The Varieties of Reference

by Gareth Evans, John McDowell (Editor)

Gareth Evans, one of the most brilliant philosophers of his generation, died in 1980 at the age of thirty-four. He had been working for many years on a book about reference, but did not complete it before his death. The work was edited for publication by John McDowell, who contributes a Preface.

Genres: Philosophy, Linguistics

432 pages, Paperback

First published November 1, 1982

About the author

Librarian Note: There is more than one author by this name in the Goodreads database.

Gareth Evans was Wilde Reader in Mental Philosophy and Fellow of University College, Oxford.

Reviews

[Alina · Rating 5 out of 5]

This was one of the most difficult reading experiences I've had in a while. I feel that I now have a general sense of the questions that Evans is after - but I can tell that there are many details of his views, and interesting arguments he rallies, which went over my head. In the introduction, Evans explains that our understanding of referring expressions (i.e., expressions that are used to refer to particulars in the world) is confused. We typically lump demonstratives, proper names, definite descriptions, some pronouns, and certain quantified expressions under this category. Referring terms are especially important to us because they're supposed to indicate to us which object is relevant to the truth-value of the expression of which it is a part. It is Evans's task to show that not all such terms are created alike. The way by which we make contact with the referent of an expression, and the way by which we think about that referent, may be just as important, for the sake of understanding the expression's semantic category.

In chapters 1 and 2, Evans introduces the aspects of Frege's and Russell's views that are relevant to this question. Frege's starting point is that the significance of a sentence consists in its being true or false. Each significant expression has a semantic power, which is its power to impact the truth-value of the sentences in which it occurs. An analysis of a singular sentence (a sentence about a particular) by a theory of meaning must account for the mechanism by which the truth-value of the sentence is determined. There are some issues in Frege's account. First, a proper name may fail to refer to any particular, and so will fail to provide an argument to any singular sentence; in effect, the sentence wouldn't have a truth-value. Frege takes this into account by saying that the sentence is neither true nor false. Second, a concept-expression may be unable to introduce a function that can yield a truth-value for each object of the domain. This led Frege to posit sense, as distinct from semantic value or meaning. Semantic value concerns the relations between expressions and the world, whereas sense concerns the relations between expressions and the understandings speakers have of them.

In chapter 2, Evans introduces Russell's theory of reference. Russell showed that there is not one but two categories for referring expressions. One is genuinely referring, and the other is the 'definite descriptions,' which refers in a different way. Their difference depends upon the way through which we come to know the individual we're thinking of; for genuinely referring terms, we must stand in a perceptual relation to the individual, and for definite descriptions, we may know of the individual through testimony or description. Often something seems to be a genuinely referring term, but in fact it's a description in disguise. Evans points out what he takes to be the major flaws in Russell's view. First, Russell is wrong to deny sense altogether. It's possible for two definite descriptions to share the same referent and yet have different senses, and understanding this has important implications (e.g., it allows for the difference between narrow- and wide-scope negation). Evans will ultimately argue that Russell was right to think that definite descriptions belong to a different semantic category than genuinely referring expressions, but Russell held this for the wrong reasons; it's not because of a difference in whether we met the individual through acquaintance or not, but rather whether the thought that underpins our use of the expression is "information-based" or not.

In chapter 3, Evans examines contemporary work (given his position in the mid-twentieth century) stemming from this Russellian/Fregean tradition. As a whole, these contemporary philosophers believe that the fundamental differences between types of referring expressions depend upon differences in how we think about the objects to which our terms refer. Evans will show that this is mistaken. Russell's theory requires that if we are to refer to an individual, it's necessary that we have a discriminating conception of that object (e.g., we perceive it, or have the capacity to recognize it). This requirement is an obstacle to forming a systematic theory of referring expressions beyond the category of demonstratives. Russell denied that language was social, but contemporary philosophers acknowledge that fact, which leads them to a distinction between what a speaker intends to say, and what they literally say, given conventions.

