| John van Benthem. The Logic of Time. A
Model-Theoretic Investigation into the Varieties of Temporal Ontology and Temporal
Discourse, Synthese Library, vol. 156, Kluwer Academic Publishers, Dordrecht, second
edition 1991. xxii + 280 pp. This is still the most comprehensive and readable book on tense and time from a logical
point of view. The present edition adds to the first edition of 1983 a one-page preface,
three pages of addenda and corrigenda (mainly short addenda), an appendix "Septs ans
après" (14 pages) and 5 pages of additional reference up to 1989. The body of the
text has remained unchanged except for a few minor corrections. The book divides into two parts, titled "Temporal ontology" and
"Temporal discourse" respectively. The first part treats the model theory of
strict partial orders; the second part investigates descriptions of these models in
suitable sentential modal languages. Within each part two options are first explored and
then compared: time as based on points and time as based on intervals. In the first part
there is a short additional chapter on events, which is more suggestive in character and
contains digressions of Zeno-type paradoxes and the causal theory of time. The book also
contains an appendix on space in which some of the formal analogies with time are
explored. In the first part the theory of point structures is cast in a first order language with a relation symbol <; there are only few digressions into higher order descriptions. The natural models for strict partial orders are the rationales, reals, or |
integers under the relation of precedence. Further axioms,
like succession (postulating a successor and predecessor under < for each object in the
domain), or density, or discreteness, narrow down the range of models in the one or the
other direction. Still, many non-standard models are caught in these nets and the task at
hand is to study how far non-standard models can be excluded by imposing further axioms
(here some consideration is given to higher-order axioms). For the study of period
structures the language is extended by an inclusion relation, Í.
Natural models here are sets of open intervals of rationales under set-inclusion and
precedence. In the second part the constitution of time from points is represented by way of Kripke structures for traditional tense logic. This covers mostly well-known ground in an elegant manner with its emphasis on correspondence theory. Completeness and incompleteness results are proved and a large number of correspondences between tense-logical postulates and first-order conditions on frames are exhibited. Traditional Priorean tense logic may also be interpreted as talking about periods. But traditional Priorean operators cannot tap the structure within periods. To Prior's F and P the author adds a new operator r with the reading: r A holds for period i iff A holds for each sub-period of i. It turns out that this extension of the FP-language is expressive enough to cover a large number of period-operators considered in the literature. This book is densely packed with ideas and results. Logical matters are presented rigorously - mainly from a model-theoretic point of view - but with a minimum of formalities. Indeed, frequently the tone is deceptively chatty. |
REVIEWS |
71 |
| Many of the proofs are little more than hints; filling in the
details is, in many cases, a challenging exercise. The ideal reader will be a philosophically minded logician; but the book will also appeal to
philosophers who may (but should not!) be inclined to skip the proofs and to logicians who
may (but should not!) be inclined to skip the philosophy. André Fuhrmann Department of Philosophy, University of Konstanz John R. Searle. The Rediscovery of the Mind. The MIT Press, Cambridge,
Massachusetts and London, 1992, 270 pp. John Searles' new book, which title is an obvious homage to Bruno Snell's classic, "The Discovery of the Mind", undoubtedly must be regarded as a milestone in the philosophical discussion on the nature of mind. The main aim of Searle's work is to criticize and overcome the dominant traditions in the philosophical study of mind, both "materialist" and "dualist". Because he thinks consciousness is the central mental phenomenon, the author wants to begin a serious examination of consciousness ion its own terms, he tries to put the final nail in the coffin of the theory that the mind is a computer program and he makes proposals for reforming our study of mental phenomena in a way that would justify the hope of rediscovering the mind. |
The first three chapters of Searles' study contain criticisms
of the dominant views in the philosophy of mind. Starting with a discussion of what's
wrong with the philosophy of mind, Searle characterizes and refutes the philosophical
doctrine of materialism in its various forms (Logical behaviorism, Type identity theory,
Token identity theory, Black box functionalism, Strong Artificial Intelligence,
Eliminative materialism, Naturalizing intentionality). He claims that the basic assumption
behind materialism is essentially the Cartesian assumption that materialism implies
antimentalism and mentalism implies antimaterialism. Answering his own question "Why
should materialists have a fear of consciousness?" Searle points out: "The
deepest reason for the fear of consciousness is that consciousness has the essentially
terrifying feature of subjectivity. Materialists are reluctant to accept that feature
because they believe that to accept the existence of subjective consciousness would be
inconsistent with their conception of what the world must be like." (p. 55). To
undermine the view of the world as completely objective nap to show that as far as the
ontology of the mind is concerned, behaviour is simply irrelevant, Searle uses his thought
experiments of Silicon Brains and Conscious Robots. But the author does not embrace dualism. Both sides are profoundly mistaken, he claims. Once you start counting types of phenomena, you are on the wrong way, whether you stop at one or two. According to Searle it is essential to show that both dualism and materialism are false because it is generally supposed that these exhaust the field, having no other options. To determine his own philosophical position Searle writes "My own views do not fit any of the traditional labels, but to many philosophers the idea that one might hold a view that does not fit these categories seems incomprehensible." (p.15) |
REVIEWS |
72 |
| The next five chapters are a series of attempts to give a
characterization of consciousness and to elaborate a philosophical theory of
consciousness. Searle tries here to give an answer to the following crucial questions: How
do we locate consciousness in relation to our scientific world view? How do we account for
its apparent irreducibility according to the standard patterns of scientific reduction?
