Motto : Crapula ingenium offuscat. Traduction : "le bec du perroquet qu'il essuie, quoiqu'il soit net" (Pascal).

Ce blog est ouvert pour faire connaître les activités d'un groupe de recherches, le Séminaire de métaphysique d'Aix en Provence (ou SEMa). Créé fin 2004, ce séminaire est un lieu d'échanges et de propositions. Accueilli par l'IHP (EA 3276) à l'Université d'Aix Marseille (AMU), il est animé par Jean-Maurice Monnoyer, bien que ce blog lui-même ait été mis en place par ses étudiants le 4 mai 2013.

Mots-clefs : Métaphysique analytique, Histoire de la philosophie classique, moderne et contemporaine,

Métaphysique de la cognition et de la perception. Méta-esthétique.

Austrian philosophy. Philosophie du réalisme scientifique.

lundi 10 juin 2013

Review of David Armstrong : Sketch for a Systematic Metaphysics, Oxford, 2010, 125 pp.

Erwin Tegtmeier

The title of the book announces a systematic metaphysics.  Formerly that would have been pleonastic but these days it is remarkable since the metaphysics of mainstream analytical philosophy is mostly rudimentary and unsystematic. There are a few systematic metaphysicians but they are unpopular and widely ignored.  Traditionally a metaphysical or ontological system is a tree of categories and subcategories. Then there is a theory about connections between entities and complexes of entities which takes into account their categories. The system has to be comprehensive, i.e., each phenomenon in the world has to be classifiable and analysable with those categories which is not as easy as it sounds.
Thus we have to look at Armstrong’s tree of categories in this book. It seems that Armstrong has two highest categories below the category of existent: particulars and universals. He even seems to take for granted that any existent must be either a particular or a universal implying that this must hold in any ontology. In Russell’s ontology of 1918 as well as in the ontology of Gustav Bergmann who continued Russell’s Logical Atomism the category of facts is co-ordinated with the categories of particulars and universals with the consequence that facts are neither particulars nor universals. There is even a stark contrast between particulars and universals on the one hand and facts on the other. The former are simple while the latter are complex. Armstrong earlier did not bother about bringing together simples (so-called “thin particulars”) and complexes (facts) in one and the same category. The former are no longer mentioned in the book. Now, If the category of fact is subcategory of the category of particulars then all facts have to be particulars and there have to be some particulars which are not facts. Armstrong’s former thin particulars were such particulars. It seems that his facts consist ultimately only of universals. That was actually Russell’s later view which is sometimes called “the bundle view” because ordinary objects are ontologically analysed as bundles of universals. Armstrong refers to it without identifying it as Russell’s later view. Armstrong had mentioned earlier (p.13 ) that Russell  in his later years adopted the bundle theory but he seems not to realise that the quotation by which he supports his view that facts are particulars is an expression of that bundle theory. Armstrong had rejected the bundle theory at that place. He contrasts bundle and attribute theory and subscribes to the latter emphasising that particulars have properties (universals) rather than consisting of them. However, if thin particulars are dropped Armstrong inevitably ends up in the bundle theory.
There is a chapter headed “particulars” but its subject is not the comprehensive category of particulars which includes facts, nor is the subject thin particulars but only ordinary persistent and changing objects. Clearly, Armstrong analyses these objects as complexes, not as thin particulars. Hence, they should be members of the category of facts. Moreover, in the absence of thin particulars all particular are facts and the categories of particulars and facts coincide.
The particular chapter deals also with the identity of changing objects. Armstrong adopts an 18th Century distinction between strict and loose identity and he argues that the identity over time and change is only loose identity.  Since loose identity is opposed to strict identity it implies the negation of strict identity. Strict identity holds only between an existent and itself.  Does Armstrong then think that changing objects are not strictly identical with themselves though they are existents? He seems to adopt the view of Berkeley and Hume that changing objects are series rather than simple substances. However, not only simple but also complex entities are self-identical, i.e. strictly identical with themselves. In Berkeley and Hume the nature of complex entities is not clear but the  implication is that they do not exist.  The point of the Empiricist and presumably also Armstrong’s view is that series do not exist. Series are not recognized as entities by the Empiricists because their ontology is fundamentally reist. Armstong’s ontology is officially anti-reist in that it is an ontology in which facts play a central role, as he frequently emphasises. Such ontologies can acknowledge series by analysing them as facts.
It turned out that the issue here is existence, not identity. Existence and ontological status is certainly not the strength of Armstrong’s metaphysics. Armstrong plays with existence and existents proclaiming that a lot of existents supervene and are therefore, as he expresses himself, “free lunches”. “They do not add to being” as he likes to say. How can an existent not contribute to being though it is a being (existent)?  Should they not be considered then rather as non-existents? If one takes “add to being” in the arithmetical sense one could think of them as zero-existents, so to speak, existent without numerical identity. That would also lead to absurdity.
