Internalism & Externalism
[Second Draft: comments welcome]
1. A Short Trip
Gottlob Frege [1848-1925] was convinced that his own Frege cases, according to which a subject may assent a certain sentence and dissent from another sentence which is exactly like the previous one but for the fact that a singular term of that sentence is there replaced by a co-referential expression, did not undermine the idea (which he implicitly endorsed) that the sense expressed by a sentence is identical with its truth-conditions. For the truth-conditions of a sentence mirror not only the sentence’s informational value - roughly, what is said of what the sentence is about (in the simplest case, this amounts to be a possible state of affairs made out of certain objects and properties) - but also its cognitive value - the way the subject conceives what the sentence is about. This depends on the fact that the thought expressed by a sentence, i.e. its truth-conditions, is made out of the senses of the sentence’s components, which according to Frege are the modes of presentation of those components’ referents.
In the Fregean camp, the notion of mode of presentation has been further conceived in a plurality of ways. A mode of presentation may indeed be taken as having i) a descriptive nature (as a consequence, the presented object is the entity (if any) which uniquely satisfies certain conceptual requisites) ii) as having a semi-descriptive nature - (in such a case, the mode is empirically object-dependent: the presented object is the entity (if any) which contextually satisfies certain conceptual requisites) iii) as conceptually object-dependent (no mode of presentation without an object that mode presents) iv) as intimately linked with the object it conceptually depends upon (so that the thought expressed by means of it is not only object-dependent, but also object-constituted).
Some thought experiments, attuned with Kripke’s new theory of reference (according to which the meaning of proper and (most) common names collapses on their reference), have contributed to discredit Frege’s afore-mentioned conviction. Putnam (1975) and Burge (1979) have indeed made up two stories which are intended to show that there may be two physically identical subjects (two twins; or one and the same subject located both in an actual and in a counterfactual situation) who share the same conception of the world although the morphologically identical sentences they utter (as well as the beliefs these sentences express) have different truth-conditions, insofar as they have different informational value, i.e. they are about different things. A similar result has been obtained by Kaplan (1989) with respect to sentences containing indexicals, i.e. context-sensitive expressions.
Once we take the sentences uttered in the Putnam/Burge cases as precisely expressing the beliefs possessed by the subjects involved in those cases, the above semantical results have immediate consequences on the issue of the individuation of intentional states. An intentional state such as a belief has two components, namely what makes it that kind of mental state (i.e. a belief rather than a desire, or a state of fear etc.) and what makes it a representational state, i.e. a mental state endowed with a certain representational content. (In the old-fashioned phenomenological terminology, this is the distinction between the quality and the matter of a mental state.) Moreover, according to whether we conceive the representational content either as relational, i.e. as depending for its own identity on the external reality(s) it is about, or as non-relational, i.e. as being what it is regardless of whether it is about an external reality, we get two ways of individuating one and the same intentional state. These ways made Putnam (1975) originally speak of a broad and a narrow psychological state.
Taken in their simplest versions, externalism and internalism are the conceptions according to which, pending on the broad vs. the narrow identification of an intentional state, the content of such a state can legitimately be conceived only either as relational or as non-relational respectively. For externalists, the representational content of an intentional state depends on a reality lying outside the subject of such a state. For internalists, no external object or event which lies or occurs outside a subject’s brain (or at most its body) is relevant for the individuation of the content of an intentional state.
To be sure, however, different versions of both externalism and internalism are possible. First of all, as suggested by the difference between the Putnam- and the Burge-cases, externalism divides itself into natural and social externalism, depending on whether the environment which is relevant for the determination of a thought’s content is taken to be the physical environment that thought’s subject is related with or the social community that subject is taken to (really or ideally) belong to. Social externalism has a broader scope than natural externalism. For unlike the latter, which primarily concerns natural kinds- and mass-terms (as well as the corresponding conceptual constituents of the thought expressed by a sentence having those terms), it theoretically applies to all conceptual/lexical constituents of a thought vs. of the sentence which expresses it respectively. Moreover, it makes the normativity issue enter the stage. For a term’s uses which conform to/deviate from the communitarian (authoritative) ones are (judged to be) correct/incorrect expressions of the communitarianly determined concept denoted by that term.