What are the structural features of consciousness? How do we account for the unconscious
and its relation to consciousness? What are the relations between consciousness,
intentionality and the Background capacities that enable us to function as conscious
beings in the world? Despite of Searle's critical remarks on property dualism, introspectionist, incorrigibility and other points of view in the philosophy of mind, the main effort in these central chapters is not critical. His aim is to show the place of consciousness in our general conception of the world and the rest of our mental life. According to Searle, consciousness "is a biological feature of human and certain other brains. It is caused by neurobiological processes and is as much a part of the biological order as any other biological feature such as photosynthesis, digestion, or mitosis" (p. 90). It must be clear that in that sense consciousness is a part of the natural world. Maybe the most important feature of consciousness is its subjectivity. Once the existence of subjective consciousness is taken for granted, then there should be nothing strange, wonderful, or mysterious about its irreducibility. "Given its existence, its irreducibility is a trivial consequence of our definitional practices." (p. 124) The gross structural features of normal, everyday consciousness mentioned by Searle consist of the following dozen: Finite |
Modalities, Unity; Intentionality; Subjective Feeling;
The Connection between Consciousness and Intentionality; The Figure-Ground, Gestalt
Structure of Conscious Experience; The Aspect of Familiarity; Overflow; The Center and the
Periphery; Boundary Conditions; Mood; The Pleasure/Unpleasure Dimension. An argument
often used by the author for identifying a feature is the absence of the feature in
pathological forms. Searle concludes this part of his book with a discussion of the
following three theses about conscious states 1. All conscious states are self-conscious. 2. Consciousness is known by a special faculty of introspection. 3. Knowledge of our own conscious states is incorrigible. which seem (not very surprisingly) to him false. Explaining the relations between unconscious mental states and consciousness Searle formulates his connection principle "all unconscious intentional states are in principle accessible to consciousness" (p. 156). Comparing his conception of the unconscious and its relation to consciousness with Freud's the author characterizes Freud's theory as incoherent in at least two respects. "First, I can't make his account of the ontology of the unconscious consistent with what we know about the brain, and second, I can't formulate a coherent version of the analogy between perception and consciousness." (p. 169) Discussing the relationship between consciousness, intentionality on the one hand, and the capacities, abilities, and general know- |
REVIEWS |
73 |
| how that enable our mental states to function on the other,
Searle modifies his previous conception of the Background. "I now think the real
mistake was to suppose that there is an inventory of mental states, some conscious, some
unconscious. Both language and culture tend to force this picture on us. We think of
memory as a storehouse of proposition and images, as a kind of big library or filling
cabinet of representations. But we should think of memory rather as a mechanism for
generating current performance, including conscious thoughts and actions, based on past
experience." (p. 187) It is important to emphasize that the Background in Searle's
sense is not a system of rules. Chapter 9 of the present book deals with cognitive science. Searle that discipline suffers from the fact that several of its most cherished foundational assumptions are mistaken. His criticism is directed against the so-called computational claim of cognitive science, against its basic assumption that the brain is a computer and that mental processes are computational. Searle's argumentation is conclusive and shows clearly: "The brain, as far as its intrinsic operations are concerned, does no information processing. It its a specific biological organ and its specific neurobiological processes cause specific forms of intentionality. In the brain, intrinsically, there are neurobiological processes and sometimes they cause consciousness. But that is the end of the story." (p. 226) The aim of the final chapter is to explore some of the consequences of the general philosophical position that the author has been advocating for the study of the mind. It starts with a discussion of the connection principle and its implications and ends with a criticism of the postulation of unconscious rules in universal grammar. |
I am convinced that Searle's interesting and polemical book
will stimulate the further discussion in the philosophy of mind. It is surely not so
difficult to disagree with some of its arguments but, as I guess, not so easy to refute
them. Klaus Wuttich
Hubert L. Dreyfus. What Computers Still Can't Do. A
Critique of Artificial Reason, The MIT Press, Cambridge, Massachusetts and London,
England, 1992, liii + 354 p. In his book "What Computers Can't Do" Dreyfus criticizes the so called information processing approach to AI which claims that all that is relevant to intelligent behaviour can be formalized in a structured description. His own thesis is that "whenever human behaviour is analyzed in terms of rules, these rules must always contain a ceteris paribus condition, i.e., they apply 'everything else being equal,' and what 'everything else' and 'equal' means in any specific situation can never be fully spelled out without a regress. [...] Thus in the last analysis all intelligibility and all intelligent behaviour must be traced back to our sense of what we are, which is, according to this argument, necessarily, on pain of regress, something we can never explicitly know." (pp. 56-57) The new material of the present edition consists of an introduction to an introduction to a classic book against Artificial Intelligence (with |
