Åsa Wikforss, Stockholm University

Purpose

The purpose of the project has been to examine central issues concerning the determination of thought content and self-knowledge. Contemporary 'externalist' theories of content emphasize the role of the external environment in the determination of content, and the question is whether such theories pose a threat to the traditional view that we know our own thoughts directly and authoritatively. If content is determined by facts in the environment, facts that could only be known through empirical investigations, does it follow that knowing one's own thoughts require undertaking such investigations? In response to this question two camps have formed: The compatibilists, who claim that there is no conflict between externalism and the traditional picture of self-knowledge, and the incompatibilists who argue that content externalism undermines self-knowledge. To support her claim, the incompatibilist employs two arguments, the 'switching argument' and the 'reductio argument': First, she imagines a situation where the individual, without knowing it, is switched from one environment to another in such a way that the content of her thoughts change. In such a scenario, it is claimed, the individual does not know her own thoughts without undertaking empirical investigations of her environment. Second, the incompatibilist argues that if, for the sake of argument, it is assumed that externalism and self-knowledge are compatible, absurdities can be derived. For instance, on the assumption that the individual knows that she is thinking that water is wet, she should be in a position to infer that there is water in her environment - purely a priori - which is absurd.

Working hypothesis

My working hypothesis was that this debate is fundamentally misconceived since it is based on a problematic notion of 'knowledge of content'. The real difficulty concerns not self-knowledge but the assumption, integral to externalism, that speakers typically have an incomplete grasp of the content of their own thoughts. This assumption forces on us a notion of thought content that fails to capture the individual's cognitive perspective, her reasoning and actions. Instead, any plausible theory of content should be constrained by the idea that content captures the individual's cognitive perspective. If externalism cannot meet this constraint, it should be rejected. During the project time, I have developed these ideas in a number of papers.

In my paper "Self-Knowledge and Knowledge of Content", I defend the claim that the real difficulty does not concern self-knowledge but incomplete understanding. To show this I start by examining the debate over the switching argument. In response to this argument, externalists have appealed to the idea that when we self-ascribe an occurrent thought ('I am thinking, with this very thought, that water is wet') we cannot be in error since the form of the judgment guarantees its truth. Hence, it does not matter if the content of the thought switches unbeknownst to the speaker: her reflexive judgments will still be true and there is no room for any 'content errors'. The incompatibilists have found this reply deeply unsatisfactory. Even if reflexive judgments of this sort cannot be mistaken, it has been argued, there is still an important sense in which the individual who self-ascribes her thought does not know what she is thinking (on the externalist picture). However, it has proven surprisingly difficult to spell out why this should be so, and what is missing for the judgments to qualify as knowledge.

In the paper I make three main claims. First, I argue that although the compatibilist is quite right to insist that content externalism cannot threaten the knowledge we have of our own occurrent thoughts, incompatibilists are equally right to suggest that this reply fails to address the real problem. The incompatibilists are mistaken, however, in assuming that the real problem is epistemological. My second claim is that the proper question to ask is whether externalism can provide a plausible account of understanding or concept grasp. Grasping a concept or a content is sometimes characterized in terms of 'knowledge of content'; but, I argue, such 'knowledge' cannot be construed as a form of propositional knowledge, on pain of a regress. My final claim is that externalists have notorious difficulties giving a plausible account of understanding. This, I argue, is a direct result of the fact that externalism depends on the assumption that individuals have an incomplete grasp of the concepts that go into their own thoughts. According to the content externalist one can be in a position to knowingly self-ascribe a thought that one does not understand. This, I propose, identifies the real source behind the incompatibilist's intuition that appealing to the self-verifying nature of the cogito-thoughts is unsatisfactory. I also develop exactly how the relevant notion of 'incomplete understanding' is to be understood in this context. It is one thing to claim that we typically employ concepts without being able to give explicit definitions or explications of them. This, I argue, is just a result of the fact that the ability to provide such explications is a meta-ability that requires more complex cognitive capacities than merely employing the concept on the object-level. Even the internalist will grant that we have an incomplete understanding of our own concepts in this sense. The externalist, however, is committed to the further idea that the individual makes simple conceptual errors and reasoning mistakes at the object level. This is a notion of 'incomplete understanding' that is distinctive of externalism and that causes difficulties when it comes to giving an account of the individual's cognitive perspective.

The suggestion that content externalism relies on the assumption of incomplete understanding has been disputed. Thus, in a response to one of my earlier papers ('Social Externalism and Conceptual Errors') Sarah Sawyer argues that the externalist need not make this assumption but can simply rely on the idea that a difference in extension entails a difference in thought content. In my paper 'Externalism and Incomplete Understanding', I respond to Sawyer's claim. I argue that the principle Sawyer refers to is in itself neutral concerning the determination of content and, moreover, that all types of externalism (whether social or physical) must rely on the assumption that individuals think with concepts they incompletely understand.

Some externalists claim that there is only a prima facie conflict between content externalism and the idea that content should capture the cognitive perspective of the individual. Indeed, it has even been claimed that content externalism is perfectly compatible with a Fregean conception of thought content. My paper "Content Externalism and Fregean Sense" discusses this proposal. I examine Tyler Burge's attempt to reconcile externalism with Frege and argue that it fails. Jessica Brown has also argued that there is a basic conflict between the Fregean idea and content externalism, but suggests that this shows that the former should be abandoned. Indeed, she argues, the content externalist is in a position to provide an alternative explanation of the phenomena that worried Frege by rejecting a central Fregean assumption, the 'transparency of sameness of content'. I argue that Brown's strategy fails, and that, instead, we face a stark choice: Abandon the idea that content should capture cognitive role, or abandon externalism. Faced with this choice, I suggest, we should abandon content externalism.