On its turn, natural externalism might be taken both in a strong and in a weak version. There are two readings of the strong/weak dichotomy. According to the former (cf. McGinn (1989)), strong externalism holds that an intentional state depends on factors lying in the subject’s external environment, whereas weak externalism says that a state only depends on the existence of the external objects it is about. According to latter (cf. Macdonald (1990), Recanati (1993)), the strong/weak dichotomy is tied to a type-token one: unlike strong externalism, weak externalism predicates object-dependence for an intentional state only qua type, not qua token. As a consequence, the two versions give different answers, a negative and a positive one respectively, as to whether it is possible to have a non-object dependent token of an intentional state.
The weak version of externalism is more flexible than the strong one. For it allows an intentional state to be about an external object which is out of its subject’s perceptual scope (as in the case of misperceptions as well as in the one of local hallucinations). However, it does not permit a state to be about an object which does not utterly exist (as in the case of a total hallucination). In such a case, weak externalism forces the subject of a certain state to redescribe its content; for that state is not an object-dependent thought, although it so seemed to its subject. This undermines the alleged first-person authority on a state’s content, the immediate and privileged knowledge a subject is supposed to have of the content of the state (s)he entertains in his/her mind. This is an even bigger problem for strong externalism, which may ascribe an individual a thought including in its content an object with which that individual is, unbeknownst to him/her, causally related.
Internalism has no such problems. For it grows out of the Cartesian intuition according to which for a state’s phenomenological appearance it makes no difference whether it is a veridical (i.e., something corresponds to it in the outside reality) or a non-veridical state. This intuition is read by internalists as supporting the idea that a state has a certain content, transparent to its subject, no matter whether that state is veridical or not. Take this idea as representing solipsistic internalism. This position may be articulated in different forms, according to the different ways an internalist neutralizes the possibility of a relational, object-dependent, content. As to this content, i.e. the state’s informational truth-conditions, an internalist may say either i) that it may receive a syntactic (hence, non-relational) counterpart (cf. Chomsky (1986)) or ii) that it is superfluous in order to account for an individual’s conceptual structure (cf. Fauconnier (1985), Jackendoff (1989)) or even that iii) it may be methodologically bracketed out insofar as it has no role in psychological explanation (cf. Fodor (1980)). Typically, the nature of an internalistically conceived content is taken to be either as the conceptual role or the non-semantic form of the mental representation the state’s possessor is in relation with insofar as (s)he has the state of which it is the content. The conceptual role of a mental representation is taken to be its causal role in an individual’s mental life, perceptual inputs and behavioral outputs of that representation generally included (cf. Harman (1982)). The non-semantic form of that representation is instead identical with, or at least supervenes on, its physical shape placed in an individual’s brain. In the supervenience case, that form will coincide with the syntactico-computational structure of the representation (cf. Fodor (1980)).
But how can a conceptual role or a physico/syntactic form be a content? The representational feature, which intrinsically characterizes the notion of a content, appears here to be lost (cf. McDowell (1986)). Moreover, if in order to account for the Cartesian intuition in its more extreme form (the hyperbolic doubt), one expunges from a content all its relational elements, how can this be individuated as a content? To account for the latter problem, a less radical form of internalism - non-solipsistic internalism - is available, according to which content internally conceived amounts to a mental file where beliefs externally determined are stored (cf. Recanati (1993)). I can thus think about an x which does not exist insofar as I have a “x”-mental file where “x”-beliefs are kept, and these are externally determined. To appeal to mental files in order to save an internalistically-based notion of content, however, entails having a holistical conception of such a content: content is determined by the whole set of an individual’s beliefs. Content holistically conceived suffers however from a lot of problems, such as public unshareability, unlearnability, non-compositionality (cf. Lepore-Loewer (1987), Fodor-Lepore (1992)).
The problems which respectively affect internalism and externalism leave further theoretical options open. The first option may be called world-oriented internalism (traces of it may be found e.g. in Jackendoff (1989), Lakoff (1986), Marconi (1997)). According to this option, beyond an internal content intentional states (or the sentences expressing them) have truth-conditions, but these are not to be conceived as a certain semantics induced by the Putnam cases would like to understand them, i.e. as possible state of affairs subsisting utterly independently from human minds. Rather, they amount to constructed slices of reality, i.e. of reality as humans take it to be.