Kripke takes advantage of this distinction to argue that it's possible to refer to an individual, which we lack capacity to discriminate. All that's required is that there is some objective causal chain that links the the utterance of the term back to causal acquaintance of the object to which the term refers; so a speaker need not have this acquaintance personally, but may just copy the utterance of another speaker, who was acquainted. This directly challenges Russell's principle. Evans argues that Kripke is wrong about this; his error is based in hasty interpretation of the semantic value of sentences that may be appropriately spoken by people who lack familiarity with an individual, where it seems that they refer to that individual. Instead, Evans argues for Russell/Frege's original picture, on which a speaker's personal thought behind an utterance makes all the difference as to determining the semantic category of terms they use. Evans will give much weight to the difference between a person's being in direct causal relations with an object, on the one hand, and a person's apprehending an object via testimony or other means.

In chapter 5, Evans introduces his key concept: the 'information-based thought'. These are thoughts whose content is governed or controlled by information (based in causal relations) we have about the world. This contrasts the thoughts we may have of the world that are based in convention or linguistic stipulation. A thought is controlled by a certain conception of an object if our assessment of the truth-value of the thought, and the significance of the consequences we draw from that thought, are determined by the content of this conception. We can be in a certain information state regardless of whether we believe that this state is veridical or note (e.g., a visual illusion of a stick bent in water will persist even if we believe it is illusory). (I find this assumption particularly interesting and problematic; my issue is that it seems more explanatorily powerful to assume that any consciously accessible content that is information-based must come with the dimension that this content is part of reality; in the stick illusion case, this would mean that we take the visual appearance of the stick bent as part of reality, while we do not take the stick's actually being bent as part of reality -- I think Evans is confused here because he fails to define whether an informational state is necessarily conscious or experienceable, or whether it may be an underlying causal component that gives rise to conscious states ).

Evans holds that there's always a duality between the object which we represent by a thought, on the one hand, and the object whose information feeds into the emergence of our thought, on the other hand. (This adds to the confusion; does this mean that informational states in themselves are never consciously accessible but must always be 'processed' in some way, a process which might necessarily involve shaping by other cognitive factors? If so it's possible to arrive at the conclusion that we can never perceive or think of the world in a manner that is neutral with respect to the existential status of what we're experiencing, contra Evans's claim that such existentially neutral states are possible).

In chapter 6, Evans argues that any demonstrative identification of an object is significantly dissimilar from descriptive identification of that object, and the former underlies Russell's conception of genuinely referring singular terms. Demonstrative identification requires being in a right link of informational-link with the object of experience, which is not identical to perceiving that object. This informational-link may hold if we watch a non-fiction film about that object, and perceive the TV screen, for example. Evans grants that the info-link is insufficient, but holds that the subject must only also have the capacity to locate the object, which might or might not require conceptual capacities.

For example, demonstrative identification behind our utterance of 'here' doesn't require conceptual capacities. Evans takes this as marking the identification of place as a very special kind of demonstrative, which is similar to the identification of one's temporal location and one's self, the topics of the following chapters. Our very possibility of experience and of receiving info from objects at all depends upon that our bodies occupy a place in the world; this implies that we occupy an 'egocentric space'. Evans also argues that action is requisite for having an egocentric space; if we were static from birth to death we'd be incapable of awareness of any sense of directionality, which requires the body's receiving sensations from fixed points in our environment. Moving which respect to those points to determine where they are fixed with respect to the body.