Armstrong’s supervenience-gambit reminds of the Medieval concept of transcendentals.  It is highly problematic if it is not restricted to fundamental phenomena such as sameness and existence.  The extensive use of the transcendentals move gives rise to severe difficulties. If the difference of two existents as well as their mereological whole are assumed, as is done by Armstrong, as no additions to being, they ground ontologically on those two entities and nothing else and would be indistinguishable. Therefore, one can argue that they ground one transcendental and only one and difference and mereological whole should not be two and differ. Armstrong generous use of the supervenients proves mistaken.  Armstrong  seems always happy to find non-entities and may feel in accord with Occam’s principle of parsimony. But too much parsimony is disastrous.  In the Middle Ages more and more of the Aristotelian categories were construed as transcendentals because they turned out to be relational and because the Aristotelian ontology never was hospitable to relations. Aristotle grants them only the lowest ontological status. Now, the ontological status of transcendentals was elusive. That is why Occam took the epistemological turn and declared that they are merely concepts in the mind. Occam’s conclusion led to representationalism and finally to idealism.
In the Preface Armstrong draws attention to the introduction of truthmaking theory as an important revision of his ontology (p. IX). With respect to truthmaking Armstrong assumes propositions as truth bearers. But their ontological status and categorisation is unclear. He traces them back to mind.  In a footnote on p.17 he writes: “I connect propositions with intentionality, and thus with mind. I do not think there is an ontological realm of propositions as some philosophers do.” It reveals a representationalist tendency that Armstrong attributes modalities to propositions rather than to facts.
One may doubt that Armstrong’s is really a fact ontology as he claims. In Russell facts are the crucial category because they adopt the role particulars had in the tradition. This move is not possible for Armstrong because he takes facts to be particulars. He cites approvingly Wittgenstein’s statement that the world is the totality of facts not of things. But this does not make sense in his ontology in which facts are things (particulars). He hails this even as the triumph of particularity, a phrase which may summarise the triumph of Aristotelian ontologies which are definitely ontologies without facts. They are reistic ontologies with particulars as the only full-blown beings.
There is no hint in Armstrong’s book concerning the category to which propositions belong or, if they form a category of their own, as to the category of which propositions are a subcategory. It seems that Armstrong does not take propositions to be linguistic entities. He writes: “I identify propositions as what is believed, what is supposed, entertained, doubted, etc. “ and “There is something abstract about propositions ...” (p.65). Armstrong considers Russell as the originator of truthmaker theory. In Russell propositions are also truth-bearers but they are linguistic entities. They are statements.  In Russell’s 1918 paper The Philosophy of Logical Atomism to which Armstrong frequently adjoins a proposition is “a sentence in the indicative” and Russell holds that what is believed, supposed etc. are facts.  
Armstrong also wants to continue Russell’s Logical Atomism in chapter 10 headed “Limits” where he mainly deals with Russell’s general facts. However, Armstrong’s general states of affairs are very different from Russell’s. The consequences of his categorising states of affairs as particulars appear. “All the persons in the room now” is for Armstrong an instance of what he calls “a totality states of affairs” (p.75). For Russell only a complete sentence represents a fact. The sentence: “all the persons in this room are Australians”, for example, would represent a general fact.
Russell had objected to Wittgenstein’s reduction of general facts to conjunctions of particular facts that the reduction would not be successful because it does not imply that the conjuncts are all the relevant particular facts. Now, it is far from Russell to reduce general facts to conjunctions of particular facts with a that-is-all clause. Rather, he takes the general quantifier to be a formal constituent of general facts. By contrast, Armstrong tries to make ontological sense of the phrases “all” and equivalently “no more”.  He distinguishes the intension of a totality state of affairs which is a property and the extension of it which Armstrong takes to be mereological sum and suggests thinking of that state of affairs as “cutting the mereological sum up without remainder” (p.79). Then he considers it as a relation holding between the property and the mereological sum. That would be a Russellian fact but not a general fact. Rather it would be in Russell a particular relational fact. Armstrong notes that the intension of a totality states of affairs is a property but not a universal which leaves it without categorisation in his ontology. Moreover, he explains that those states of affairs are “not additions to being” (p. 79).
