Alberto Voltolini

In recent works, Chomsky has once more endorsed a computational view of rulefollowing, whereby to follow a rule is to operate certain computations on a subject’s mental representations. As is well known, this picture does not conform to what we may call the grammatical conception of rule-following outlined by Wittgenstein, whereby an elucidation of the concept of rule-following is aimed at by isolating grammatical statements regarding the phrase ‘to follow a rule’. As a result, Chomskyan and Wittgensteinian treatments of topics immediately connected with rule-following, namely linguistic competence and understanding, are utterly different from one another. There are two possible stances that computationalists like Chomsky may adopt with regard to the discrepancy between the two aforementioned modes of dealing with rule-following, namely a conciliatory and a non-conciliatory attitude. According to the former attitude, grammatical remarks on and computationallyoriented theories of rule-following investigate one and the same topic although admittedly at different levels, namely a conceptual and an empirical one. According to the latter attitude, grammatical remarks are just a preliminary step in the investigation of rule-following which scientific advancement, presently represented by computationally-oriented theories on this matter, is well entitled to put aside. In what follows, however, I will try to show that both stances are problematic. The conciliatory attitude simply does not work, for it hardly copes with the fact that the concept of rule-following does not supervene, even weakly, on the property of rule-following, namely the property instantiated in the mental/cerebral phenomena that computationally-oriented theories of rule-following study. To take the contrary attitude, on the other hand, is to end up with another disappointing result, namely that the computational treatment of rule-following ultimately deals with something different from that which we wished to gain knowledge of when we began our inquiry into rule-following..

In The Theory of Objects, Alexius Meinong used true negative existentials to argue in favour of non-existent objects: in order to assert veridically that an object O does not exist, one has to refer to O itself1. From Bertrand Russell's "On Denoting" onwards, it has become a commonplace to say that this argument does not work. For every sentence apparently concerning non-existents one can provide a paraphrase which eliminates the singular term contained in it and therefore dispels the illusion of a reference to a non-existent item2. As is well known, this strategy implies that such a singular term is a camouflaged definite description. However, Saul Kripke and other supporters of the so-called new theory of reference have claimed that singular terms directly referring to their designation (ordinary proper names, indexicals, mass and substance terms: from now on, DR terms) are not at all synonymous with descriptions. In their opinion, true negative existentials represent no exception to this claim. In what follows I would like to stress that, though Kripke's critique of the Russellian treatment of negative existentials is convincing, his treatment of true negative existentials containing a DR term is flawed. Nor do other contemporary accounts that share Kripke's critique as far as these sentences are concerned fare any better. In fact, I think that Meinong's argument may be correct where true negative existentials containing DR terms are concerned. But I shall aim to demonstrate this elsewhere.

Jerry Fodor now holds (1990) that the content of mental state types opaquely taxonomized (de dicto content: DDC) is determined by the 'orthographical' syntax + the computational/functional role of such states. Mental states whose tokens are both orthographically and truth-conditionally identical may be different with regard to the computational/functional role played by their respective representational cores. This make them tantamount to different contentful states, i.e. states with different DDCs, insofar as they are opaquely taxonomized. Indeed they cannot both be truthfully ascribed to a single subject at the same time. Some years ago (1987), Fodor postulated a notion of mental content which also went beyond that of a mental state's truth-conditions. States whose tokens differ in their truth-conditions, or broad content, might, he claimed, still share a narrow content (NC), which was causally responsible for the shared behavior of the subjects of these states. For instance, two molecularly identical individuals, living in environments in all respects the same, except for the chemical substance of the phenomenically indistinguishable liquids filling their respective lakes and rivers, would behave similarly when having truth-conditionally different thoughts regarding those liquids. According to Fodor, this sameness of behavior was causally dependent on the sameness of the NC of the two individuals' truth-conditionally different thoughts. Now, this way of individuating mental states is still of interest for semantics. Indeed, NC allows one contextually to fix the broad content of a mental state token. Echoing Kaplan's notion of character,1 Fodor explained NC as a function that mapped contexts (of thought) onto broad contents. NC was thus invoked by Fodor mainly in order to account for sameness of intentional behavior. But DDC also plays a role in explaining intentional behavior, precisely by explaining why a subject whose thought-tokens have identical truthconditions may behave differently..