We cannot take any space represented in film, or whose conscious access is mediated through some non-perceptual means, as part of our egocentric space. This is because it is impossible to move to-and-fro within that represented space; we cannot step into a television screen. Evans raises the interesting thought experiment that a person A might be fed information coming up from another person B underwater, so that A's actions above-water control the actions of B, by virtue of some special bodysuit. Even in this case, the underwater world would fail to be included in A's egocentric space; this is because the question would arise for A 'what does it mean from a place (i.e., the above-water place) to be the same as that place (i.e.,the underwater place?' - no such question can possibly arise for the location of our egocentric space, and this reflects that an additional conceptual element is present in A's sense of the underwater space, which is absent in A's sense of the above-water place. Even if A's experience of the underwater place is prolonged to the extent that A forgets about the set-up of this situation, underwater might become 'here' for A, but this doesn't imply that A has two egocentric spaces. There is necessarily always only one egocentric space, and in this abnormal situation, it might become a space that is not physically surrounding A. Thoughts about egocentric space are properly genuinely referring. If we make a judgment about "here", but there's no place at all (e.g., a brain in a vat might do this), the thought behind this doesn't exist. No intelligible proposition would've been entertained, and the judgment is not subject to truth and falsity to begin with.

Demonstrative identification displays 'immunity to error through misidentification'. In order for a thought to be vulnerable to such error, it must involve two distinguishable beliefs, that 'b is F' and 'a=b', whereby we conclude 'a is F' (i.e., the thought is 'identification-dependent'). In demonstrative identification, there's no possible gap between the deliverances of the informational link, and thoughts that employ the idea of the object that arises from this link. There is only one act of grasping that object; there are no distinguishable beliefs at hand, and no possibility of drawing identities between conceptually separable objects. For example, when having 'here'-thoughts, our conception of place is controlled by informational links to the place, and there's no junction of the generation of thoughts here at which we can form a conception of 'here' that is distinguishable from that which is controlled as such. Of course, we could say 'I am in Berkeley', but this is no longer a 'here'-thought; this involves an identification, and is not a purely demonstrative identification, employing conceptual capacities. Of course, we could be wrong about where we are (as in the abnormal case of A's controlling B's underwater movements via a body suit), but this error necessarily involves the malfunctioning of our sensory organs. As long as sensation is functioning ordinarily, demonstrative identifications are immune to errors through misidentification.

In chapter 7, Evans examines demonstrative identification of the first-person pronoun, "I," such that this is almost perfectly parallel to that of place. It turns out that "I" and "here" are two sides of the same coin; in the identification of "I" we don't move ourselves, but we ourselves move, whereas in that of place, we move ourselves as objects located in space. Sydney Shoemaker argued that memory isn't necessarily identification-free; rather, the appearance of this arises from the trivial linguistic fact that we wouldn't describe a person whose information about the past wasn't originally acquired by herself as remembering. We may imagine cases in which there are psychological duplicates A and B, where A remembers B's experiences (i.e., B's experiences provide the causal basis for A's memory). Shoemaker calls this 'quasi-remembering'. In this case, it seems reasonable to ask 'someone was F, but was it me?', which for Shoemaker, serves as the criterion for marking an identification as a condition of a thought. Evans shows that Shoemaker has made a critical mistake. An analogy with the case of place is useful here; in the aforementioned thought experiment, in order for the question for A to arise at all 'this is my egocentric space, but is it here?' the concept 'here' must first be registered; this concept typically entails the location that surrounds one's body. So in order to arrive at the concept of having an egocentric space that is not the physical space that surrounds one's body, one must first possess the concept of this physical space, the typical referent of 'here.'

This indicates that the ordinary 'here' is a priori or the most fundamental, and an atypical 'here' (e.g., the underwater place where B is located) can only be defined derivatively of this ordinary 'here.' Analogously, in the case of memory, in order for A to have a memory in which the subject of experience is B, rather than herself, this requires that A already possess the concept of ordinary memory. A must start off with knowing what it is to remember her own past experiences, and then define, derivatively from that, the possibility that the subject of that experience isn't her but is somebody else. In order for quasi-memory to be the more general or basic case of memory, it must be the case that the most basic case of memory is the result of an identification; but memory cannot operate like that - it cannot be identity neutral, and this isn't due to trivial linguistic fact. Memory must involve the identification-free demonstrative identification of the self insofar as experience involves this, and beliefs based upon experience are automatically temporally maintained. If we're wrong about this demonstrative identification, it's because abnormal circumstances obtain (e.g., the presence of psychological duplicates), not unlike if we're wrong about the demonstrative identification of 'here', it's because our sensory organs are malfunctioning.