Gustav Bergmann argued that the gist of representationalism is the assumption of a third kind of entities which are neither physical nor mental and function as representatives of the proper objects of knowledge. Armstrong’s propositions clearly are neither physical nor mental. They are contrasted by Armstrong with the real (p. 62) hence considered to be non-real. Moreover, propositions are not put into any of the categories. However, it seems at first that Armstrong does not fit into Bergmann’s genetic explanation of representationalism. Bergmann claims that the trend towards representationalism stems from a rejection of universals, from the lack of fundamental ties between particulars and universals and from the rejection of facts. Armstrong notes that in this book he assumes fundamental ties for the first time and as far as universals and facts are concerned he had advocated them for a long time. However, his universals are extraordinary. They are localised and characterised as repeatables which is contradictory since it implies that they are one but also more than one.  Armstrong’s facts are also extraordinary in that they are categorised, as was mentioned already, as particulars.
According to Bergmann, the representatives serve as bearers of features of the world which the respective ontology cannot cope with. That is just what Armstrong’s propositions do. They not only bear the property of truth but also modalities such as necessity and possibility. Most revealing is Armstrong’s explanation of propositions which implies that beliefs, suppositions etc. are not directed to facts but to propositions as their representatives. What allows Armstrong to get along with much fewer entities is his relation of truthmaking between existents and propositions. In contrast to the intentional relation of Bergmann which is one-one it is a many-one relation. Thus the same existent can make several different propositions true. He registers that reduction of existents in two principles: the entailment principle and the possibility principle. The former (p.65) says that the truthmaker of a proposition makes also its logically entailed propositions true. The latter claims that the truthmakers of a contingent proposition is also the truthmaker of the possibility of its falsity and Armstrong thinks that he can by this principle avoid any assumption possibilia. Strictly speaking, the entailment principle excludes the ontologisation of logical constants. Nevertheless, he wants to have entities with logical constants such as conjunctive universals (p. 29).
The theory of truthmakers was originally developed to loosen the connection between statement and stated such that the stated need not be a fact. Thus it was designed to be able to dispense with facts. According to this theory even single tropes can make a statement true. Correspondingly, in Armstrong many truthmakers of propositions are not states of affairs though officially his world is a world of states affairs.
It supports the diagnosis of representationalism that Armstrong views mathematics and logic as rational disciplines that deals with true propositions which are necessary and discovered a priori. (p.88). He also characterises mathematics as the science of structures without making clear to what category structures belongs. Since he talks about the instantiation of structures they may be categorised as universals. The structural view of mathematics is very common but it separates mathematics and its application leads into difficulties in Armstrong’s ontology because he advocates the principle that all universals have to be instantiated.  Armstrong tries to solve the difficulties by drawing on his theory of possibilities and taking the uninstantiated structures to be involved in possibilities.
In view of the title of the book one misses in large parts of the book the categorisations since the first question of the ontologist/systematic metaphysician is always, as we saw: what category does it belong to? In the short chapter on time for example the fundamental ontological alternative of relationism versus absolutism does not appear at all. Rather Armstrong discusses the issue of time in terms of the rather un-ontological and coarse classification of mainstream analytic philosophy. The criterion of that classification is whether only the present or also the past or also the future is taken to be real. It is not related to the categorisation of time.
In his summary of the whole book Armstrong emphasises as its main points that the world “is a structure of contingent states of affairs”, that states of affairs types have to be introduced and also negativity but not as absences (by which he means Russell’s negative facts) but as limits (i.e., as his totality states of affairs).  Nevertheless, the categorisation of structures is unclear. In the chapter on states of affairs there is talk of structural universals (p.29) but that cannot be what he means since he explicitly holds that the world is a particular (p.28). The concept of absences which he adopts from someone else (p.74) is highly misleading here since not only “there is no elephant in the zoo” represents a negative fact according to Russell but also “the elephant is not blue”.  Moreover, It is unclear to which of Armstrong’s categories the types of state of affairs belong. Armstrong’s schematisation of them makes them look like Russellian propositional functions. What is clear is that they are not universals, which were his original relata of the relation ‘causes’ and which led into the difficulty that particular instances of a causal law could not be logially derived. Obviously, in Armstrong they cannot be particulars either.



Pièces mécaniques en chocolat - Lucca