In this paper I want to show that the idea supporters of traditional creationism (TC) defend, that success of a fictional character across different works has to be accounted for in terms of the persistence of (numerically) one and the same fictional entity, is incorrect. For the supposedly commonsensical data on which those supporters claim their ideas rely are rather controversial. Once they are properly interpreted, they can rather be accommodated by moderate creationism (MC), according to which fictional characters arise out of a reflexive stance on a certain make-believe process. For MC, success of a fictional character across different works amounts to the fact that, first, different work-bound ficta are related with each other by means of a relation weaker than numerical identity, transfictional sameness, and second, that all those ficta are related by transfictional inclusion to a fictum that in some sense gather them all, the so-called general character. Since a general character is an abstract constructed entity, moreover, the more those particular ficta are generated, the more general fictional characters including all of them arise.

In (1990), Jerry Fodor has defended a naturalized conception of meaning for Mentalese expressions which relies on the notion of asymmetric dependence. According to this conception, any naturalized theory of meaning must be able to account for the fact that meaning is robust, namely that any token of a certain Mentalese expression “x” retains the expression’s meaning, X, for any Y (≠ X) which happens to cause it. Now, this robustness of “x”‘s meaning can precisely be explained in terms of the subsistence of an asymmetric dependence of any nomic connection between Ys and “x”‘s tokens on another nomic connection between Xs and “x”‘s tokens. According to Fodor, then, this relation between nomic connections can account in perfectly naturalistic terms for “x” meaning X, by providing a sufficient condition for such a meaning. In what follows, however, I will try to show, first, that the subsistence of asymmetric dependencies of the kind envisaged by Fodor is not enough for assigning meaning to a certain expression. Indeed, there are dependencies of this kind which are meaning-irrelevant. Secondly, I will claim that asymmetric dependence relations are not able even to support for the robustness of meaning. For there are cases in which according to the structure of these relations we would have to conclude that meaning is altered although robustness would require to have it unchanged.

In one of its latest papers Timothy Williamson has drawn a distinction between two readings of the phrase "possible F", where "F" is a predicate variable: the predicative and the attributive. In what follows, on the one hand I will hold that the first reading naturally applies to the phrase "possible object", thereby supporting a moderata conception of possibilia as entities that possibly exist. Moreover, I will maintain that one such conception provides the best possible account of Tractarian objects. On the other hand, I will claim that the second reading of the phrase "possible state of affairs" supports a parsimonious ontology of states of affairs, according to which a possible state of affairs is nothing over and above the de re possibility its constituents have of being related in a certain way to each other. As a result facts, i.e. possible states of affairs actually obtaining, are the only genuine states of affairs that there are, or, to put it differently, for any given state of affairs it is redundant that it is a fact. Besides, I will also claim that, appearances notwithstanding, this ontology of Sachverhalten underlies the Tractatus itself. Finally, I will try to show how, again appearances notwithstanding, this thrifty ontology of Sachverhalten squares with the semantic theses of the Tractatus according to which an elementary proposition presents a possible state of affairs as its sense and it is true if ‘that’ possible states of affairs obtains.

Da Hegel fino a Bradley, l'attacco idealista ad una concezione pluralistica della realtà come una credenza non suffragata dalla verità delle cose si è valso dell'argomento semantico secondo il quale le espressioni indicali, su cui da ultimo riposerebbe tutta la valenza referenziale del linguaggio, non si riferiscono a segmenti discreti del reale ma si limitano ad esprimere universali. Dal versante ontologico opposto, Russell ha guidato la reazione all'idealismo assoluto (inaugurando così uno dei filoni di riflessione della filosofia analitica) facendo perno proprio sulla capacità da parte di tali espressioni di designare direttamente atomi distinti di realtà. L'argomento idealista, nelle sue linee generali, funziona così. Anche ammesso che un termine singolare "N" non possa denotare un'entità reale finché non è disambiguato da un completamento indicale (del tipo "questo N"), un'operazione del genere è inutile perché un'espressione indicale è a sua volta incapace di designare univocamente un individuo. Quando uso "questo", ad es., un tale uso non designa alcunché perché ogni cosa è un questo e il questo in sé stesso non è alcun individuo. Quello che faccio in tale uso è piuttosto dare luogo all'espressione di un universale, nella fattispecie del concetto (individuale) di questità. L'argomento idealista è fallace. Ma la replica russelliana, che pure ha dalla sua tutta la forza del senso comune, non è in grado di metterlo fuori gioco integralmente. Infatti, entrambi assumono - l'uno come premessa all'interno di un ragionamento ad absurdum, l'altro come tesi positiva - che riferirsi mediante un indicale ad un individuo reale corrisponde a discriminarlo da tutti gli altri individui del suo dominio. Questo significa, in altre parole, che entrambi assumono la tesi dell'indistinzione di semantica ed epistemologia. Solo una dottrina che condivida le tesi semantiche di Russell sugli indicali ma non presupponga il collassare di semantica ed epistemologia può far decadere del tutto l'argomento idealista..