In my review of Brown's book from 2004, I develop the criticisms of Brown. Although Brown used to argue that externalism and self-knowledge are incompatible, presenting her own version of the reductio argument, in the book she rejects the reductio argument and endorses compatibilism. In my review I argue that Brown provides no good reasons for this change of heart. In particular, I suggest that her reply to the reductio is very problematic. The reply relies on the idea that since we cannot know a priori whether a purported natural kind term in fact denotes a natural kind, we cannot know a priori whether a term expresses a natural kind concept or not. Since this is so, she argues, knowing that one is thinking that water is wet does not put one in a position to infer that there is water in one's environment - an empirical premise must be added, that the concept of water in fact is a natural kind concept. This move, I argue, is very problematic since it commits the externalist to an even more radical form of incomplete understanding: i.e. to the idea that we cannot know a priori whether a term expresses a directly referential concept (a natural kind concept) or a descriptive one ('the transparent, thirst-quenching, potable liquid....').

This criticism is developed in two other papers. First, in '"Externalism and A Posteriori Semantics", Sören Häggqvist and I jointly criticize various versions of the suggestion that whether or not a term has an externalist semantics or not depends on the external environment, a thesis we dub 'a posteriori semantics'. One motivation for this thesis is the aim to account for cases where a putative natural kind term fails to pick out a natural kind: The term may have a standard externalist semantics (if it picks out a natural kind) or a descriptivist one (if it does not). Knowing which semantics applies will therefore require detailed empirical knowledge. This move has also been employed in cases where a singular term, such as a name, fails to have a reference. We argue that a posteriori semantics should be rejected. A number of difficulties for the position - metaphysical, epistemological and methodological - are articulated. Finally, we suggest that a posteriori semantics misconstrues the way in which semantics is empirical.

Second, I have written a critical reply to a recent paper by Daniel Korman. Korman argues that the externalist can fall back on certain 'default conditionals' in order to deal with cases where a purported natural kind term fails to pick out a natural kind. For instance, Korman suggests, if 'water' fails to pick out any liquid at all (if we are on 'Dry Earth'), the externalist can simply hold that 'water' expresses a certain descriptive concept. I argue that the appeal to default conditionals is flawed in a number of ways, and that, in the end it forces the externalist to endorse a very problematic version of a posteriori semantics.

In the debate, content externalism is taken to be correct without much argument. The question debated is not whether externalism is correct but what its implications are, for example its epistemological implications. The widely shared assumption is that if it turns out that some of these implications are unpalatable then this is something we have to live with, since the externalist position is not to be questioned. By contrast, I think that the difficulties encountered by the standard externalist accounts of thought content indicate that the time has come to question the externalist orthodoxy. However, there is much confusion concerning how the externalist thesis is to be understood in the first place, and this makes it difficult to assess it. For this reason, I have written a survey paper attempting to clear up the terminological confusions that abound, "Semantic externalism. The theory and its foundations".

In my paper "Externalism and World-Dependence" I assess one particular externalist claim: i.e. the claim that externalism implies the thesis that one cannot think thoughts containing natural kind concepts unless one has been in causal contact with instances of the kind. I argue that contrary to received opinion dependency claims of this sort cannot be defended - even if one accepts the standard externalist Twin Earth arguments. The tendency to construe Twin Earth-externalism as a form of dependency thesis, I suggest, results from a failure to keep apart two distinct types of externalism, what I call 'foundational externalism', a thesis concerning the determination of meaning and content, and 'externalist semantics', a thesis concerning the semantic value of certain types of terms (such as natural kind terms). Foundational externalism does not support any thesis concerning the necessary conditions for concept possession, and although externalist semantics entails that certain terms have an object-dependent semantic value, it is only available in the case of singular terms. This, I suggest, has implications for the debate over self-knowledge, in particular the debate over the reductio argument. The argument is directed against externalist semantics, not foundational externalism, and there is no reason to believe it applies in the case of general terms, such as 'water'. It also follows that psychological externalism, the view that certain types of psychological states are relational in character, does not have the semantic support it is generally is assumed to have. At most, psychological externalism applies to a rather limited class of psychological states.

In assessing the externalist thesis I have focused on natural kind terms since these terms are very generally assumed to require an externalist account. Thus, I have written a paper, "Naming Natural Kinds" criticizing foundational externalism with respect to these terms. Together with Sören Häggqvist, I have also written a paper examining Kripke's arguments against descriptivist accounts of the meaning of natural kind terms. We argue that Kripke's contention that natural kind terms are semantically close to singular terms cannot be upheld and that his objections to descriptivist account of proper names do not carry over to the case of kind terms. We end by suggesting that a version of the cluster theory can be defended, and that standard externalist accounts should be rejected.

In addition, I have written on epistemological issues concerning self-knowledge. I have written two papers on the epistemology of first-person knowledge of beliefs ("Knowing Our Own Beliefs" and "Self-Knowledge and Evans's Principle") discussing what it means for judgments about our own beliefs to be justified. I have also written on the question of what it is to know one's own language, arguing against the common idea that knowing one's own language should be characterized as standing in a cognitive relation to one's language (in "Objectivity and Error").

I have presented versions of all of these papers at a large number of conferences and workshops and around the world. I have also participated in a number of public events in Sweden, including TV and radio performances where philosophical issues relating to my research have been discussed.