectiveness of the secondorder objectification of an invariant generator of statistical distributions. We thus realize that a considerable part of the quantum formalism has a deep meaning that extends far beyond microphysics. This formalism actually expresses a universal general epistemological situation: a stepping up of objectification in the face of the fundamental obstacles that it encounters.

Subsequently, several philosophers have taken advantage of these pioneering analyzes of the creators of quantum mechanics and elaborated a new conception of knowledge on that basis. This is the case of Karl-Otto Apel, who devoted a book to this endeavor. Starting from the two di erences between natural sciences and human sciences mentioned by Habermas, Apel shows that they vanish in the quantum paradigm. The first opposition, between the (natural) sciences of detachment and the (human) sciences of commitment, collapses from the outset: “In the [sciences of nature as in the human sciences], writes Apel, it is necessary to give up the representation of an objective external world of which a multiplicity of perspectives falls, in principle, under (...) theoretical control. Instead, there are aspects of the world that are incompatible, complementary, [because they are indissolubly linked to each perspective, to each mode of intervention]” [1]. The aspects of nature seen in microphysics, “(...) are objectively incompatible and, for this reason, comparable to the mutually exclusive worldviews of the (human) sciences (Geisteswissenschaften)”.

The second epistemological di erence mentioned by Habermas between the natural sciences (supposed to describe actual properties) and the human sciences (supposed to focus on potentialities) is also analyzed and criticized by Apel.

12 M. Bitbol

Indeed, this di erence was canceled out by the advent of quantum physics. Just as the human sciences, quantum theory manipulates symbols (the state vectors) that describe a potentiality rather than some actual event. It serves to anticipate what can happen in the future in various experimental contexts, rather than context-independent present findings. Apel attempts to identify the reason for this shift from classical physics to quantum mechanics: “(Quantum mechanics), he writes, succeeds in separating the subject from the object in the statistical explanation of the behavior of sets of particles, but it fails at the level of the individual particles”. In other words, the use of probabilities in quantum physics is the mark left on the limit of first-order objectification, and the expression a second-order objectification: the indirect objectification of statistical distributions of spatio-temporal phenomena, rather than the direct objectification of a set of spatio-temporal entities.

This being granted, the modalities of the connection between quantum physics and the human sciences are of two quite distinct types. On the one hand, we can qualitatively develop the similarity of the epistemological situation between microscopic physics and each human science considered separately. On the other hand, we can seek to state the quantitative or at least formal consequences of this type of epistemological situations.

4Qualitative Parallels Between Quantum Theory and the Human Sciences

As we have just seen, qualitative parallels between quantum theory and the human sciences can be based on the Bohrian concept of complementarity. However, complementarity being a “broad-spectrum concept” (Putnam), its modalities of applications can vary a lot. Already, in quantum mechanics, several variants of this concept have been listed by Bohr.

Taken in the broadest sense, Bohr’s complementarity expresses the impossibility of getting rid of the holistic features of experimentation. But these holistic features manifest in three di erent ways:

•The mutual exclusivity of two variables that are inseparable from experimental contexts. An exemple of such pairs of conjugate variables is position and momentum.

•The mutual exclusivity of two pictures, respectively associated with partial experimental contexts and a global experimental context. This is the case of the corpuscular and wave pictures in the double slit experiment. The corpuscular picture is associated with the partial context of detection of the passage of an object through one slit ; and the wave picture is associated with the global context of indistinguishability of the paths corresponding to the two slits.

•The mutual exclusivity of potentiality and actuality. They correspond respectively to: (i) the context of forecasting future measurement results after the initial preparation, and (ii) the context of the final measurement. This latter

pair is represented in quantum theory by the continuous and discontinuous modes of evolution of the state vector. The “causal” mode of evolution by the Schr¨odinger equation excludes the “acausal” mode of evolution associated with an act of experimental localization in space-time.

Two examples of qualitative use of the Bohrian concept of complementarity in the human sciences will now be developed. They manifest the same amplitude of variation around the common theme of the holistic character of knowledge, as in quantum physics.