REVIEWS |
74 |
| an introduction). Comparing with the foreword of the '72
edition, the style of the level-3-introduction reveals much more self-confidence. As in
1972 Dreyfus carefully prepares his defence positions in face of the foreseen counter
attacks of the enraged Al encampment ["I want to make absolutely clear from the
outset that what I am criticizing is the implicit and explicit philosophical assumptions
[...] not their technical work." (P. 87)], twenty years later it sounds like that:
"After fifty years of effort, however, it is now clear to all but a few diehards that
this attempt to produce general intelligence [that a high-speed digital computer might
exhibit intelligent behaviour - M.U.] has failed. [...] the research program based on the
assumption that human beings produce intelligence using facts and rules has reached a dead
end, and there is no reason to think it could ever succeed. Indeed, what John Haugeland
had called Good Old-Fashioned AI (GOFAI) is a paradigm case of what philosophers of
science call a degenerating research program." (p. ix) Dreyfus received vivid reactions to his criticism after publishing the first edition of "What Computers Can't Do". "Dreyfus's complaints were broadly perceived within the AI community, and within the discipline of philosophy as well, as shortsighted and unsympathetic [...]" (P. & P. Churchland "Could a Machine Think"? Scientific American January 1990, p. 33) Nevertheless, he held his view tenaciously criticizing all the news defence strategies the AI literature came up with. Dreyfus, "the black knight of AI", has fought it for more then 20 years, and eventually it's gone. He compares the dynamics of the |
breakdown of GOFAI with the amazingly quick disintegration of
the Soviet Union (cf. p. xiv) Dreyfus's work might serve as an encouraging example to the
aim that sticking to one's own convictions pays off in the end - if there indeed would be
and end of GOFAI. Unfortunately, Dreyfus has to notice somewhat astonishedly that
"the GOFAI research program has refused to degenerate gracefully and is fighting
on" (p. xv) although it had lost almost all support on the part of the U.S., Defense
Department. (cf. p. xlvii). So one still may doubt whether GOFAI is in fact as dead as
Soviet Union and the dinosaurs. In the '92 - introduction he mainly deals with a new trend
in Artificial Reason: research on neutral-network modelling, which is alledgly concerned
to some extent with irrationalistic philosophical theories. However, his evaluation of
this new approach, i.e. of network based reinforcement learning does not leave much hope
for AI research either: "All work in AI, then, seems to face a deep dilemma. If one
tries to build a GOFAI system, one finds that one has to represent in a belief system all
that a human being understands simply by being a skilled human being. [...] Happily,
recent research in machine learning does not requirethat one represent everything that
human beings understand simply by being human. But then, as we have just seen, one
encounters the other horn of the dilemma. One needs a learning device that shares enough
human concerns and human structure to learn to generalize the way human beings do. And as
improbable as it was that one could build a device that could capture our humanity in a
physical symbol system, it seems at least as unlikely that one could build a device
sufficiently like us to act and learn in our world." (pp. xlv-xlvi) It is a rather obvious thesis that under completely new circumstances "we would not |
REVIEWS |
75 |
| know what to pay attention to or how to act appropriately." (p. xxix) Just take an inhabitant of N.Y.C. away to the desert (or vice versa) and observe how he/she fails to solve the frame problem. For a robot, any part of our human world is completely new, so why should we expect it to act faultlessly from the very beginning? It is completely obvious that it will make mistakes. The point is the ability to provide a reasonable analysis of these mistakes so as to avoid them in future. What is at stage here is skilful evaluation of situations rather then checking the storage of a structured knowledge basis. What does the process look like in case of the humans? The most natural way appears to be learning by doing (or else, reinforcement learning). As it comes to explanation, however, there is nothing better than rule systems. And in case pedagogics comes up with something new some day - why should that be incapable for artificial trainees as well? | According to Kazimierz Ajdukiewicz's third dogma of
rationality one should avoid claiming something with a degree of certainty which surpasses
it degree of foundedness. So, maybe, at present it would be better to formulate the final
valuation of GOFAI a bit more cautiously. According to a German saying, what is prematury claimed to be dead often enjoys a long and prosper life thereafter. "What Computers Can't Do" is a classic in critique of Artificial Reason. Beyond doubt, it is still very worth reading, both as a reference book and as a monograph on the philosophical foundations of AI. The present edition deserves additional attention as a case study in history of science. Max Urchs Wrocław University |