I believe that Levine’s response to the knowledge argument is subject to at least two kinds ofobjections. First, it is not as obvious as Levine assumes that Mary, who knows everything about thephysical, would really learn something new upon her release. In fact there have been several attemptsto reject this assumption (e.g. Paul M Churchland, ‘Reduction, Qualia, and the Direct Introspection ofBrain States’, Journal of Philosophy 82, 1985; Daniel C. Dennett, Consciousness Explained 1991).Second, there are a number of objections against Levine’s way of undermining premises of theknowledge argument (e.g. David J. Chalmers, The Conscious Mind 1996; Daniel Stoljar,‘Physicalism and the Necessary A Posteriori’ Journal of Philosophy 97, 2000). Levine says ‘For thesame reasons I do not accept the conceivability argument, I do not accept the knowledge argument’(p. 77), but the knowledge argument is obviously quite distinct from the conceivability argument.

If Levine’s explanatory gap were merely apparent then there would be several possible ways formaterialists to overcome the gap. They might be able to show that the gap is filled becausephenomenal consciousness is to be reductively explained. Or they might be able to demonstrate thatthe gap is an illusion because phenomenal experience is to be eliminated from their ontology. Inchapters 4 and 5 Levine tries to refute those attempts. As a consequence he is committed to the‘modest qualophile’ which says that, although there are good reasons to hold materialism, ‘nomaterialist theory seems to really explain our experience, to make intelligible how a system satisfyingthe materialist’s description could be a subject of conscious experience’ (p. 128). Finally in chapter 6Levine closes his exploration of the puzzle of consciousness by analysing the ‘replacement argument’and the ‘zombie epistemology argument’, both of which have arisen from the recent debate onconsciousness and cognition.

One weakness that I find in the book is Levine’s over-dependence on the conceivability argument.On the one hand I think that Levine’s notion of the explanatory gap is extremely compelling and thatit is at least very difficult for materialists to provide a fully acceptable solution that bridges the gap.But on the other hand I worry that readers might have a false impression that Levine’s concept of theexplanatory gap is almost solely derived from the highly controversial conceivability argument. Foralthough Levine examines other challenges to materialism, his discussion of them is significantlybriefer than that of the conceivability argument. I nevertheless recommend the book to anyone who isinterested in the problem of consciousness. Though the recent popularity of consciousness studies hasproduced an increase in books on the topic, I find few more provocative and well-written thanLevine’s Purple Haze.

Yujin Nagasawa Australian National University

Fisette, Denis (ed.), Consciousness and Intentionality: Models and Modalities of Attribution,Dordrecht, Kluwer Academic Publishers, 1999, pp. viii + 361, US$140, £88 (cloth).

Consciousness and Intentionality is a collection of papers based on talks delivered at Montreal in1995, at a conference with the same title. The collection is divided into four thematic sections:(1) consciousness, with an introduction by Pierre Poirier; (2) the problem of qualia, with anintroduction by Luc Faucher; (3) the ascription of intentional states, with an introduction by MartinMontminy; and (4) externalism and mental causation, with an introduction by Paul Bernier. The foursections are loosely related to each other and even the papers within each section cover a wide varietyof issues, as often happens when conference presentations are collected together. In spite of theheterogeneity of topics and philosophical stands represented, it is interesting to note that there areunifying features. For instance, the philosophers whose work is most often quoted across the collectedpapers, and considered as a starting point for discussion, are Donald Davidson for an accountof intentional ascription and mental causation and Fred Dretske for a naturalized theory ofrepresentation. The reference to Davidson’s and Dretske’s writings allows us to identify a commonbackground for some of the topics that receive attention here. More generally, the collection in itsentirety can be seen as a contribution to the investigation of the limits and the virtues of naturalizationprojects, and a very useful reading for the philosopher of mind who wants to keep up with the recentdebates.