Klaus Meyer-Abich [21] developed a psychological variety of complementarity, in the spirit of Bohr and H¨o ding. According to Meyer-Abich, this kind of complementarity reflects an incompatibility between the act of aiming at objects and the act of reflecting on objectification. This incompatibility was especially highlighted by Kurt Goldstein in his empathic observations of patients who suffered brain damage during the First World War. For such patients, the psychical attitudes of intentionality and reflexivity are so dissociated that even their succession becomes impossible. Surprising as it may seem, this alteration observed by Goldstein is not a consequence of certain focused lesions, but is found in virtually all patients with extensive lesions of the cerebral cortex. In every patient of this kind, “everything that forces him to go beyond the sphere of ‘actual reality’ to reach what is ‘simply possible’, brings a failure” [12]. Patients adhere to what is immediately experienced, without being able to distantiate from it and without being capable of embedding it into a representation. They remain bound to intentional directedness towards objects without being able to step back and acquire a reflective knowledge of themselves. In a later reflection, Goldstein characterizes this deficit of patients having cortical lesions with words that were also used by Bohr: patients “act in the world instead of thinking of it or talking about it”; in other words, they become pure actors because they have lost the degrees of freedom that would have allowed them to behave as spectators of themselves as well. From this, one may infer that in organisms, the actor-stance, the stance of self-adherence to oneself, is fundamental. By contrast, the complementary stance of a detached spectator: (a) can only be incomplete, and (b) requires resources in excess of that of the actor. Holistic integration imposes the actor-stance. It allows only incidentally and fragmentarily the stance of a detached spectator, which supposes that one suspends for a time the fundamental actor-stance.

The analogy between this cognitive pathology and the complementarity of corpuscular and wave representations is striking. In the latter type of complementarity, the corpuscular representation prevails in the context of a local detection on one branch of the interferometer, whereas the wave representation is relevant in the context of an evaluation of the e ects produced by the interferometer as a whole. The corpuscular representation prevails when only one path is available, while the wave representation prevails when all possible paths have been left open. Similarly, Goldstein’s pathological configuration highlights a type of complementarity wherein the local approach of an act is exclusive of a global approach of action. It also corresponds to a duality of attitudes between adher-

14 M. Bitbol

ence to a certain act, and the reflective distancing that allows every possibility of acting to be displayed before the eyes.

A di erent variety of complementarity, that still manifests the holistic nature of knowledge, is mobilized by Michael Rasmussen’s reflection on linguistics [24]. Rasmussen confronts Bohr with another great Danish thinker who is almost contemporary with him: the linguist Louis Hjelmslev. However, his comparison of linguistics with the epistemological configuration of quantum physics is not limited to one author; it extends to any structuralist conception of language, such as Saussure’s. The comparison is established in two steps.

The first step is a definition of observation: observing means restricting the initial conditions on the basis of which a prediction can be made. The question of the impact of the observation on the observed domain can then be replaced with a request concerning the impact of the forecast on the forecasted events. But, says Rasmussen, in linguistics, this impact is by construction considerable. When a linguist tries to predict the future evolution of her own language, she modifies it by her very act of forecasting. For, as a speaker of her language, the linguist is bound to give a normative or prescriptive value to her prediction. When she foresees the future state of her language, she prescribes a condition of identity (this future language must still be “English”, despite all its transformation). And since her speech is guided by such prescription, she influences the evolution of her own language. To sum up, the prediction “disturbs” the language. The work of the linguist influences the evolutionary dynamics of her own language. Linguistic analysis cannot be detached from the metabolism of language.

The second step in this parallel between quantum mechanics and linguistic analysis consists in describing a form of complementarity. There are two mutually exclusive approaches to language: the synchronic approach and the diachronic approach. The synchronic approach tends to immobilize the language in its present form, namely in a present system of di erences between signs. The diachronic approach, instead, follows in the short term the developments of the practice of speech, and it tends to identify in the long term the drifts of the system of semantic di erences. Clearly, extracting a synchronic structure (from a snapshot of language), and making a diachronic analysis (of the history of language), are mutually exclusive operations. We find in Louis Hjelmslev’s work a detailed description of this di erence [20]. As a preliminary, Hjelmslev points out that while signs follow each other in speech, they coexist in the text that transcribes it. Their succession constitutes speech, and their coexistence constitutes texts. Both speech and texts involve a conjunction of signs. In both cases one sign comes, and then another, and then another etc. But this conjunction unfolds in time for speech, and it unfolds in space for texts.

Now, according to structuralism, each sign has a unique position which defines it by its di erences with respect to the others. No other sign can really replace it, since to take its place would be tantamount to endorse the same pattern of di erences and therefore to identify with its unique function. In this case, the signs form a disjunction: at each given position, one sign may be used or another, or another one etc. ; and this substitution changes nothing since