Chapters 8 and 9 were less foundational to Evans's view, so I will skip summarizing them.

In chapter 10, Evans examines the semantic values of existential statements that employ empty referring terms (e.g., "Santa Claus isn't real" and "Santa Claus wears red"). These may be broken down into negative existential statements (which are true) and "conniving" existential statements, by which we know that the referring term is empty, and only pretend that they successfully refer (e.g., as in games of make-believe and certain statements about fictions; depending upon the way in which they are used, they may be true or false; but the interesting feature of them is their pragmatic force, whose investigation will let us better understand the flexibility of information-based thoughts). In make-believe reference, in which we pretend that a certain physical object, a prop, is something that it's not, or some other imaginary object, we stand in the usual informational-links to that prop as we do in cases of ordinary reference. But in playing game of make-believe, we will make assertions about that prop's having certain unreal properties, which are not conveyed by those info-links directly, but which we imagine forth on the basis of the prop's actual properties, as conveyed by those info-links. To understand negative existential statements, we must be engaged, or willing to engage, in make-believe, in order to understand the empty referring term, and then to understand how its purported referent doesn't exist.

In chapter 12, Evans examines proper names. This was less foundational, so I'll skip summarizing it.

Here are some of my personal thoughts from reading this work: It seems to me one major achievement of Evans is his showing that questions in the philosophy of mind can be crucial in examining questions in the philosophy of language. For example, basic facts of the embodiment of our cognition, and the possibility of being informationally hooked up to parts of the world, allow for radically different dynamics of the semantics of certain expressions that pertain to these facts (e.g., the demonstratives "I" and "here"). I do philosophy of mind generally, and so this inspires me to want to think of potential repercussions of ideas I've been considering upon issues in language - these would be very distant and difficult projects to undertake though.

Also, I found myself often preoccupied by the question of how we can refer to the values (based in affect or inclinations towards certain actions, i.e., affordances) of particular objects. Can we be informationally linked onto these? It seems like we can, but the source of the information would not be straightforwardly mind-independent as in the case with the physical presence of objects. Instead, this information would be based upon past experiences of the ways the object can impact our emotion, or the ways we can act upon the object, which is always partially based upon its mind-independent properties.

[Larry · Rating 4 out of 5]

Really takes off 200 pages in. Deservedly famous discussions of recognition-based identification; basically phenomenology done right. The key point of Evans' subtle discussion is how it relates to 'Russell's principle', namely the idea that you can only think of an object if you can discriminate it from others. Then, Evans proceeds to relativize this epistemic capacity to a context of exercise, distinguishing recognition from 'recall' in absentia (the former being a kind of non-conceptual ability, i.e. an entirely context-dependent capacity). It's pretty clear around the edges that Evans had a firm grasp of unarticulated constituents (see his theory of modes).

[Chant · Rating 5 out of 5]

Classic philosophy of language text that deserves more attention (imo).

[Nat · Rating 5 out of 5]

This is a fantastic work of philosophy. Not because it is particularly clear (in places it is incredibly dense and is full of weird, idiosyncratic, technical language that you have to learn to figure out what's going on), but because it draws materials from a bunch of different fields and makes all kinds of provocative claims.

[Kathleen · Rating 3 out of 5]

This is an excellent book if you are a philosopher who wants to learn a really good theory of reference with implications for Phil mind and Phil language (and which also, indirectly, helped me understand a lot of Kant way better); it is full deserving of its status as a classic. If you are a normal person maybe avoid it

[Fernando]

no entiendo una puta madre... me acostaría de buena gana con quien me explicara la teoría de la referencia de este señor --ok, ok, ése tal vez no sea el mejor de los pagos, pero no tego dinero :S

¿alguien me explica? :P

[Tudor · Rating 5 out of 5]

well worth several more reads