As we are told in the Preface and in the Introduction to the first section, the necessity to deal withthe topics of consciousness and intentionality comes from the attempt to dissolve an apparent tension.Most of the contributors embrace a naturalistic view of the mind, according to which mental states,events or properties are natural states, events or properties. While natural phenomena are supposedto be satisfactorily explained within the framework of a natural science, the phenomena of

Vol. 80, No. 2; June 2002 247

consciousness and intentionality seem still mysterious and not fully characterizable in naturalisticterms. The objective of the philosopher of mind is to contribute (with the co-operation of thepsychologist, the biologist and the neuroscientist) to a naturalistic explanation of intentionality,consciousness and rationality in physical organisms.

Not all the contributors to this collection endorse naturalization projects with equal enthusiasm,but they do not challenge such projects of naturalization on the basis of their presumed explanatoryinadequacy. On the contrary, they accept naturalism as a framework that probably cannot be assessedindependently of a metaphysical view of the world. Poirier writes in the Introduction to the firstsection:

This is not the place to question the value of so-called naturalization programs. In fact, there mayvery well be no metaphysical standpoint from which to evaluate this type of philosophicalprogram (2).

Let us briefly consider Joëlle Proust’s very interesting paper on intentionality, to be found in thefirst section, as an example of a philosophical enquiry that takes the naturalistic framework forgranted. The question Proust focuses on is whether an information can qualify as mental content onlyif it is accessible to the subject. A first step is to specify what ‘accessible’ means. In the literature ithas been argued that a piece of information is cognitively available if it is available to the wholesystem for its survival (according to the principles of globality and functionality), and if it is self-acquired and not just passively inherited (according to the principle of self-acquisition). In otherwords, the system uses the information in a cognitive way if, for instance, that information can bedescribed as a reason for the system’s behaviour. The definition of cognitive availability by appeal tothe three principles mentioned above is very controversial, and Proust shows that the principlesinvolved can also be found to be incompatible with one another. These difficulties call for an analysisthat either restricts the class of accessible information, or relaxes it, and the result can have importantconsequences for an analysis of conscious reports as well. Proust’s conclusion is that what we callcognitive information might not be accessible to the whole system, and that the subject might also failto report coherently and reliably its own conscious states, since the same notion of a ‘singleperspective’ seems undermined.

Proust’s paper is a revisionist one, as it does challenge the traditional philosophical role of theintentional subject. Demystification is not a consistent target in the book. The collection hosts thediscussion of problems that have little bearing on the question of whether naturalization projects willever be explanatorily adequate (see Kevin Mulligan’s paper on perception, Shaun Nichols andStephen Stich’s on empathy or Daniel Laurier’s on the indeterminacy of interpretation). Moreover,there are papers that prefer to highlight the limits of a naturalistic perspective rather than its benefits.One of these is Frances Egan’s paper in defence of interpretational semantics.

Egan expresses scepticism about the success of naturalistic semantics and suggests that contentascription should be construed as partly pragmatic. The reasons why Egan thinks that naturalisticsemantics is doomed to failure are mainly its incapacity to deal with the possibility of error and thelack of determination that is likely to affect the meaning relation as physical relation. Egan describesthe problem of misrepresentation in the following way. Suppose meaning is a relation R betweentokenings of certain internal non-semantically individuated structures and X. If only Xs can bearrelation R to these structures, then all tokenings of these structures are true and misrepresentationcannot occur. If, in circumstances that are not ideal (C), Ys (not Xs) bear relation R to the structures,then the structures can mean Xs or Ys-in-circumstance-C. If late at night I come to believe that thereare cows standing in the fields, but actually the animals I see are horses, then ‘cows’ means cows orhorses-on-a-dark-night. Egan seems to dismiss the solutions to this puzzle that have been put forwardin the literature without even introducing them.

The second problem seems to be the Quinean problem of content specification. Do ‘cows’tokenings bear a meaning relation to cows only, or to undetached cow parts as well? According toEgan, naturalistic semantics cannot easily provide a way to select among properties that are co-instantiated. Egan offers solutions to these problems by adopting a version of computationalsemantics according to which content is ascribed by taking into consideration the interests of theinterpreter. Misrepresentation is characterized as a mismatch between a content associated by aninterpreter to some behaviour of A (e.g. the ethologist’s explanation of the behaviour of the frog asdetecting and eating flies) and some instances of A’s same behaviour that are associated with adifferent content (e.g. the frog detecting and eating black spots). Egan believes that only for thepurposes of an explanation of the relation between frogs and flies in the natural environment, say, thedifference in content matters.