A second, and more popular, option is dualism, according to which all Frege-, Putnam/Burge-, and Kaplan-cases prompt one to bifurcate the content of an intentional state into a narrow and a broad content (cf. Block (1986), McGinn (1982); Loar (1988), who adheres to a social-externalist conception of broad content, speaks of psychological and social content). Narrow content is content as internalists take it to be; typically, it is conceived as conceptual role. Broad content is instead given by the state’s truth-conditions, taken as a metaphysically independent possible state of affairs.
A main drawback of this position is that it is conceivable that one and the same intentional state has two completely irrelated contents. To mend this drawback, Fodor (1987) proposed a contextualist variant of dualism - inspired by Kaplan’s 1989) notion of character as a function from contexts to contents - according to which narrow content is what the state’s broad content is given a context. So conceived, narrow content may contain relational elements, i.e. those which enable an intentional state, when entertained in a certain context, to pick up a particular broad content. Nevertheless, identified in terms of its narrow content an intentional state may still supervene on its physical basis; for to identify a state in such a way is to individuate it in terms of its causal powers, which ultimately depend on its physical structure (non-relationally conceived).
Different problems with this ‘mapping function’-notion of narrow content (such as its apparent inexpressibility and its schematic nature) may have ultimately led Fodor (1994) to reject it. Intentional states only have broad content, conceived in purely informational terms: the content of an intentional state is what appropriately covaries with it (the theory of the asymmetric dependence between causal laws first advanced by Fodor (1987) is charged to explain what “appropriately” here means). Insofar as states are so identified, they do not supervene on, but reliably covary with, their physical non-relational structure. What prompted one to put forward a notion of narrow content - all the cases recalled two paragraphs above - may be explained away in a way or other, so that there are no relevant narrow content-based generalizations which a scientifically oriented psychology may be charged to miss.
By so doing, however, Fodor throws out the baby with the bath-water. For in his rejection of narrow content he implicitly makes the ‘mapping function’- and the internalistically-based notion of narrow content collapses. This is not to be taken for granted; for the needs that prompt one to defend the former - basically, to account for the Putnam-Burge cases - are not the same as those that prompt one to defend the latter - basically, to account for Frege-cases. Thus, before dispensing with either notion, one has to acknowledge that these notions are distinct.
A different possibility is to altogether avoid the narrow-broad content distinction and make again use a unitary notion of content, suitably attuned to deal with the different potential counterexamples to it represented by Frege-cases on the one hand and Putnam-Burge cases on the other. This is made in Bilgrami (1992) soi-disant externalist conception of content. To be sure, Bilgrami’s notion of content is externalist, insofar as in order for a concept constituting an intentional state’s content to subsist, the beliefs which characterize such a content must be somehow determined by the causal interaction of their subject with an external world. However, such a notion also fits internalist requirements, precisely because such a content consists in a mental file where indefinitely many beliefs, somehow relationally determined, are listed. Frege-cases are thus solved by making different co-referential terms expressing different concepts holistically characterized by a different list of such beliefs. Putnam-cases are instead solved by selecting in the beliefs’ list which constitutes a given concept a subset of beliefs shared by the Putnamian twins and pretending that this subset characterizes a local content responsible for twins’ commonality of behavior.
This is not the only way to recover unitariety of content, however. Another possibility is given by reflecting that to be forced to choose between an internalist and an externalist conception of content presupposes to endorse a inner/outer dichotomy regarding the relationship between mind on the one hand and objecthood on the other. In other terms, it presupposes to take the question, how can something inner - the mind - pick up via the representational content of an intentional state of its something outer - an object, as a genuine problem. There are however several ways to trivialize this problem, based on making the object a genuine constituent of an intentional state. This may be achieved by taking the state as being dependent on, if not constituted by, the entities it is about, where these may be further conceived either in realist terms (cf. McDowell (1986)) or as intensional entities (cf. Castañeda (1989)) or anyway as (directly or indirectly) mentally constituted items (objects of thought, or of human discourse).
2. Externalism and Self-Knowledge (in preparation)
3. Externalism and Mental Causation (in preparation)