248 Reviews

Pragmatic factors also have a major role in preventing indetermination. ‘Cows’ tokenings co-varyboth with cows and with undetached cow parts, but the interpreter who is interested in the behaviourof subjects in relation to middle-sized objects will not consider the correlation of tokenings withtemporal segments of these objects as a relevant alternative. Egan’s arguments are just sketched. Theyare not sufficiently detailed nor cogent to be persuasive, as in the paper there is no indication of howthe naturalist has replied (or could reply) to the challenges, and no discussion of whether thepragmatic version of computational semantics could also get into trouble. The problems Eganconsiders are nonetheless worth attending to.

I have chosen to focus on Proust’s and Egan’s papers because they represent two oppositeattitudes in the debate on the limits of naturalism, but in the collection there are many othercontributors that deserve mention: in the section on consciousness, Pierre Jacob, William Seager andPierre Livet; in the section on qualia, David Rosenthal, Elisabeth Pacherie and Evan Thompson; inthe section on intentional description, David Davies; in the section on mental causation andexternalism, Ausonio Marras, Steven Davis and Michel Seymour. In general, the desirability ofnaturalization projects is rarely questioned by the contributors, but the central idea in the collection isthat, when we think about phenomena like consciousness and intentionality, we cannot pretend thereis nothing left to explain. We just have to work harder in order to fill the gaps.

Lisa Bortolotti The Australian National University

Carstairs-McCarthy, Andrew The origins of complex language, Oxford, Oxford University Press,1999, pp vi + 260.

There are three features of contemporary languages that stand in need of explanation because, theauthor claims, they are not inevitable. These features are: large vocabulary, duality of patterning (theanalysability of utterance into morphemes and phonemes), and the distinctness between the nounphrase and the sentence.

Almost all signs function in the fact that they contrast with other signs, a point made by Saussure.Another way to put this is that a new utterance in a language system will be taken to mean somethingnew, as against be assumed to have the same meaning as an existing utterance. Carstairs-McCarthyspeaks about such ‘synonymy avoidance’ as a principle driving the expansion of vocabulary—anexpansion given elbow-room by the improved articulatory range resulting from the evolution(literally, the descent) of the larynx. This anatomical change helps to allow the phonemic contraststhat mark the difference between human utterance and that of other apes. The author’s hypothesis isthat greater articulatory range provided the selective pressure, given synonymy avoidance, for greaterneural resources devoted to remembering the greater number of utterable signs. Duality of patterningis the related phenomenon allowing an increase in lexical items on the efficient principle that a newitem may be constructed by, for example, joining two existing ones. Thus, the brain does not need toexpand relentlessly, as it might under a regime where each new lexical item was unique in all itsarticulatory elements.

The third feature of languages, the noun phrase as distinct from the sentence, is accounted for in along discussion of the primitive status of the syllable. This is understood as a fundamental articulatorycomponent of utterance, and Carstairs-McCarthy examines the prospect that its constituents stand as amodel and miniature for the components of sentences. Thus, he argues for a phonological basis forsyntax.

Having set out the case that the structure of complex language derives from the way utterance hasbecome produced by humans, the author proceeds to consider whether the archaeological and fossilrecord support such an idea. The evolution of the larynx in hominids has occurred over the lastseveral million years, and Carstairs-McCarthy acknowledges that only in the last 50–100,000 yearsare there signs of behaviour unambiguously different from that of other modern apes. This is not amajor problem if it is assumed, (1) that the utterances of fellow primates are not fundamentally ofa different sort to human ones, and (2) that there need be no close link between hominid utterance andother kinds of behaviour.

As regards the first point, Carstairs-McCarthy relies on the prospect that animal calls may achievereference, as in the different cries made by vervet monkeys in the face of different predator classes.Even granting this contestable claim, an issue that remains to be explained (to this reviewer, anyway)is the capacity for reference in the absence of the object or event referred to. This detachability ofsigns from contexts, their transportability and re-use in achieving reference, relies on an ongoing

Vol. 80, No. 2; June 2002 249